[llvm] [AArch64] Generate zeroing forms of certain SVE2.2 instructions (11/11) (PR #116837)
Momchil Velikov via llvm-commits
llvm-commits at lists.llvm.org
Mon Dec 23 06:32:52 PST 2024
https://github.com/momchil-velikov updated https://github.com/llvm/llvm-project/pull/116837
>From 482b828db88326bc2e356f53fe11f03443a7989e Mon Sep 17 00:00:00 2001
From: Momchil Velikov <momchil.velikov at arm.com>
Date: Thu, 19 Dec 2024 15:43:05 +0000
Subject: [PATCH 01/16] [AArch64] Prefer SVE2.2 zeroing forms of certain
instructions with an all-true predicate
When the predicate of a destructive operation is known to be all-true,
for example
fabs z0.s, p0/m, z1.s
then the entire output register is written and we can use a zeroing
(instead of a merging) form of the instruction, for example
fabs z0.s, p0/z, z1.s
thus eliminate the dependency on the input-output destination register
without the need to insert a `movprfx`.
This patch complements (and in the case of 2b3266c1701f, fixes a regression)
the following:
7f4414b2a1a4 [AArch64] Generate zeroing forms of certain SVE2.2 instructions (4/11) (#116830)
2474cf7ad123 [AArch64] Generate zeroing forms of certain SVE2.2 instructions (3/11) (#116829)
6f285d311595 [AArch64] Generate zeroing forms of certain SVE2.2 instructions (2/11) (#116828)
2b3266c1701f [AArch64] Generate zeroing forms of certain SVE2.2 instructions (1/11) (#116259)
---
.../Target/AArch64/AArch64ISelDAGToDAG.cpp | 2 +
llvm/lib/Target/AArch64/AArch64InstrInfo.td | 3 -
.../lib/Target/AArch64/AArch64SVEInstrInfo.td | 8 -
llvm/lib/Target/AArch64/SVEInstrFormats.td | 102 ++-
.../CodeGen/AArch64/zeroing-forms-abs-neg.ll | 600 +++++++++++++++++-
.../AArch64/zeroing-forms-fcvt-bfcvt.ll | 287 ++++++++-
.../AArch64/zeroing-forms-fcvtlt-fcvtx.ll | 123 +++-
.../CodeGen/AArch64/zeroing-forms-fcvtzsu.ll | 580 ++++++++++++++++-
8 files changed, 1577 insertions(+), 128 deletions(-)
diff --git a/llvm/lib/Target/AArch64/AArch64ISelDAGToDAG.cpp b/llvm/lib/Target/AArch64/AArch64ISelDAGToDAG.cpp
index ff3ca8a24fc04a..6aa8cd4f0232ac 100644
--- a/llvm/lib/Target/AArch64/AArch64ISelDAGToDAG.cpp
+++ b/llvm/lib/Target/AArch64/AArch64ISelDAGToDAG.cpp
@@ -228,6 +228,8 @@ class AArch64DAGToDAGISel : public SelectionDAGISel {
return false;
}
+ bool SelectAny(SDValue) { return true; }
+
bool SelectDupZero(SDValue N) {
switch(N->getOpcode()) {
case AArch64ISD::DUP:
diff --git a/llvm/lib/Target/AArch64/AArch64InstrInfo.td b/llvm/lib/Target/AArch64/AArch64InstrInfo.td
index d015cc15581ad0..b1c61c8f84ef45 100644
--- a/llvm/lib/Target/AArch64/AArch64InstrInfo.td
+++ b/llvm/lib/Target/AArch64/AArch64InstrInfo.td
@@ -381,9 +381,6 @@ def NoUseScalarIncVL : Predicate<"!Subtarget->useScalarIncVL()">;
def UseSVEFPLD1R : Predicate<"!Subtarget->noSVEFPLD1R()">;
-def UseUnaryUndefPseudos
- : Predicate<"!(Subtarget->isSVEorStreamingSVEAvailable() && (Subtarget->hasSVE2p2() || Subtarget->hasSME2p2()))">;
-
def AArch64LocalRecover : SDNode<"ISD::LOCAL_RECOVER",
SDTypeProfile<1, 1, [SDTCisSameAs<0, 1>,
SDTCisInt<1>]>>;
diff --git a/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td b/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
index 4a4412f9df6a1a..1e297eb5feb266 100644
--- a/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
+++ b/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
@@ -675,14 +675,6 @@ let Predicates = [HasSVEorSME] in {
defm FABS_ZPmZ : sve_int_un_pred_arit_bitwise_fp<0b100, "fabs", AArch64fabs_mt>;
defm FNEG_ZPmZ : sve_int_un_pred_arit_bitwise_fp<0b101, "fneg", AArch64fneg_mt>;
- let Predicates = [HasSVEorSME, UseUnaryUndefPseudos] in {
- defm FABS_ZPmZ : sve_fp_un_pred_arit_hsd<AArch64fabs_mt>;
- defm FNEG_ZPmZ : sve_fp_un_pred_arit_hsd<AArch64fneg_mt>;
-
- defm ABS_ZPmZ : sve_int_un_pred_arit_bhsd<AArch64abs_mt>;
- defm NEG_ZPmZ : sve_int_un_pred_arit_bhsd<AArch64neg_mt>;
- }
-
foreach VT = [nxv2bf16, nxv4bf16, nxv8bf16] in {
// No dedicated instruction, so just clear the sign bit.
def : Pat<(VT (fabs VT:$op)),
diff --git a/llvm/lib/Target/AArch64/SVEInstrFormats.td b/llvm/lib/Target/AArch64/SVEInstrFormats.td
index f8a633872cb966..a1bce3601aab60 100644
--- a/llvm/lib/Target/AArch64/SVEInstrFormats.td
+++ b/llvm/lib/Target/AArch64/SVEInstrFormats.td
@@ -484,6 +484,7 @@ let Predicates = [HasSVEorSME] in {
//===----------------------------------------------------------------------===//
def SVEDup0 : ComplexPattern<vAny, 0, "SelectDupZero", []>;
def SVEDup0Undef : ComplexPattern<vAny, 0, "SelectDupZeroOrUndef", []>;
+def SVEAny : ComplexPattern<vAny, 0, "SelectAny", []>;
class SVE_1_Op_Pat<ValueType vtd, SDPatternOperator op, ValueType vt1,
Instruction inst>
@@ -504,10 +505,15 @@ multiclass SVE_1_Op_PassthruUndef_Pat<ValueType vtd, SDPatternOperator op, Value
(inst $Op3, $Op1, $Op2)>;
}
-class SVE_1_Op_PassthruUndefZero_Pat<ValueType vtd, SDPatternOperator op, ValueType pg,
- ValueType vts, Instruction inst>
- : Pat<(vtd (op pg:$Op1, vts:$Op2, (vtd (SVEDup0Undef)))),
- (inst $Op1, $Op2)>;
+multiclass SVE_1_Op_PassthruUndefZero_Pat<ValueType vtd, SDPatternOperator op, ValueType pg,
+ ValueType vts, Instruction inst> {
+ let AddedComplexity = 1 in {
+ def : Pat<(vtd (op pg:$Op1, vts:$Op2, (vtd (SVEDup0Undef)))),
+ (inst $Op1, $Op2)>;
+ def : Pat<(vtd (op (pg (SVEAllActive:$Op1)), vts:$Op2, (vtd (SVEAny)))),
+ (inst $Op1, $Op2)>;
+ }
+}
// Used to match FP_ROUND_MERGE_PASSTHRU, which has an additional flag for the
// type of rounding. This is matched by timm0_1 in pattern below and ignored.
@@ -576,10 +582,15 @@ multiclass SVE_3_Op_Undef_Pat<ValueType vtd, SDPatternOperator op, ValueType vt1
(inst $Op1, $Op2, $Op3)>;
}
-class SVE_3_Op_UndefZero_Pat<ValueType vtd, SDPatternOperator op, ValueType vt1,
- ValueType vt2, ValueType vt3, Instruction inst>
- : Pat<(vtd (op (vt1 (SVEDup0Undef)), vt2:$Op1, vt3:$Op2)),
- (inst $Op1, $Op2)>;
+multiclass SVE_3_Op_UndefZero_Pat<ValueType vtd, SDPatternOperator op, ValueType vt1,
+ ValueType vt2, ValueType vt3, Instruction inst> {
+ let AddedComplexity = 1 in {
+ def : Pat<(vtd (op (vt1 (SVEDup0Undef)), vt2:$Op1, vt3:$Op2)),
+ (inst $Op1, $Op2)>;
+ def : Pat<(vtd (op (vt1 (SVEAny)), (vt2 (SVEAllActive:$Op2)), vt3:$Op3)),
+ (inst $Op2, $Op3)>;
+ }
+}
class SVE_4_Op_Pat<ValueType vtd, SDPatternOperator op, ValueType vt1,
ValueType vt2, ValueType vt3, ValueType vt4,
@@ -2862,8 +2873,8 @@ multiclass sve_fp_fcvtltz<string asm, string op> {
def _HtoS : sve_fp_fcvt2z<0b1001, asm, ZPR32, ZPR16>;
def _StoD : sve_fp_fcvt2z<0b1111, asm, ZPR64, ZPR32>;
- def : SVE_3_Op_UndefZero_Pat<nxv4f32, !cast<SDPatternOperator>(op # _f32f16), nxv4f32, nxv4i1, nxv8f16, !cast<Instruction>(NAME # _HtoS)>;
- def : SVE_3_Op_UndefZero_Pat<nxv2f64, !cast<SDPatternOperator>(op # _f64f32), nxv2f64, nxv2i1, nxv4f32, !cast<Instruction>(NAME # _StoD)>;
+ defm : SVE_3_Op_UndefZero_Pat<nxv4f32, !cast<SDPatternOperator>(op # _f32f16), nxv4f32, nxv4i1, nxv8f16, !cast<Instruction>(NAME # _HtoS)>;
+ defm : SVE_3_Op_UndefZero_Pat<nxv2f64, !cast<SDPatternOperator>(op # _f64f32), nxv2f64, nxv2i1, nxv4f32, !cast<Instruction>(NAME # _StoD)>;
}
//===----------------------------------------------------------------------===//
@@ -3273,7 +3284,7 @@ class sve_fp_z2op_p_zd<bits<7> opc,string asm, RegisterOperand i_zprtype,
multiclass sve_fp_z2op_p_zd<string asm, SDPatternOperator op> {
def _DtoS : sve_fp_z2op_p_zd<0b0001010, asm, ZPR64, ZPR32>;
- def : SVE_3_Op_UndefZero_Pat<nxv4f32, op, nxv4f32, nxv2i1, nxv2f64, !cast<Instruction>(NAME # _DtoS)>;
+ defm : SVE_3_Op_UndefZero_Pat<nxv4f32, op, nxv4f32, nxv2i1, nxv2f64, !cast<Instruction>(NAME # _DtoS)>;
}
multiclass sve_fp_z2op_p_zd_hsd<bits<5> opc, string asm> {
@@ -3290,7 +3301,7 @@ multiclass sve_fp_z2op_p_zd_frint<bits<2> opc, string asm> {
multiclass sve_fp_z2op_p_zd_bfcvt<string asm, SDPatternOperator op> {
def NAME : sve_fp_z2op_p_zd<0b1001010, asm, ZPR32, ZPR16>;
- def : SVE_3_Op_UndefZero_Pat<nxv8bf16, op, nxv8bf16, nxv4i1, nxv4f32, !cast<Instruction>(NAME)>;
+ defm : SVE_3_Op_UndefZero_Pat<nxv8bf16, op, nxv8bf16, nxv4i1, nxv4f32, !cast<Instruction>(NAME)>;
}
multiclass sve_fp_z2op_p_zd_d<bit U, string asm, string int_op, SDPatternOperator ir_op> {
@@ -3302,14 +3313,14 @@ multiclass sve_fp_z2op_p_zd_d<bit U, string asm, string int_op, SDPatternOperato
def _DtoS : sve_fp_z2op_p_zd<{ 0b111100, U }, asm, ZPR64, ZPR32>;
def _DtoD : sve_fp_z2op_p_zd<{ 0b111111, U }, asm, ZPR64, ZPR64>;
- def : SVE_3_Op_UndefZero_Pat<nxv4i32, !cast<SDPatternOperator>(int_op # _i32f64), nxv4i32, nxv2i1, nxv2f64, !cast<Instruction>(NAME # _DtoS)>;
- def : SVE_3_Op_UndefZero_Pat<nxv2i64, !cast<SDPatternOperator>(int_op # _i64f32), nxv2i64, nxv2i1, nxv4f32, !cast<Instruction>(NAME # _StoD)>;
- def : SVE_3_Op_UndefZero_Pat<nxv4i32, !cast<SDPatternOperator>(int_op # _i32f16), nxv4i32, nxv4i1, nxv8f16, !cast<Instruction>(NAME # _HtoS)>;
- def : SVE_3_Op_UndefZero_Pat<nxv2i64, !cast<SDPatternOperator>(int_op # _i64f16), nxv2i64, nxv2i1, nxv8f16, !cast<Instruction>(NAME # _HtoD)>;
+ defm : SVE_3_Op_UndefZero_Pat<nxv4i32, !cast<SDPatternOperator>(int_op # _i32f64), nxv4i32, nxv2i1, nxv2f64, !cast<Instruction>(NAME # _DtoS)>;
+ defm : SVE_3_Op_UndefZero_Pat<nxv2i64, !cast<SDPatternOperator>(int_op # _i64f32), nxv2i64, nxv2i1, nxv4f32, !cast<Instruction>(NAME # _StoD)>;
+ defm : SVE_3_Op_UndefZero_Pat<nxv4i32, !cast<SDPatternOperator>(int_op # _i32f16), nxv4i32, nxv4i1, nxv8f16, !cast<Instruction>(NAME # _HtoS)>;
+ defm : SVE_3_Op_UndefZero_Pat<nxv2i64, !cast<SDPatternOperator>(int_op # _i64f16), nxv2i64, nxv2i1, nxv8f16, !cast<Instruction>(NAME # _HtoD)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv8i16, ir_op, nxv8i1, nxv8f16, !cast<Instruction>(NAME # _HtoH)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv4i32, ir_op, nxv4i1, nxv4f32, !cast<Instruction>(NAME # _StoS)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv2i64, ir_op, nxv2i1, nxv2f64, !cast<Instruction>(NAME # _DtoD)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv8i16, ir_op, nxv8i1, nxv8f16, !cast<Instruction>(NAME # _HtoH)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv4i32, ir_op, nxv4i1, nxv4f32, !cast<Instruction>(NAME # _StoS)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv2i64, ir_op, nxv2i1, nxv2f64, !cast<Instruction>(NAME # _DtoD)>;
}
multiclass sve_fp_z2op_p_zd_c<bit U, string asm> {
@@ -3336,12 +3347,12 @@ multiclass sve_fp_z2op_p_zd_b_0<string asm, string op> {
def _DtoS : sve_fp_z2op_p_zd<0b1101010, asm, ZPR64, ZPR32>;
def _StoD : sve_fp_z2op_p_zd<0b1101011, asm, ZPR32, ZPR64>;
- def : SVE_3_Op_UndefZero_Pat<nxv8f16, !cast<SDPatternOperator>(op # _f16f32), nxv8f16, nxv4i1, nxv4f32, !cast<Instruction>(NAME # _StoH)>;
- def : SVE_3_Op_UndefZero_Pat<nxv8f16, !cast<SDPatternOperator>(op # _f16f64), nxv8f16, nxv2i1, nxv2f64, !cast<Instruction>(NAME # _DtoH)>;
- def : SVE_3_Op_UndefZero_Pat<nxv4f32, !cast<SDPatternOperator>(op # _f32f64), nxv4f32, nxv2i1, nxv2f64, !cast<Instruction>(NAME # _DtoS)>;
- def : SVE_3_Op_UndefZero_Pat<nxv4f32, !cast<SDPatternOperator>(op # _f32f16), nxv4f32, nxv4i1, nxv8f16, !cast<Instruction>(NAME # _HtoS)>;
- def : SVE_3_Op_UndefZero_Pat<nxv2f64, !cast<SDPatternOperator>(op # _f64f16), nxv2f64, nxv2i1, nxv8f16, !cast<Instruction>(NAME # _HtoD)>;
- def : SVE_3_Op_UndefZero_Pat<nxv2f64, !cast<SDPatternOperator>(op # _f64f32), nxv2f64, nxv2i1, nxv4f32, !cast<Instruction>(NAME # _StoD)>;
+ defm : SVE_3_Op_UndefZero_Pat<nxv8f16, !cast<SDPatternOperator>(op # _f16f32), nxv8f16, nxv4i1, nxv4f32, !cast<Instruction>(NAME # _StoH)>;
+ defm : SVE_3_Op_UndefZero_Pat<nxv8f16, !cast<SDPatternOperator>(op # _f16f64), nxv8f16, nxv2i1, nxv2f64, !cast<Instruction>(NAME # _DtoH)>;
+ defm : SVE_3_Op_UndefZero_Pat<nxv4f32, !cast<SDPatternOperator>(op # _f32f64), nxv4f32, nxv2i1, nxv2f64, !cast<Instruction>(NAME # _DtoS)>;
+ defm : SVE_3_Op_UndefZero_Pat<nxv4f32, !cast<SDPatternOperator>(op # _f32f16), nxv4f32, nxv4i1, nxv8f16, !cast<Instruction>(NAME # _HtoS)>;
+ defm : SVE_3_Op_UndefZero_Pat<nxv2f64, !cast<SDPatternOperator>(op # _f64f16), nxv2f64, nxv2i1, nxv8f16, !cast<Instruction>(NAME # _HtoD)>;
+ defm : SVE_3_Op_UndefZero_Pat<nxv2f64, !cast<SDPatternOperator>(op # _f64f32), nxv2f64, nxv2i1, nxv4f32, !cast<Instruction>(NAME # _StoD)>;
}
//===----------------------------------------------------------------------===//
@@ -4859,6 +4870,16 @@ multiclass sve_int_un_pred_arit<bits<3> opc, string asm,
def : SVE_1_Op_Passthru_Pat<nxv8i16, op, nxv8i1, nxv8i16, !cast<Instruction>(NAME # _H)>;
def : SVE_1_Op_Passthru_Pat<nxv4i32, op, nxv4i1, nxv4i32, !cast<Instruction>(NAME # _S)>;
def : SVE_1_Op_Passthru_Pat<nxv2i64, op, nxv2i1, nxv2i64, !cast<Instruction>(NAME # _D)>;
+
+ def _B_UNDEF : PredOneOpPassthruPseudo<NAME # _B, ZPR8>;
+ def _H_UNDEF : PredOneOpPassthruPseudo<NAME # _H, ZPR16>;
+ def _S_UNDEF : PredOneOpPassthruPseudo<NAME # _S, ZPR32>;
+ def _D_UNDEF : PredOneOpPassthruPseudo<NAME # _D, ZPR64>;
+
+ defm : SVE_1_Op_PassthruUndef_Pat<nxv16i8, op, nxv16i1, nxv16i8, !cast<Pseudo>(NAME # _B_UNDEF)>;
+ defm : SVE_1_Op_PassthruUndef_Pat<nxv8i16, op, nxv8i1, nxv8i16, !cast<Pseudo>(NAME # _H_UNDEF)>;
+ defm : SVE_1_Op_PassthruUndef_Pat<nxv4i32, op, nxv4i1, nxv4i32, !cast<Pseudo>(NAME # _S_UNDEF)>;
+ defm : SVE_1_Op_PassthruUndef_Pat<nxv2i64, op, nxv2i1, nxv2i64, !cast<Pseudo>(NAME # _D_UNDEF)>;
}
multiclass sve_int_un_pred_arit_z<bits<3> opc, string asm, SDPatternOperator op> {
@@ -4867,10 +4888,10 @@ multiclass sve_int_un_pred_arit_z<bits<3> opc, string asm, SDPatternOperator op>
def _S : sve_int_un_pred_arit_z<0b10, { opc, 0b0 }, asm, ZPR32>;
def _D : sve_int_un_pred_arit_z<0b11, { opc, 0b0 }, asm, ZPR64>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv16i8, op, nxv16i1, nxv16i8, !cast<Instruction>(NAME # _B)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv8i16, op, nxv8i1, nxv8i16, !cast<Instruction>(NAME # _H)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv4i32, op, nxv4i1, nxv4i32, !cast<Instruction>(NAME # _S)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv2i64, op, nxv2i1, nxv2i64, !cast<Instruction>(NAME # _D)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv16i8, op, nxv16i1, nxv16i8, !cast<Instruction>(NAME # _B)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv8i16, op, nxv8i1, nxv8i16, !cast<Instruction>(NAME # _H)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv4i32, op, nxv4i1, nxv4i32, !cast<Instruction>(NAME # _S)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv2i64, op, nxv2i1, nxv2i64, !cast<Instruction>(NAME # _D)>;
}
multiclass sve_int_un_pred_arit_h<bits<3> opc, string asm,
@@ -4984,6 +5005,17 @@ multiclass sve_int_un_pred_arit_bitwise_fp<bits<3> opc, string asm,
def : SVE_1_Op_Passthru_Pat<nxv4f32, op, nxv4i1, nxv4f32, !cast<Instruction>(NAME # _S)>;
def : SVE_1_Op_Passthru_Pat<nxv2f32, op, nxv2i1, nxv2f32, !cast<Instruction>(NAME # _S)>;
def : SVE_1_Op_Passthru_Pat<nxv2f64, op, nxv2i1, nxv2f64, !cast<Instruction>(NAME # _D)>;
+
+ def _H_UNDEF : PredOneOpPassthruPseudo<NAME # _H, ZPR16>;
+ def _S_UNDEF : PredOneOpPassthruPseudo<NAME # _S, ZPR32>;
+ def _D_UNDEF : PredOneOpPassthruPseudo<NAME # _D, ZPR64>;
+
+ defm : SVE_1_Op_PassthruUndef_Pat<nxv8f16, op, nxv8i1, nxv8f16, !cast<Pseudo>(NAME # _H_UNDEF)>;
+ defm : SVE_1_Op_PassthruUndef_Pat<nxv4f16, op, nxv4i1, nxv4f16, !cast<Pseudo>(NAME # _H_UNDEF)>;
+ defm : SVE_1_Op_PassthruUndef_Pat<nxv2f16, op, nxv2i1, nxv2f16, !cast<Pseudo>(NAME # _H_UNDEF)>;
+ defm : SVE_1_Op_PassthruUndef_Pat<nxv4f32, op, nxv4i1, nxv4f32, !cast<Pseudo>(NAME # _S_UNDEF)>;
+ defm : SVE_1_Op_PassthruUndef_Pat<nxv2f32, op, nxv2i1, nxv2f32, !cast<Pseudo>(NAME # _S_UNDEF)>;
+ defm : SVE_1_Op_PassthruUndef_Pat<nxv2f64, op, nxv2i1, nxv2f64, !cast<Pseudo>(NAME # _D_UNDEF)>;
}
multiclass sve_int_un_pred_arit_bitwise_fp_z<bits<3> opc, string asm, SDPatternOperator op> {
@@ -4991,12 +5023,12 @@ multiclass sve_int_un_pred_arit_bitwise_fp_z<bits<3> opc, string asm, SDPatternO
def _S : sve_int_un_pred_arit_z<0b10, { opc, 0b1 }, asm, ZPR32>;
def _D : sve_int_un_pred_arit_z<0b11, { opc, 0b1 }, asm, ZPR64>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv8f16, op, nxv8i1, nxv8f16, !cast<Instruction>(NAME # _H)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv4f16, op, nxv4i1, nxv4f16, !cast<Instruction>(NAME # _H)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv2f16, op, nxv2i1, nxv2f16, !cast<Instruction>(NAME # _H)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv4f32, op, nxv4i1, nxv4f32, !cast<Instruction>(NAME # _S)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv2f32, op, nxv2i1, nxv2f32, !cast<Instruction>(NAME # _S)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv2f64, op, nxv2i1, nxv2f64, !cast<Instruction>(NAME # _D)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv8f16, op, nxv8i1, nxv8f16, !cast<Instruction>(NAME # _H)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv4f16, op, nxv4i1, nxv4f16, !cast<Instruction>(NAME # _H)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv2f16, op, nxv2i1, nxv2f16, !cast<Instruction>(NAME # _H)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv4f32, op, nxv4i1, nxv4f32, !cast<Instruction>(NAME # _S)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv2f32, op, nxv2i1, nxv2f32, !cast<Instruction>(NAME # _S)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv2f64, op, nxv2i1, nxv2f64, !cast<Instruction>(NAME # _D)>;
}
multiclass sve_fp_un_pred_arit_hsd<SDPatternOperator op> {
diff --git a/llvm/test/CodeGen/AArch64/zeroing-forms-abs-neg.ll b/llvm/test/CodeGen/AArch64/zeroing-forms-abs-neg.ll
index 1caee994220f05..510d4576646f12 100644
--- a/llvm/test/CodeGen/AArch64/zeroing-forms-abs-neg.ll
+++ b/llvm/test/CodeGen/AArch64/zeroing-forms-abs-neg.ll
@@ -18,7 +18,7 @@ define <vscale x 2 x double> @test_svabs_f64_x_1(<vscale x 2 x i1> %pg, <vscale
; CHECK-2p2-NEXT: fabs z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fabs.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fabs.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x double> %0
}
@@ -34,7 +34,7 @@ define <vscale x 2 x double> @test_svabs_f64_x_2(<vscale x 2 x i1> %pg, double %
; CHECK-2p2-NEXT: fabs z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fabs.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fabs.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x double> %0
}
@@ -65,7 +65,7 @@ define <vscale x 4 x float> @test_svabs_f32_x_1(<vscale x 4 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: fabs z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fabs.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fabs.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x float> %0
}
@@ -81,7 +81,7 @@ define <vscale x 4 x float> @test_svabs_f32_x_2(<vscale x 4 x i1> %pg, double %z
; CHECK-2p2-NEXT: fabs z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fabs.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fabs.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x float> %0
}
@@ -112,7 +112,7 @@ define <vscale x 8 x half> @test_svabs_f16_x_1(<vscale x 8 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: fabs z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fabs.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fabs.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x half> %0
}
@@ -128,7 +128,7 @@ define <vscale x 8 x half> @test_svabs_f16_x_2(<vscale x 8 x i1> %pg, double %z0
; CHECK-2p2-NEXT: fabs z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fabs.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fabs.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x half> %0
}
@@ -159,7 +159,7 @@ define <vscale x 16 x i8> @test_svabs_s8_x_1(<vscale x 16 x i1> %pg, <vscale x 1
; CHECK-2p2-NEXT: abs z0.b, p0/z, z0.b
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.abs.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.abs.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
ret <vscale x 16 x i8> %0
}
@@ -175,8 +175,8 @@ define <vscale x 16 x i8> @test_svabs_s8_x_2(<vscale x 16 x i1> %pg, double %z0,
; CHECK-2p2-NEXT: abs z0.b, p0/z, z1.b
; CHECK-2p2-NEXT: ret
entry:
- %1 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.abs.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
- ret <vscale x 16 x i8> %1
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.abs.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
}
define <vscale x 16 x i8> @test_svabs_s8_z(<vscale x 16 x i1> %pg, double %z0, <vscale x 16 x i8> %x) {
@@ -191,8 +191,8 @@ define <vscale x 16 x i8> @test_svabs_s8_z(<vscale x 16 x i1> %pg, double %z0, <
; CHECK-2p2-NEXT: abs z0.b, p0/z, z1.b
; CHECK-2p2-NEXT: ret
entry:
- %1 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.abs.nxv16i8(<vscale x 16 x i8> zeroinitializer, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
- ret <vscale x 16 x i8> %1
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.abs.nxv16i8(<vscale x 16 x i8> zeroinitializer, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
}
define <vscale x 8 x i16> @test_svabs_s16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %x) {
@@ -206,7 +206,7 @@ define <vscale x 8 x i16> @test_svabs_s16_x_1(<vscale x 8 x i1> %pg, <vscale x 8
; CHECK-2p2-NEXT: abs z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.abs.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.abs.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x i16> %0
}
@@ -222,7 +222,7 @@ define <vscale x 8 x i16> @test_svabs_s16_x_2(<vscale x 8 x i1> %pg, double %z0,
; CHECK-2p2-NEXT: abs z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.abs.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.abs.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x i16> %0
}
@@ -253,7 +253,7 @@ define <vscale x 4 x i32> @test_svabs_s32_x_1(<vscale x 4 x i1> %pg, <vscale x 4
; CHECK-2p2-NEXT: abs z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.abs.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.abs.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -269,7 +269,7 @@ define <vscale x 4 x i32> @test_svabs_s32_x_2(<vscale x 4 x i1> %pg, double %z0,
; CHECK-2p2-NEXT: abs z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.abs.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.abs.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -300,7 +300,7 @@ define <vscale x 2 x i64> @test_svabs_s64_x_1(<vscale x 2 x i1> %pg, <vscale x 2
; CHECK-2p2-NEXT: abs z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.abs.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.abs.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -316,7 +316,7 @@ define <vscale x 2 x i64> @test_svabs_s64_x_2(<vscale x 2 x i1> %pg, double %z0,
; CHECK-2p2-NEXT: abs z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.abs.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.abs.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -347,7 +347,7 @@ define <vscale x 2 x double> @test_svneg_f64_x_1(<vscale x 2 x i1> %pg, <vscale
; CHECK-2p2-NEXT: fneg z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fneg.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fneg.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x double> %0
}
@@ -363,7 +363,7 @@ define <vscale x 2 x double> @test_svneg_f64_x_2(<vscale x 2 x i1> %pg, double %
; CHECK-2p2-NEXT: fneg z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fneg.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fneg.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x double> %0
}
@@ -394,7 +394,7 @@ define <vscale x 4 x float> @test_svneg_f32_x_1(<vscale x 4 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: fneg z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fneg.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fneg.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x float> %0
}
@@ -410,7 +410,7 @@ define <vscale x 4 x float> @test_svneg_f32_x_2(<vscale x 4 x i1> %pg, double %z
; CHECK-2p2-NEXT: fneg z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fneg.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fneg.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x float> %0
}
@@ -441,7 +441,7 @@ define <vscale x 8 x half> @test_svneg_f16_x_1(<vscale x 8 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: fneg z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fneg.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fneg.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x half> %0
}
@@ -457,7 +457,7 @@ define <vscale x 8 x half> @test_svneg_f16_x_2(<vscale x 8 x i1> %pg, double %z0
; CHECK-2p2-NEXT: fneg z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fneg.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fneg.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x half> %0
}
@@ -488,7 +488,7 @@ define <vscale x 16 x i8> @test_svneg_s8_x_1(<vscale x 16 x i1> %pg, <vscale x 1
; CHECK-2p2-NEXT: neg z0.b, p0/z, z0.b
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.neg.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.neg.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
ret <vscale x 16 x i8> %0
}
@@ -504,8 +504,8 @@ define <vscale x 16 x i8> @test_svneg_s8_x_2(<vscale x 16 x i1> %pg, double %z0,
; CHECK-2p2-NEXT: neg z0.b, p0/z, z1.b
; CHECK-2p2-NEXT: ret
entry:
- %1 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.neg.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
- ret <vscale x 16 x i8> %1
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.neg.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
}
define <vscale x 16 x i8> @test_svneg_s8_z(<vscale x 16 x i1> %pg, double %z0, <vscale x 16 x i8> %x) {
@@ -520,8 +520,8 @@ define <vscale x 16 x i8> @test_svneg_s8_z(<vscale x 16 x i1> %pg, double %z0, <
; CHECK-2p2-NEXT: neg z0.b, p0/z, z1.b
; CHECK-2p2-NEXT: ret
entry:
- %1 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.neg.nxv16i8(<vscale x 16 x i8> zeroinitializer, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
- ret <vscale x 16 x i8> %1
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.neg.nxv16i8(<vscale x 16 x i8> zeroinitializer, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
}
define <vscale x 8 x i16> @test_svneg_s16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %x) {
@@ -535,7 +535,7 @@ define <vscale x 8 x i16> @test_svneg_s16_x_1(<vscale x 8 x i1> %pg, <vscale x 8
; CHECK-2p2-NEXT: neg z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.neg.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.neg.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x i16> %0
}
@@ -551,7 +551,7 @@ define <vscale x 8 x i16> @test_svneg_s16_x_2(<vscale x 8 x i1> %pg, double %z0,
; CHECK-2p2-NEXT: neg z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.neg.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.neg.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x i16> %0
}
@@ -582,7 +582,7 @@ define <vscale x 4 x i32> @test_svneg_s32_x_1(<vscale x 4 x i1> %pg, <vscale x 4
; CHECK-2p2-NEXT: neg z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.neg.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.neg.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -598,7 +598,7 @@ define <vscale x 4 x i32> @test_svneg_s32_x_2(<vscale x 4 x i1> %pg, double %z0,
; CHECK-2p2-NEXT: neg z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.neg.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.neg.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -629,7 +629,7 @@ define <vscale x 2 x i64> @test_svneg_s64_x_1(<vscale x 2 x i1> %pg, <vscale x 2
; CHECK-2p2-NEXT: neg z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.neg.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.neg.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -645,7 +645,7 @@ define <vscale x 2 x i64> @test_svneg_s64_x_2(<vscale x 2 x i1> %pg, double %z0,
; CHECK-2p2-NEXT: neg z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.neg.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.neg.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -664,3 +664,535 @@ entry:
%0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.neg.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
+
+define <vscale x 2 x double> @test_svfabs_f64_ptrue_u(double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svfabs_f64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: fabs z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfabs_f64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fabs z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fabs.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svfabs_f64_ptrue(double %z0, <vscale x 2 x double> %x, <vscale x 2 x double> %y) {
+; CHECK-LABEL: test_svfabs_f64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: fabs z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfabs_f64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fabs z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fabs.nxv2f64(<vscale x 2 x double> %x, <vscale x 2 x i1> %pg, <vscale x 2 x double> %y)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 4 x float> @test_svfabs_f32_ptrue_u(double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svfabs_f32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: fabs z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfabs_f32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: fabs z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fabs.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svfabs_f32_ptrue(double %z0, <vscale x 4 x float> %x, <vscale x 4 x float> %y) {
+; CHECK-LABEL: test_svfabs_f32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: fabs z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfabs_f32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: fabs z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fabs.nxv4f32(<vscale x 4 x float> %x, <vscale x 4 x i1> %pg, <vscale x 4 x float> %y)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 8 x half> @test_svfabs_f16_ptrue_u(double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svfabs_f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: fabs z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfabs_f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: fabs z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fabs.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svfabs_f16_ptrue(double %z0, <vscale x 8 x half> %x, <vscale x 8 x half> %y) {
+; CHECK-LABEL: test_svfabs_f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: fabs z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfabs_f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: fabs z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fabs.nxv8f16(<vscale x 8 x half> %x, <vscale x 8 x i1> %pg, <vscale x 8 x half> %y)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 16 x i8> @test_svabs_s8_ptrue_u(double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svabs_s8_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.b
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: abs z0.b, p0/m, z1.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svabs_s8_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.b
+; CHECK-2p2-NEXT: abs z0.b, p0/z, z1.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.abs.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svabs_s8_ptrue(double %z0, <vscale x 16 x i8> %x, <vscale x 16 x i8> %y) {
+; CHECK-LABEL: test_svabs_s8_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.b
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: abs z0.b, p0/m, z2.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svabs_s8_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.b
+; CHECK-2p2-NEXT: abs z0.b, p0/z, z2.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.abs.nxv16i8(<vscale x 16 x i8> %x, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %y)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 8 x i16> @test_svabs_s16_ptrue_u(double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svabs_s16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: abs z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svabs_s16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: abs z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nx84i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.abs.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svabs_s16_ptrue(double %z0, <vscale x 8 x i16> %x, <vscale x 8 x i16> %y) {
+; CHECK-LABEL: test_svabs_s16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: abs z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svabs_s16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: abs z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.abs.nxv8i16(<vscale x 8 x i16> %x, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %y)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svabs_s32_ptrue_u(double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svabs_s32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: abs z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svabs_s32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: abs z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.abs.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svabs_s32_ptrue(double %z0, <vscale x 4 x i32> %x, <vscale x 4 x i32> %y) {
+; CHECK-LABEL: test_svabs_s32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: abs z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svabs_s32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: abs z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.abs.nxv4i32(<vscale x 4 x i32> %x, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %y)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svabs_s64_ptrue_u(double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svabs_s64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: abs z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svabs_s64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: abs z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.abs.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svabs_s64_ptrue(double %z0, <vscale x 2 x i64> %x, <vscale x 2 x i64> %y) {
+; CHECK-LABEL: test_svabs_s64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: abs z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svabs_s64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: abs z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.abs.nxv2i64(<vscale x 2 x i64> %x, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %y)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x double> @test_svfneg_f64_ptrue_u(double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svfneg_f64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: fneg z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfneg_f64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fneg z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fneg.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svfneg_f64_ptrue(double %z0, <vscale x 2 x double> %x, <vscale x 2 x double> %y) {
+; CHECK-LABEL: test_svfneg_f64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: fneg z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfneg_f64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fneg z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fneg.nxv2f64(<vscale x 2 x double> %x, <vscale x 2 x i1> %pg, <vscale x 2 x double> %y)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 4 x float> @test_svfneg_f32_ptrue_u(double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svfneg_f32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: fneg z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfneg_f32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: fneg z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fneg.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svfneg_f32_ptrue(double %z0, <vscale x 4 x float> %x, <vscale x 4 x float> %y) {
+; CHECK-LABEL: test_svfneg_f32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: fneg z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfneg_f32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: fneg z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fneg.nxv4f32(<vscale x 4 x float> %x, <vscale x 4 x i1> %pg, <vscale x 4 x float> %y)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 8 x half> @test_svfneg_f16_ptrue_u(double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svfneg_f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: fneg z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfneg_f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: fneg z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fneg.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svfneg_f16_ptrue(double %z0, <vscale x 8 x half> %x, <vscale x 8 x half> %y) {
+; CHECK-LABEL: test_svfneg_f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: fneg z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfneg_f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: fneg z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fneg.nxv8f16(<vscale x 8 x half> %x, <vscale x 8 x i1> %pg, <vscale x 8 x half> %y)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 16 x i8> @test_svneg_s8_ptrue_u(double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svneg_s8_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.b
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: neg z0.b, p0/m, z1.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svneg_s8_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.b
+; CHECK-2p2-NEXT: neg z0.b, p0/z, z1.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.neg.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svneg_s8_ptrue(double %z0, <vscale x 16 x i8> %x, <vscale x 16 x i8> %y) {
+; CHECK-LABEL: test_svneg_s8_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.b
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: neg z0.b, p0/m, z2.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svneg_s8_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.b
+; CHECK-2p2-NEXT: neg z0.b, p0/z, z2.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.neg.nxv16i8(<vscale x 16 x i8> %x, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %y)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 8 x i16> @test_svneg_s16_ptrue_u(double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svneg_s16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: neg z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svneg_s16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: neg z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.neg.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svneg_s16_ptrue(double %z0, <vscale x 8 x i16> %x, <vscale x 8 x i16> %y) {
+; CHECK-LABEL: test_svneg_s16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: neg z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svneg_s16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: neg z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.neg.nxv8i16(<vscale x 8 x i16> %x, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %y)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svneg_s32_ptrue_u(double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svneg_s32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: neg z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svneg_s32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: neg z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.neg.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svneg_s32_ptrue(double %z0, <vscale x 4 x i32> %x, <vscale x 4 x i32> %y) {
+; CHECK-LABEL: test_svneg_s32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: neg z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svneg_s32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: neg z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.neg.nxv4i32(<vscale x 4 x i32> %x, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %y)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svneg_s64_ptrue_u(double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svneg_s64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: neg z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svneg_s64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: neg z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.neg.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svneg_s64_ptrue(double %z0, <vscale x 2 x i64> %x, <vscale x 2 x i64> %y) {
+; CHECK-LABEL: test_svneg_s64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: neg z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svneg_s64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: neg z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.neg.nxv2i64(<vscale x 2 x i64> %x, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %y)
+ ret <vscale x 2 x i64> %0
+}
diff --git a/llvm/test/CodeGen/AArch64/zeroing-forms-fcvt-bfcvt.ll b/llvm/test/CodeGen/AArch64/zeroing-forms-fcvt-bfcvt.ll
index cf9ac49ca7b236..855bf9a3b3c491 100644
--- a/llvm/test/CodeGen/AArch64/zeroing-forms-fcvt-bfcvt.ll
+++ b/llvm/test/CodeGen/AArch64/zeroing-forms-fcvt-bfcvt.ll
@@ -18,7 +18,7 @@ define <vscale x 8 x half> @test_svcvt_f16_f32_x_1(<vscale x 4 x i1> %pg, <vscal
; CHECK-2p2-NEXT: fcvt z0.h, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fcvt.f16f32(<vscale x 8 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fcvt.f16f32(<vscale x 8 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 8 x half> %0
}
@@ -33,7 +33,7 @@ define <vscale x 8 x half> @test_svcvt_f16_f32_x_2(<vscale x 4 x i1> %pg, double
; CHECK-2p2-NEXT: fcvt z0.h, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fcvt.f16f32(<vscale x 8 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fcvt.f16f32(<vscale x 8 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 8 x half> %0
}
@@ -64,7 +64,7 @@ define <vscale x 8 x bfloat> @test_svcvt_bf16_f32_x_1(<vscale x 4 x i1> %pg, <vs
; CHECK-2p2-NEXT: bfcvt z0.h, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x bfloat> @llvm.aarch64.sve.fcvt.bf16f32.v2(<vscale x 8 x bfloat> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 8 x bfloat> @llvm.aarch64.sve.fcvt.bf16f32.v2(<vscale x 8 x bfloat> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 8 x bfloat> %0
}
@@ -79,7 +79,7 @@ define <vscale x 8 x bfloat> @test_svcvt_bf16_f32_x_2(<vscale x 4 x i1> %pg, dou
; CHECK-2p2-NEXT: bfcvt z0.h, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x bfloat> @llvm.aarch64.sve.fcvt.bf16f32.v2(<vscale x 8 x bfloat> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 8 x bfloat> @llvm.aarch64.sve.fcvt.bf16f32.v2(<vscale x 8 x bfloat> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 8 x bfloat> %0
}
@@ -110,7 +110,7 @@ define <vscale x 8 x half> @test_svcvt_f16_f64_x_1(<vscale x 2 x i1> %pg, <vscal
; CHECK-2p2-NEXT: fcvt z0.h, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fcvt.f16f64(<vscale x 8 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fcvt.f16f64(<vscale x 8 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 8 x half> %0
}
@@ -125,7 +125,7 @@ define <vscale x 8 x half> @test_svcvt_f16_f64_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: fcvt z0.h, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fcvt.f16f64(<vscale x 8 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fcvt.f16f64(<vscale x 8 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 8 x half> %0
}
@@ -156,7 +156,7 @@ define <vscale x 4 x float> @test_svcvt_f32_f64_x_1(<vscale x 2 x i1> %pg, <vsca
; CHECK-2p2-NEXT: fcvt z0.s, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fcvt.f32f64(<vscale x 4 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fcvt.f32f64(<vscale x 4 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 4 x float> %0
}
@@ -171,7 +171,7 @@ define <vscale x 4 x float> @test_svcvt_f32_f64_x_2(<vscale x 2 x i1> %pg, doubl
; CHECK-2p2-NEXT: fcvt z0.s, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fcvt.f32f64(<vscale x 4 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fcvt.f32f64(<vscale x 4 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 4 x float> %0
}
@@ -202,7 +202,7 @@ define <vscale x 4 x float> @test_svcvt_f32_f16_x_1(<vscale x 4 x i1> %pg, <vsca
; CHECK-2p2-NEXT: fcvt z0.s, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fcvt.f32f16(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fcvt.f32f16(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 4 x float> %0
}
@@ -217,7 +217,7 @@ define <vscale x 4 x float> @test_svcvt_f32_f16_x_2(<vscale x 4 x i1> %pg, doubl
; CHECK-2p2-NEXT: fcvt z0.s, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fcvt.f32f16(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fcvt.f32f16(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 4 x float> %0
}
@@ -248,7 +248,7 @@ define <vscale x 2 x double> @test_svcvt_f64_f16_x_1(<vscale x 2 x i1> %pg, <vsc
; CHECK-2p2-NEXT: fcvt z0.d, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fcvt.f64f16(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fcvt.f64f16(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 2 x double> %0
}
@@ -263,7 +263,7 @@ define <vscale x 2 x double> @test_svcvt_f64_f16_x_2(<vscale x 2 x i1> %pg, doub
; CHECK-2p2-NEXT: fcvt z0.d, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fcvt.f64f16(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fcvt.f64f16(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 2 x double> %0
}
@@ -294,7 +294,7 @@ define <vscale x 2 x double> @test_svcvt_f64_f32_x_1(<vscale x 2 x i1> %pg, <vsc
; CHECK-2p2-NEXT: fcvt z0.d, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fcvt.f64f32(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fcvt.f64f32(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 2 x double> %0
}
@@ -309,7 +309,7 @@ define <vscale x 2 x double> @test_svcvt_f64_f32_x_2(<vscale x 2 x i1> %pg, doub
; CHECK-2p2-NEXT: fcvt z0.d, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fcvt.f64f32(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fcvt.f64f32(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 2 x double> %0
}
@@ -328,3 +328,262 @@ entry:
%0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fcvt.f64f32(<vscale x 2 x double> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 2 x double> %0
}
+
+define <vscale x 8 x half> @test_svcvt_f16_f32_ptrue_u(double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svcvt_f16_f32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: fcvt z0.h, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_f16_f32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: fcvt z0.h, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fcvt.f16f32(<vscale x 8 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svcvt_f16_f32_ptrue(double %z0, <vscale x 8 x half> %x, <vscale x 4 x float> %y ) {
+; CHECK-LABEL: test_svcvt_f16_f32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: fcvt z0.h, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_f16_f32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: fcvt z0.h, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fcvt.f16f32(<vscale x 8 x half> %x, <vscale x 4 x i1> %pg, <vscale x 4 x float> %y)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x bfloat> @test_svcvt_bf16_f32_ptrue_u(double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svcvt_bf16_f32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: bfcvt z0.h, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_bf16_f32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: bfcvt z0.h, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 8 x bfloat> @llvm.aarch64.sve.fcvt.bf16f32.v2(<vscale x 8 x bfloat> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 8 x bfloat> %0
+}
+
+define <vscale x 8 x bfloat> @test_svcvt_bf16_f32_ptrue(double %z0, <vscale x 8 x bfloat> %x, <vscale x 4 x float> %y ) {
+; CHECK-LABEL: test_svcvt_bf16_f32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: bfcvt z0.h, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_bf16_f32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: bfcvt z0.h, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 8 x bfloat> @llvm.aarch64.sve.fcvt.bf16f32.v2(<vscale x 8 x bfloat> %x, <vscale x 4 x i1> %pg, <vscale x 4 x float> %y)
+ ret <vscale x 8 x bfloat> %0
+}
+
+define <vscale x 8 x half> @test_svcvt_f16_f64_ptrue_u(double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svcvt_f16_f64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: fcvt z0.h, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_f16_f64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fcvt z0.h, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fcvt.f16f64(<vscale x 8 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svcvt_f16_f64_ptrue(double %z0, <vscale x 8 x half> %x, <vscale x 2 x double> %y ) {
+; CHECK-LABEL: test_svcvt_f16_f64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: fcvt z0.h, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_f16_f64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fcvt z0.h, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fcvt.f16f64(<vscale x 8 x half> %x, <vscale x 2 x i1> %pg, <vscale x 2 x double> %y)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 4 x float> @test_svcvt_f32_f64_ptrue_u(double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svcvt_f32_f64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: fcvt z0.s, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_f32_f64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fcvt z0.s, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fcvt.f32f64(<vscale x 4 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svcvt_f32_f64_ptrue(double %z0, <vscale x 4 x float> %x, <vscale x 2 x double> %y ) {
+; CHECK-LABEL: test_svcvt_f32_f64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: fcvt z0.s, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_f32_f64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fcvt z0.s, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fcvt.f32f64(<vscale x 4 x float> %x, <vscale x 2 x i1> %pg, <vscale x 2 x double> %y)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svcvt_f32_f16_ptrue_u(double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svcvt_f32_f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: fcvt z0.s, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_f32_f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: fcvt z0.s, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fcvt.f32f16(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svcvt_f32_f16_ptrue(double %z0, <vscale x 4 x float> %x, <vscale x 8 x half> %y ) {
+; CHECK-LABEL: test_svcvt_f32_f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: fcvt z0.s, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_f32_f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: fcvt z0.s, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fcvt.f32f16(<vscale x 4 x float> %x, <vscale x 4 x i1> %pg, <vscale x 8 x half> %y)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 2 x double> @test_svcvt_f64_f16_ptrue_u(double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svcvt_f64_f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: fcvt z0.d, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_f64_f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fcvt z0.d, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fcvt.f64f16(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svcvt_f64_f16_ptrue(double %z0, <vscale x 2 x double> %x, <vscale x 8 x half> %y ) {
+; CHECK-LABEL: test_svcvt_f64_f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: fcvt z0.d, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_f64_f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fcvt z0.d, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fcvt.f64f16(<vscale x 2 x double> %x, <vscale x 2 x i1> %pg, <vscale x 8 x half> %y)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svcvt_f64_f32_ptrue_u(double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svcvt_f64_f32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: fcvt z0.d, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_f64_f32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fcvt z0.d, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fcvt.f64f32(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svcvt_f64_f32_ptrue(double %z0, <vscale x 2 x double> %x, <vscale x 4 x float> %y ) {
+; CHECK-LABEL: test_svcvt_f64_f32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: fcvt z0.d, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_f64_f32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fcvt z0.d, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fcvt.f64f32(<vscale x 2 x double> %x, <vscale x 2 x i1> %pg, <vscale x 4 x float> %y)
+ ret <vscale x 2 x double> %0
+}
diff --git a/llvm/test/CodeGen/AArch64/zeroing-forms-fcvtlt-fcvtx.ll b/llvm/test/CodeGen/AArch64/zeroing-forms-fcvtlt-fcvtx.ll
index 60879b1529230f..c7431e11c21ca3 100644
--- a/llvm/test/CodeGen/AArch64/zeroing-forms-fcvtlt-fcvtx.ll
+++ b/llvm/test/CodeGen/AArch64/zeroing-forms-fcvtlt-fcvtx.ll
@@ -18,7 +18,7 @@ define <vscale x 4 x float> @test_svcvtlt_f32_f16_x_1(<vscale x 4 x i1> %pg, <vs
; CHECK-2p2-NEXT: fcvtlt z0.s, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fcvtlt.f32f16(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fcvtlt.f32f16(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 4 x float> %0
}
@@ -33,7 +33,7 @@ define <vscale x 4 x float> @test_svcvtlt_f32_f16_x_2(<vscale x 4 x i1> %pg, dou
; CHECK-2p2-NEXT: fcvtlt z0.s, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fcvtlt.f32f16(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fcvtlt.f32f16(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 4 x float> %0
}
@@ -64,7 +64,7 @@ define <vscale x 2 x double> @test_svcvtlt_f64_f32_x_1(<vscale x 2 x i1> %pg, <v
; CHECK-2p2-NEXT: fcvtlt z0.d, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fcvtlt.f64f32(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fcvtlt.f64f32(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 2 x double> %0
}
@@ -79,7 +79,7 @@ define <vscale x 2 x double> @test_svcvtlt_f64_f32_x_2(<vscale x 2 x i1> %pg, do
; CHECK-2p2-NEXT: fcvtlt z0.d, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fcvtlt.f64f32(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fcvtlt.f64f32(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 2 x double> %0
}
@@ -110,7 +110,7 @@ define <vscale x 4 x float> @test_svcvtx_f32_f64_x_1(<vscale x 2 x i1> %pg, <vsc
; CHECK-2p2-NEXT: fcvtx z0.s, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fcvtx.f32f64(<vscale x 4 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fcvtx.f32f64(<vscale x 4 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 4 x float> %0
}
@@ -125,7 +125,7 @@ define <vscale x 4 x float> @test_svcvtx_f32_f64_x_2(<vscale x 2 x i1> %pg, doub
; CHECK-2p2-NEXT: fcvtx z0.s, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fcvtx.f32f64(<vscale x 4 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fcvtx.f32f64(<vscale x 4 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 4 x float> %0
}
@@ -144,3 +144,114 @@ entry:
%0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fcvtx.f32f64(<vscale x 4 x float> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 4 x float> %0
}
+
+define <vscale x 4 x float> @test_svcvtlt_f32_f16_ptrue_u(double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svcvtlt_f32_f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: fcvtlt z0.s, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvtlt_f32_f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: fcvtlt z0.s, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fcvtlt.f32f16(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svcvtlt_f32_f16_ptrue(double %z0, <vscale x 4 x float> %x, <vscale x 8 x half> %y) {
+; CHECK-LABEL: test_svcvtlt_f32_f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: fcvtlt z0.s, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvtlt_f32_f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: fcvtlt z0.s, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fcvtlt.f32f16(<vscale x 4 x float> %x, <vscale x 4 x i1> %pg, <vscale x 8 x half> %y)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 2 x double> @test_svcvtlt_f64_f32_ptrue_u(double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svcvtlt_f64_f32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: fcvtlt z0.d, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvtlt_f64_f32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fcvtlt z0.d, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fcvtlt.f64f32(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svcvtlt_f64_f32_ptrue(double %z0, <vscale x 2 x double> %x, <vscale x 4 x float> %y) {
+; CHECK-LABEL: test_svcvtlt_f64_f32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: fcvtlt z0.d, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvtlt_f64_f32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fcvtlt z0.d, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fcvtlt.f64f32(<vscale x 2 x double> %x, <vscale x 2 x i1> %pg, <vscale x 4 x float> %y)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 4 x float> @test_svcvtx_f32_f64_ptrue_u(double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svcvtx_f32_f64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: fcvtx z0.s, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvtx_f32_f64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fcvtx z0.s, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fcvtx.f32f64(<vscale x 4 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svcvtx_f32_f64_ptrue(double %z0, <vscale x 4 x float> %x, <vscale x 2 x double> %y) {
+; CHECK-LABEL: test_svcvtx_f32_f64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: fcvtx z0.s, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvtx_f32_f64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fcvtx z0.s, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fcvtx.f32f64(<vscale x 4 x float> %x, <vscale x 2 x i1> %pg, <vscale x 2 x double> %y)
+ ret <vscale x 4 x float> %0
+}
diff --git a/llvm/test/CodeGen/AArch64/zeroing-forms-fcvtzsu.ll b/llvm/test/CodeGen/AArch64/zeroing-forms-fcvtzsu.ll
index b8b36d390330af..7259502bf44002 100644
--- a/llvm/test/CodeGen/AArch64/zeroing-forms-fcvtzsu.ll
+++ b/llvm/test/CodeGen/AArch64/zeroing-forms-fcvtzsu.ll
@@ -18,7 +18,7 @@ define <vscale x 4 x i32> @test_fcvtzs_s32_f64_x_1(<vscale x 2 x i1> %pg, <vscal
; CHECK-2p2-NEXT: fcvtzs z0.s, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzs.i32f64(<vscale x 4 x i32> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzs.i32f64(<vscale x 4 x i32> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 4 x i32> %0
}
@@ -33,7 +33,7 @@ define <vscale x 4 x i32> @test_fcvtzs_s32_f64_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: fcvtzs z0.s, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzs.i32f64(<vscale x 4 x i32> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzs.i32f64(<vscale x 4 x i32> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 4 x i32> %0
}
@@ -64,7 +64,7 @@ define <vscale x 2 x i64> @test_fcvtzs_s64_f32_x_1(<vscale x 2 x i1> %pg, <vscal
; CHECK-2p2-NEXT: fcvtzs z0.d, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzs.i64f32(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzs.i64f32(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 2 x i64> %0
}
@@ -79,7 +79,7 @@ define <vscale x 2 x i64> @test_fcvtzs_s64_f32_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: fcvtzs z0.d, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzs.i64f32(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzs.i64f32(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 2 x i64> %0
}
@@ -110,7 +110,7 @@ define <vscale x 4 x i32> @test_fcvtzs_s32_f16_x_1(<vscale x 4 x i1> %pg, <vscal
; CHECK-2p2-NEXT: fcvtzs z0.s, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzs.i32f16(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzs.i32f16(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 4 x i32> %0
}
@@ -125,7 +125,7 @@ define <vscale x 4 x i32> @test_fcvtzs_s32_f16_x_2(<vscale x 4 x i1> %pg, double
; CHECK-2p2-NEXT: fcvtzs z0.s, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzs.i32f16(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzs.i32f16(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 4 x i32> %0
}
@@ -156,7 +156,7 @@ define <vscale x 2 x i64> @test_fcvtzs_s64_f16_x_1(<vscale x 2 x i1> %pg, <vscal
; CHECK-2p2-NEXT: fcvtzs z0.d, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzs.i64f16(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzs.i64f16(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 2 x i64> %0
}
@@ -171,7 +171,7 @@ define <vscale x 2 x i64> @test_fcvtzs_s64_f16_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: fcvtzs z0.d, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzs.i64f16(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzs.i64f16(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 2 x i64> %0
}
@@ -202,7 +202,7 @@ define <vscale x 4 x i32> @test_fcvtzu_u32_f64_x_1(<vscale x 2 x i1> %pg, <vscal
; CHECK-2p2-NEXT: fcvtzu z0.s, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzu.i32f64(<vscale x 4 x i32> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzu.i32f64(<vscale x 4 x i32> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 4 x i32> %0
}
@@ -217,7 +217,7 @@ define <vscale x 4 x i32> @test_fcvtzu_u32_f64_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: fcvtzu z0.s, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzu.i32f64(<vscale x 4 x i32> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzu.i32f64(<vscale x 4 x i32> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 4 x i32> %0
}
@@ -248,7 +248,7 @@ define <vscale x 2 x i64> @test_fcvtzu_u64_f32_x_1(<vscale x 2 x i1> %pg, <vscal
; CHECK-2p2-NEXT: fcvtzu z0.d, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzu.i64f32(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzu.i64f32(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 2 x i64> %0
}
@@ -263,7 +263,7 @@ define <vscale x 2 x i64> @test_fcvtzu_u64_f32_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: fcvtzu z0.d, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzu.i64f32(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzu.i64f32(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 2 x i64> %0
}
@@ -294,7 +294,7 @@ define <vscale x 4 x i32> @test_fcvtzu_u32_f16_x_1(<vscale x 4 x i1> %pg, <vscal
; CHECK-2p2-NEXT: fcvtzu z0.s, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzu.i32f16(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzu.i32f16(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 4 x i32> %0
}
@@ -309,7 +309,7 @@ define <vscale x 4 x i32> @test_fcvtzu_u32_f16_x_2(<vscale x 4 x i1> %pg, double
; CHECK-2p2-NEXT: fcvtzu z0.s, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzu.i32f16(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzu.i32f16(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 4 x i32> %0
}
@@ -340,7 +340,7 @@ define <vscale x 2 x i64> @test_fcvtzu_u64_f16_x_1(<vscale x 2 x i1> %pg, <vscal
; CHECK-2p2-NEXT: fcvtzu z0.d, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzu.i64f16(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzu.i64f16(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 2 x i64> %0
}
@@ -355,7 +355,7 @@ define <vscale x 2 x i64> @test_fcvtzu_u64_f16_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: fcvtzu z0.d, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzu.i64f16(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzu.i64f16(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 2 x i64> %0
}
@@ -387,7 +387,7 @@ define <vscale x 8 x i16> @test_svcvt_s16_f16_x_1(<vscale x 8 x i1> %pg, <vscale
; CHECK-2p2-NEXT: fcvtzs z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.fcvtzs.nxv8i16.nxv8f16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.fcvtzs.nxv8i16.nxv8f16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x i16> %0
}
@@ -403,7 +403,7 @@ define <vscale x 8 x i16> @test_svcvt_s16_f16_x_2(<vscale x 8 x i1> %pg, double
; CHECK-2p2-NEXT: fcvtzs z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.fcvtzs.nxv8i16.nxv8f16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.fcvtzs.nxv8i16.nxv8f16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x i16> %0
}
@@ -434,7 +434,7 @@ define <vscale x 8 x i16> @test_svcvt_u16_f16_x_1(<vscale x 8 x i1> %pg, <vscale
; CHECK-2p2-NEXT: fcvtzu z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.fcvtzu.nxv8i16.nxv8f16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.fcvtzu.nxv8i16.nxv8f16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x i16> %0
}
@@ -450,7 +450,7 @@ define <vscale x 8 x i16> @test_svcvt_u16_f16_x_2(<vscale x 8 x i1> %pg, double
; CHECK-2p2-NEXT: fcvtzu z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.fcvtzu.nxv8i16.nxv8f16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.fcvtzu.nxv8i16.nxv8f16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x i16> %0
}
@@ -481,7 +481,7 @@ define <vscale x 4 x i32> @test_svcvt_s32_f32_x_1(<vscale x 4 x i1> %pg, <vscale
; CHECK-2p2-NEXT: fcvtzs z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzs.nxv4i32.nxv4f32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzs.nxv4i32.nxv4f32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x i32> %0
}
@@ -497,7 +497,7 @@ define <vscale x 4 x i32> @test_svcvt_s32_f32_x_2(<vscale x 4 x i1> %pg, double
; CHECK-2p2-NEXT: fcvtzs z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzs.nxv4i32.nxv4f32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzs.nxv4i32.nxv4f32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x i32> %0
}
@@ -528,7 +528,7 @@ define <vscale x 4 x i32> @test_svcvt_u32_f32_x_1(<vscale x 4 x i1> %pg, <vscale
; CHECK-2p2-NEXT: fcvtzu z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzu.nxv4i32.nxv4f32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzu.nxv4i32.nxv4f32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x i32> %0
}
@@ -544,7 +544,7 @@ define <vscale x 4 x i32> @test_svcvt_u32_f32_x_2(<vscale x 4 x i1> %pg, double
; CHECK-2p2-NEXT: fcvtzu z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzu.nxv4i32.nxv4f32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzu.nxv4i32.nxv4f32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x i32> %0
}
@@ -575,7 +575,7 @@ define <vscale x 2 x i64> @test_svcvt_s64_f64_x_1(<vscale x 2 x i1> %pg, <vscale
; CHECK-2p2-NEXT: fcvtzs z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzs.nxv2i64.nxv2f64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzs.nxv2i64.nxv2f64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x i64> %0
}
@@ -591,7 +591,7 @@ define <vscale x 2 x i64> @test_svcvt_s64_f64_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: fcvtzs z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzs.nxv2i64.nxv2f64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzs.nxv2i64.nxv2f64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x i64> %0
}
@@ -622,7 +622,7 @@ define <vscale x 2 x i64> @test_svcvt_u64_f64_x_1(<vscale x 2 x i1> %pg, <vscale
; CHECK-2p2-NEXT: fcvtzu z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzu.nxv2i64.nxv2f64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzu.nxv2i64.nxv2f64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x i64> %0
}
@@ -638,7 +638,7 @@ define <vscale x 2 x i64> @test_svcvt_u64_f64_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: fcvtzu z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzu.nxv2i64.nxv2f64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzu.nxv2i64.nxv2f64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x i64> %0
}
@@ -657,3 +657,527 @@ entry:
%0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzu.nxv2i64.nxv2f64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x i64> %0
}
+
+define <vscale x 4 x i32> @test_fcvtzs_i32_f64_ptrue_u(double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_fcvtzs_i32_f64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: fcvtzs z0.s, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_fcvtzs_i32_f64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fcvtzs z0.s, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzs.i32f64(<vscale x 4 x i32> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_fcvtzs_i32_f64_ptrue(double %z0, <vscale x 4 x i32> %x, <vscale x 2 x double> %y) {
+; CHECK-LABEL: test_fcvtzs_i32_f64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: fcvtzs z0.s, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_fcvtzs_i32_f64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fcvtzs z0.s, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzs.i32f64(<vscale x 4 x i32> %x, <vscale x 2 x i1> %pg, <vscale x 2 x double> %y)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_fcvtzu_i32_f64_ptrue_u(double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_fcvtzu_i32_f64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: fcvtzu z0.s, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_fcvtzu_i32_f64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fcvtzu z0.s, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzu.i32f64(<vscale x 4 x i32> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_fcvtzu_i32_f64_ptrue(double %z0, <vscale x 4 x i32> %x, <vscale x 2 x double> %y) {
+; CHECK-LABEL: test_fcvtzu_i32_f64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: fcvtzu z0.s, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_fcvtzu_i32_f64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fcvtzu z0.s, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzu.i32f64(<vscale x 4 x i32> %x, <vscale x 2 x i1> %pg, <vscale x 2 x double> %y)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_fcvtzs_i64_f32_ptrue_u(double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_fcvtzs_i64_f32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: fcvtzs z0.d, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_fcvtzs_i64_f32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fcvtzs z0.d, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzs.i64f32(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_fcvtzs_i64_f32_ptrue(double %z0, <vscale x 2 x i64> %x, <vscale x 4 x float> %y) {
+; CHECK-LABEL: test_fcvtzs_i64_f32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: fcvtzs z0.d, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_fcvtzs_i64_f32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fcvtzs z0.d, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzs.i64f32(<vscale x 2 x i64> %x, <vscale x 2 x i1> %pg, <vscale x 4 x float> %y)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_fcvtzu_i64_f32_ptrue_u(double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_fcvtzu_i64_f32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: fcvtzu z0.d, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_fcvtzu_i64_f32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fcvtzu z0.d, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzu.i64f32(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_fcvtzu_i64_f32_ptrue(double %z0, <vscale x 2 x i64> %x, <vscale x 4 x float> %y) {
+; CHECK-LABEL: test_fcvtzu_i64_f32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: fcvtzu z0.d, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_fcvtzu_i64_f32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fcvtzu z0.d, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzu.i64f32(<vscale x 2 x i64> %x, <vscale x 2 x i1> %pg, <vscale x 4 x float> %y)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 4 x i32> @test_fcvtzs_i32_f16_ptrue_u(double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_fcvtzs_i32_f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: fcvtzs z0.s, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_fcvtzs_i32_f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: fcvtzs z0.s, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzs.i32f16(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_fcvtzs_i32_f16_ptrue(double %z0, <vscale x 4 x i32> %x, <vscale x 8 x half> %y) {
+; CHECK-LABEL: test_fcvtzs_i32_f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: fcvtzs z0.s, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_fcvtzs_i32_f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: fcvtzs z0.s, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzs.i32f16(<vscale x 4 x i32> %x, <vscale x 4 x i1> %pg, <vscale x 8 x half> %y)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_fcvtzu_i32_f16_ptrue_u(double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_fcvtzu_i32_f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: fcvtzu z0.s, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_fcvtzu_i32_f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: fcvtzu z0.s, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzu.i32f16(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_fcvtzu_i32_f16_ptrue(double %z0, <vscale x 4 x i32> %x, <vscale x 8 x half> %y) {
+; CHECK-LABEL: test_fcvtzu_i32_f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: fcvtzu z0.s, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_fcvtzu_i32_f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: fcvtzu z0.s, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzu.i32f16(<vscale x 4 x i32> %x, <vscale x 4 x i1> %pg, <vscale x 8 x half> %y)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_fcvtzs_i64_f16_ptrue_u(double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_fcvtzs_i64_f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: fcvtzs z0.d, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_fcvtzs_i64_f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fcvtzs z0.d, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzs.i64f16(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_fcvtzs_i64_f16_ptrue(double %z0, <vscale x 2 x i64> %x, <vscale x 8 x half> %y) {
+; CHECK-LABEL: test_fcvtzs_i64_f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: fcvtzs z0.d, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_fcvtzs_i64_f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fcvtzs z0.d, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzs.i64f16(<vscale x 2 x i64> %x, <vscale x 2 x i1> %pg, <vscale x 8 x half> %y)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_fcvtzu_i64_f16_ptrue_u(double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_fcvtzu_i64_f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: fcvtzu z0.d, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_fcvtzu_i64_f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fcvtzu z0.d, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzu.i64f16(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_fcvtzu_i64_f16_ptrue(double %z0, <vscale x 2 x i64> %x, <vscale x 8 x half> %y) {
+; CHECK-LABEL: test_fcvtzu_i64_f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: fcvtzu z0.d, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_fcvtzu_i64_f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fcvtzu z0.d, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzu.i64f16(<vscale x 2 x i64> %x, <vscale x 2 x i1> %pg, <vscale x 8 x half> %y)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 8 x i16> @test_fcvtzs_i16_f16_ptrue_u(double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_fcvtzs_i16_f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: fcvtzs z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_fcvtzs_i16_f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: fcvtzs z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.fcvtzs.nxv8i16.nxv8f16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_fcvtzs_i16_f16_ptrue(double %z0, <vscale x 8 x i16> %x, <vscale x 8 x half> %y) {
+; CHECK-LABEL: test_fcvtzs_i16_f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: fcvtzs z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_fcvtzs_i16_f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: fcvtzs z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.fcvtzs.nxv8i16.nxv8f16(<vscale x 8 x i16> %x, <vscale x 8 x i1> %pg, <vscale x 8 x half> %y)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_fcvtzu_i16_f16_ptrue_u(double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_fcvtzu_i16_f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: fcvtzu z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_fcvtzu_i16_f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: fcvtzu z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.fcvtzu.nxv8i16.nxv8f16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_fcvtzu_i16_f16_ptrue(double %z0, <vscale x 8 x i16> %x, <vscale x 8 x half> %y) {
+; CHECK-LABEL: test_fcvtzu_i16_f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: fcvtzu z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_fcvtzu_i16_f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: fcvtzu z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.fcvtzu.nxv8i16.nxv8f16(<vscale x 8 x i16> %x, <vscale x 8 x i1> %pg, <vscale x 8 x half> %y)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_fcvtzs_i32_f32_ptrue_u(double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_fcvtzs_i32_f32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: fcvtzs z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_fcvtzs_i32_f32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: fcvtzs z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzs.nxv4i32.nxv4f32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_fcvtzs_i32_f32_ptrue(double %z0, <vscale x 4 x i32> %x, <vscale x 4 x float> %y) {
+; CHECK-LABEL: test_fcvtzs_i32_f32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: fcvtzs z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_fcvtzs_i32_f32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: fcvtzs z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzs.nxv4i32.nxv4f32(<vscale x 4 x i32> %x, <vscale x 4 x i1> %pg, <vscale x 4 x float> %y)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_fcvtzu_i32_f32_ptrue_u(double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_fcvtzu_i32_f32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: fcvtzu z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_fcvtzu_i32_f32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: fcvtzu z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzu.nxv4i32.nxv4f32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_fcvtzu_i32_f32_ptrue(double %z0, <vscale x 4 x i32> %x, <vscale x 4 x float> %y) {
+; CHECK-LABEL: test_fcvtzu_i32_f32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: fcvtzu z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_fcvtzu_i32_f32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: fcvtzu z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.fcvtzu.nxv4i32.nxv4f32(<vscale x 4 x i32> %x, <vscale x 4 x i1> %pg, <vscale x 4 x float> %y)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_fcvtzs_i64_f64_ptrue_u(double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_fcvtzs_i64_f64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: fcvtzs z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_fcvtzs_i64_f64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fcvtzs z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzs.nxv2i64.nxv2f64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_fcvtzs_i64_f64_ptrue(double %z0, <vscale x 2 x i64> %x, <vscale x 2 x double> %y) {
+; CHECK-LABEL: test_fcvtzs_i64_f64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: fcvtzs z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_fcvtzs_i64_f64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fcvtzs z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzs.nxv2i64.nxv2f64(<vscale x 2 x i64> %x, <vscale x 2 x i1> %pg, <vscale x 2 x double> %y)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_fcvtzu_i64_f64_ptrue_u(double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_fcvtzu_i64_f64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: fcvtzu z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_fcvtzu_i64_f64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fcvtzu z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzu.nxv2i64.nxv2f64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_fcvtzu_i64_f64_ptrue(double %z0, <vscale x 2 x i64> %x, <vscale x 2 x double> %y) {
+; CHECK-LABEL: test_fcvtzu_i64_f64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: fcvtzu z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_fcvtzu_i64_f64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fcvtzu z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.fcvtzu.nxv2i64.nxv2f64(<vscale x 2 x i64> %x, <vscale x 2 x i1> %pg, <vscale x 2 x double> %y)
+ ret <vscale x 2 x i64> %0
+}
>From 335fe0bfa26145b4723dfff475d5c6c33e090e28 Mon Sep 17 00:00:00 2001
From: Momchil Velikov <momchil.velikov at arm.com>
Date: Tue, 19 Nov 2024 11:53:56 +0000
Subject: [PATCH 02/16] [AArch64] Generate zeroing forms of certain SVE2.2
instructions (5/11)
---
.../lib/Target/AArch64/AArch64SVEInstrInfo.td | 4 +-
llvm/lib/Target/AArch64/SVEInstrFormats.td | 11 +-
.../CodeGen/AArch64/zeroing-forms-uscvtf.ll | 658 ++++++++++++++++++
3 files changed, 670 insertions(+), 3 deletions(-)
create mode 100644 llvm/test/CodeGen/AArch64/zeroing-forms-uscvtf.ll
diff --git a/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td b/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
index 1e297eb5feb266..23d626fe786630 100644
--- a/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
+++ b/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
@@ -4277,8 +4277,8 @@ let Predicates = [HasSVE2p2orSME2p2] in {
defm FCVTZS_ZPzZ : sve_fp_z2op_p_zd_d<0b0, "fcvtzs", "int_aarch64_sve_fcvtzs", AArch64fcvtzs_mt>;
defm FCVTZU_ZPzZ : sve_fp_z2op_p_zd_d<0b1, "fcvtzu", "int_aarch64_sve_fcvtzu", AArch64fcvtzu_mt>;
// Integer convert to floating-point, zeroing predicate
- defm SCVTF_ZPzZ : sve_fp_z2op_p_zd_c<0b0, "scvtf">;
- defm UCVTF_ZPzZ : sve_fp_z2op_p_zd_c<0b1, "ucvtf">;
+ defm SCVTF_ZPzZ : sve_fp_z2op_p_zd_c<0b0, "scvtf", "int_aarch64_sve_scvtf", AArch64scvtf_mt>;
+ defm UCVTF_ZPzZ : sve_fp_z2op_p_zd_c<0b1, "ucvtf", "int_aarch64_sve_ucvtf", AArch64ucvtf_mt>;
// Signed integer base 2 logarithm of fp value, zeroing predicate
defm FLOGB_ZPzZ : sve_fp_z2op_p_zd_d_flogb<"flogb">;
diff --git a/llvm/lib/Target/AArch64/SVEInstrFormats.td b/llvm/lib/Target/AArch64/SVEInstrFormats.td
index a1bce3601aab60..027f83a6a50442 100644
--- a/llvm/lib/Target/AArch64/SVEInstrFormats.td
+++ b/llvm/lib/Target/AArch64/SVEInstrFormats.td
@@ -3323,7 +3323,7 @@ multiclass sve_fp_z2op_p_zd_d<bit U, string asm, string int_op, SDPatternOperato
defm : SVE_1_Op_PassthruUndefZero_Pat<nxv2i64, ir_op, nxv2i1, nxv2f64, !cast<Instruction>(NAME # _DtoD)>;
}
-multiclass sve_fp_z2op_p_zd_c<bit U, string asm> {
+multiclass sve_fp_z2op_p_zd_c<bit U, string asm, string int_op, SDPatternOperator ir_op> {
def _HtoH : sve_fp_z2op_p_zd<{ 0b011001, U }, asm, ZPR16, ZPR16>;
def _StoH : sve_fp_z2op_p_zd<{ 0b011010, U }, asm, ZPR32, ZPR16>;
def _StoS : sve_fp_z2op_p_zd<{ 0b101010, U }, asm, ZPR32, ZPR32>;
@@ -3331,6 +3331,15 @@ multiclass sve_fp_z2op_p_zd_c<bit U, string asm> {
def _DtoS : sve_fp_z2op_p_zd<{ 0b111010, U }, asm, ZPR64, ZPR32>;
def _DtoH : sve_fp_z2op_p_zd<{ 0b011011, U }, asm, ZPR64, ZPR16>;
def _DtoD : sve_fp_z2op_p_zd<{ 0b111011, U }, asm, ZPR64, ZPR64>;
+
+ def : SVE_3_Op_UndefZero_Pat<nxv4f32, !cast<SDPatternOperator>(int_op # _f32i64), nxv4f32, nxv2i1, nxv2i64, !cast<Instruction>(NAME # _DtoS)>;
+ def : SVE_3_Op_UndefZero_Pat<nxv2f64, !cast<SDPatternOperator>(int_op # _f64i32), nxv2f64, nxv2i1, nxv4i32, !cast<Instruction>(NAME # _StoD)>;
+ def : SVE_3_Op_UndefZero_Pat<nxv8f16, !cast<SDPatternOperator>(int_op # _f16i32), nxv8f16, nxv4i1, nxv4i32, !cast<Instruction>(NAME # _StoH)>;
+ def : SVE_3_Op_UndefZero_Pat<nxv8f16, !cast<SDPatternOperator>(int_op # _f16i64), nxv8f16, nxv2i1, nxv2i64, !cast<Instruction>(NAME # _DtoH)>;
+
+ def : SVE_1_Op_PassthruUndefZero_Pat<nxv8f16, ir_op, nxv8i1,nxv8i16, !cast<Instruction>(NAME # _HtoH)>;
+ def : SVE_1_Op_PassthruUndefZero_Pat<nxv4f32, ir_op, nxv4i1,nxv4i32, !cast<Instruction>(NAME # _StoS)>;
+ def : SVE_1_Op_PassthruUndefZero_Pat<nxv2f64, ir_op, nxv2i1,nxv2i64, !cast<Instruction>(NAME # _DtoD)>;
}
multiclass sve_fp_z2op_p_zd_d_flogb<string asm> {
diff --git a/llvm/test/CodeGen/AArch64/zeroing-forms-uscvtf.ll b/llvm/test/CodeGen/AArch64/zeroing-forms-uscvtf.ll
new file mode 100644
index 00000000000000..1ca3b1a6e31cb9
--- /dev/null
+++ b/llvm/test/CodeGen/AArch64/zeroing-forms-uscvtf.ll
@@ -0,0 +1,658 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
+; RUN: llc -mattr=+sve < %s | FileCheck %s
+; RUN: llc -mattr=+sve2p2 < %s | FileCheck %s -check-prefix CHECK-2p2
+
+; RUN: llc -mattr=+sme -force-streaming < %s | FileCheck %s
+; RUN: llc -mattr=+sme2p2 -force-streaming < %s | FileCheck %s -check-prefix CHECK-2p2
+
+target triple = "aarch64-linux"
+
+define <vscale x 4 x float> @test_scvtf_f32_s64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_scvtf_f32_s64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: scvtf z0.s, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_scvtf_f32_s64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: scvtf z0.s, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.scvtf.f32i64(<vscale x 4 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_scvtf_f32_s64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_scvtf_f32_s64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: scvtf z0.s, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_scvtf_f32_s64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: scvtf z0.s, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.scvtf.f32i64(<vscale x 4 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_scvtf_f32_s64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_scvtf_f32_s64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: scvtf z0.s, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_scvtf_f32_s64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: scvtf z0.s, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.scvtf.f32i64(<vscale x 4 x float> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 2 x double> @test_scvtf_f64_s32_x_1(<vscale x 2 x i1> %pg, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_scvtf_f64_s32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: scvtf z0.d, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_scvtf_f64_s32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: scvtf z0.d, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.scvtf.f64i32(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_scvtf_f64_s32_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_scvtf_f64_s32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: scvtf z0.d, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_scvtf_f64_s32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: scvtf z0.d, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.scvtf.f64i32(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_scvtf_f64_s32_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_scvtf_f64_s32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: scvtf z0.d, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_scvtf_f64_s32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: scvtf z0.d, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.scvtf.f64i32(<vscale x 2 x double> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 8 x half> @test_scvtf_f16_s32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_scvtf_f16_s32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: scvtf z0.h, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_scvtf_f16_s32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: scvtf z0.h, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.scvtf.f16i32(<vscale x 8 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_scvtf_f16_s32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_scvtf_f16_s32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: scvtf z0.h, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_scvtf_f16_s32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: scvtf z0.h, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.scvtf.f16i32(<vscale x 8 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_scvtf_f16_s32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_scvtf_f16_s32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: scvtf z0.h, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_scvtf_f16_s32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: scvtf z0.h, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.scvtf.f16i32(<vscale x 8 x half> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_scvtf_f16_s64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_scvtf_f16_s64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: scvtf z0.h, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_scvtf_f16_s64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: scvtf z0.h, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.scvtf.f16i64(<vscale x 8 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_scvtf_f16_s64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_scvtf_f16_s64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: scvtf z0.h, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_scvtf_f16_s64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: scvtf z0.h, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.scvtf.f16i64(<vscale x 8 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_scvtf_f16_s64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_scvtf_f16_s64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: scvtf z0.h, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_scvtf_f16_s64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: scvtf z0.h, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.scvtf.f16i64(<vscale x 8 x half> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 4 x float> @test_ucvtf_f32_u64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_ucvtf_f32_u64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ucvtf z0.s, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_ucvtf_f32_u64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ucvtf z0.s, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.ucvtf.f32i64(<vscale x 4 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_ucvtf_f32_u64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_ucvtf_f32_u64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ucvtf z0.s, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_ucvtf_f32_u64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ucvtf z0.s, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.ucvtf.f32i64(<vscale x 4 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_ucvtf_f32_u64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_ucvtf_f32_u64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: ucvtf z0.s, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_ucvtf_f32_u64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ucvtf z0.s, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.ucvtf.f32i64(<vscale x 4 x float> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 2 x double> @test_ucvtf_f64_u32_x_1(<vscale x 2 x i1> %pg, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_ucvtf_f64_u32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ucvtf z0.d, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_ucvtf_f64_u32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ucvtf z0.d, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.ucvtf.f64i32(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_ucvtf_f64_u32_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_ucvtf_f64_u32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ucvtf z0.d, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_ucvtf_f64_u32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ucvtf z0.d, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.ucvtf.f64i32(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_ucvtf_f64_u32_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_ucvtf_f64_u32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: ucvtf z0.d, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_ucvtf_f64_u32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ucvtf z0.d, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.ucvtf.f64i32(<vscale x 2 x double> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 8 x half> @test_ucvtf_f16_u32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_ucvtf_f16_u32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ucvtf z0.h, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_ucvtf_f16_u32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ucvtf z0.h, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.ucvtf.f16i32(<vscale x 8 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_ucvtf_f16_u32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_ucvtf_f16_u32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ucvtf z0.h, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_ucvtf_f16_u32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ucvtf z0.h, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.ucvtf.f16i32(<vscale x 8 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_ucvtf_f16_u32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_ucvtf_f16_u32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: ucvtf z0.h, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_ucvtf_f16_u32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ucvtf z0.h, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.ucvtf.f16i32(<vscale x 8 x half> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_ucvtf_f16_u64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_ucvtf_f16_u64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ucvtf z0.h, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_ucvtf_f16_u64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ucvtf z0.h, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.ucvtf.f16i64(<vscale x 8 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_ucvtf_f16_u64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_ucvtf_f16_u64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ucvtf z0.h, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_ucvtf_f16_u64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ucvtf z0.h, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.ucvtf.f16i64(<vscale x 8 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_ucvtf_f16_u64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_ucvtf_f16_u64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: ucvtf z0.h, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_ucvtf_f16_u64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ucvtf z0.h, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.ucvtf.f16i64(<vscale x 8 x half> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svcvt_f16_s16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svcvt_f16_s16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: scvtf z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_f16_s16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: scvtf z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.scvtf.nxv8f16.nxv8i16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svcvt_f16_s16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svcvt_f16_s16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: scvtf z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_f16_s16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: scvtf z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.scvtf.nxv8f16.nxv8i16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svcvt_f16_s16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svcvt_f16_s16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: scvtf z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_f16_s16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: scvtf z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.scvtf.nxv8f16.nxv8i16(<vscale x 8 x half> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svcvt_f16_u16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svcvt_f16_u16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ucvtf z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_f16_u16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ucvtf z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.ucvtf.nxv8f16.nxv8i16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svcvt_f16_u16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svcvt_f16_u16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: ucvtf z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_f16_u16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ucvtf z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.ucvtf.nxv8f16.nxv8i16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svcvt_f16_u16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svcvt_f16_u16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: ucvtf z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_f16_u16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ucvtf z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.ucvtf.nxv8f16.nxv8i16(<vscale x 8 x half> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 4 x float> @test_svcvt_f32_s32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svcvt_f32_s32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: scvtf z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_f32_s32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: scvtf z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.scvtf.nxv4f32.nxv4i32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svcvt_f32_s32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svcvt_f32_s32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: scvtf z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_f32_s32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: scvtf z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.scvtf.nxv4f32.nxv4i32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svcvt_f32_s32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svcvt_f32_s32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: scvtf z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_f32_s32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: scvtf z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.scvtf.nxv4f32.nxv4i32(<vscale x 4 x float> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svcvt_f32_u32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svcvt_f32_u32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ucvtf z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_f32_u32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ucvtf z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.ucvtf.nxv4f32.nxv4i32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svcvt_f32_u32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svcvt_f32_u32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: ucvtf z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_f32_u32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ucvtf z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.ucvtf.nxv4f32.nxv4i32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svcvt_f32_u32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svcvt_f32_u32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: ucvtf z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_f32_u32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ucvtf z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.ucvtf.nxv4f32.nxv4i32(<vscale x 4 x float> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 2 x double> @test_svcvt_f64_s64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svcvt_f64_s64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: scvtf z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_f64_s64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: scvtf z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.scvtf.nxv2f64.nxv2i64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svcvt_f64_s64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svcvt_f64_s64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: scvtf z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_f64_s64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: scvtf z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.scvtf.nxv2f64.nxv2i64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svcvt_f64_s64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svcvt_f64_s64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: scvtf z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_f64_s64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: scvtf z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.scvtf.nxv2f64.nxv2i64(<vscale x 2 x double> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svcvt_f64_u64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svcvt_f64_u64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ucvtf z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_f64_u64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ucvtf z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.ucvtf.nxv2f64.nxv2i64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svcvt_f64_u64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svcvt_f64_u64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: ucvtf z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_f64_u64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ucvtf z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.ucvtf.nxv2f64.nxv2i64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svcvt_f64_u64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svcvt_f64_u64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: ucvtf z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcvt_f64_u64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ucvtf z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.ucvtf.nxv2f64.nxv2i64(<vscale x 2 x double> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x double> %0
+}
>From 983d8f12cbb6577bebaeea904b7270753966c68f Mon Sep 17 00:00:00 2001
From: Momchil Velikov <momchil.velikov at arm.com>
Date: Fri, 20 Dec 2024 14:41:57 +0000
Subject: [PATCH 03/16] [fixup] Rebase and add tests for the new all-true
patterns
---
llvm/lib/Target/AArch64/SVEInstrFormats.td | 16 +-
.../CodeGen/AArch64/zeroing-forms-uscvtf.ll | 525 ++++++++++++++++++
2 files changed, 533 insertions(+), 8 deletions(-)
diff --git a/llvm/lib/Target/AArch64/SVEInstrFormats.td b/llvm/lib/Target/AArch64/SVEInstrFormats.td
index 027f83a6a50442..64d0d3749efdc4 100644
--- a/llvm/lib/Target/AArch64/SVEInstrFormats.td
+++ b/llvm/lib/Target/AArch64/SVEInstrFormats.td
@@ -3332,14 +3332,14 @@ multiclass sve_fp_z2op_p_zd_c<bit U, string asm, string int_op, SDPatternOperato
def _DtoH : sve_fp_z2op_p_zd<{ 0b011011, U }, asm, ZPR64, ZPR16>;
def _DtoD : sve_fp_z2op_p_zd<{ 0b111011, U }, asm, ZPR64, ZPR64>;
- def : SVE_3_Op_UndefZero_Pat<nxv4f32, !cast<SDPatternOperator>(int_op # _f32i64), nxv4f32, nxv2i1, nxv2i64, !cast<Instruction>(NAME # _DtoS)>;
- def : SVE_3_Op_UndefZero_Pat<nxv2f64, !cast<SDPatternOperator>(int_op # _f64i32), nxv2f64, nxv2i1, nxv4i32, !cast<Instruction>(NAME # _StoD)>;
- def : SVE_3_Op_UndefZero_Pat<nxv8f16, !cast<SDPatternOperator>(int_op # _f16i32), nxv8f16, nxv4i1, nxv4i32, !cast<Instruction>(NAME # _StoH)>;
- def : SVE_3_Op_UndefZero_Pat<nxv8f16, !cast<SDPatternOperator>(int_op # _f16i64), nxv8f16, nxv2i1, nxv2i64, !cast<Instruction>(NAME # _DtoH)>;
-
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv8f16, ir_op, nxv8i1,nxv8i16, !cast<Instruction>(NAME # _HtoH)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv4f32, ir_op, nxv4i1,nxv4i32, !cast<Instruction>(NAME # _StoS)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv2f64, ir_op, nxv2i1,nxv2i64, !cast<Instruction>(NAME # _DtoD)>;
+ defm : SVE_3_Op_UndefZero_Pat<nxv4f32, !cast<SDPatternOperator>(int_op # _f32i64), nxv4f32, nxv2i1, nxv2i64, !cast<Instruction>(NAME # _DtoS)>;
+ defm : SVE_3_Op_UndefZero_Pat<nxv2f64, !cast<SDPatternOperator>(int_op # _f64i32), nxv2f64, nxv2i1, nxv4i32, !cast<Instruction>(NAME # _StoD)>;
+ defm : SVE_3_Op_UndefZero_Pat<nxv8f16, !cast<SDPatternOperator>(int_op # _f16i32), nxv8f16, nxv4i1, nxv4i32, !cast<Instruction>(NAME # _StoH)>;
+ defm : SVE_3_Op_UndefZero_Pat<nxv8f16, !cast<SDPatternOperator>(int_op # _f16i64), nxv8f16, nxv2i1, nxv2i64, !cast<Instruction>(NAME # _DtoH)>;
+
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv8f16, ir_op, nxv8i1,nxv8i16, !cast<Instruction>(NAME # _HtoH)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv4f32, ir_op, nxv4i1,nxv4i32, !cast<Instruction>(NAME # _StoS)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv2f64, ir_op, nxv2i1,nxv2i64, !cast<Instruction>(NAME # _DtoD)>;
}
multiclass sve_fp_z2op_p_zd_d_flogb<string asm> {
diff --git a/llvm/test/CodeGen/AArch64/zeroing-forms-uscvtf.ll b/llvm/test/CodeGen/AArch64/zeroing-forms-uscvtf.ll
index 1ca3b1a6e31cb9..06c87a70c19192 100644
--- a/llvm/test/CodeGen/AArch64/zeroing-forms-uscvtf.ll
+++ b/llvm/test/CodeGen/AArch64/zeroing-forms-uscvtf.ll
@@ -656,3 +656,528 @@ entry:
%0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.ucvtf.nxv2f64.nxv2i64(<vscale x 2 x double> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x double> %0
}
+
+define <vscale x 4 x float> @test_scvtf_f32_i64_ptrue_u(double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_scvtf_f32_i64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: scvtf z0.s, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_scvtf_f32_i64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: scvtf z0.s, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.scvtf.f32i64(<vscale x 4 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_scvtf_f32_i64_ptrue(double %z0,<vscale x 4 x float> %x, <vscale x 2 x i64> %y) {
+; CHECK-LABEL: test_scvtf_f32_i64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: scvtf z0.s, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_scvtf_f32_i64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: scvtf z0.s, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.scvtf.f32i64(<vscale x 4 x float> %x, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %y)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 2 x double> @test_scvtf_f64_i32_ptrue_u(double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_scvtf_f64_i32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: scvtf z0.d, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_scvtf_f64_i32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: scvtf z0.d, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.scvtf.f64i32(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_scvtf_f64_i32_ptrue(double %z0,<vscale x 2 x double> %x, <vscale x 4 x i32> %y) {
+; CHECK-LABEL: test_scvtf_f64_i32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: scvtf z0.d, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_scvtf_f64_i32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: scvtf z0.d, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.scvtf.f64i32(<vscale x 2 x double> %x, <vscale x 2 x i1> %pg, <vscale x 4 x i32> %y)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 8 x half> @test_scvtf_f16_i32_ptrue_u(double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_scvtf_f16_i32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: scvtf z0.h, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_scvtf_f16_i32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: scvtf z0.h, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.scvtf.f16i32(<vscale x 8 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_scvtf_f16_i32_ptrue(double %z0,<vscale x 8 x half> %x, <vscale x 4 x i32> %y) {
+; CHECK-LABEL: test_scvtf_f16_i32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: scvtf z0.h, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_scvtf_f16_i32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: scvtf z0.h, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.scvtf.f16i32(<vscale x 8 x half> %x, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %y)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_scvtf_f16_i64_ptrue_u(double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_scvtf_f16_i64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: scvtf z0.h, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_scvtf_f16_i64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: scvtf z0.h, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.scvtf.f16i64(<vscale x 8 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_scvtf_f16_i64_ptrue(double %z0,<vscale x 8 x half> %x, <vscale x 2 x i64> %y) {
+; CHECK-LABEL: test_scvtf_f16_i64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: scvtf z0.h, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_scvtf_f16_i64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: scvtf z0.h, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.scvtf.f16i64(<vscale x 8 x half> %x, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %y)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 4 x float> @test_ucvtf_f32_i64_ptrue_u(double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_ucvtf_f32_i64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: ucvtf z0.s, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_ucvtf_f32_i64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: ucvtf z0.s, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.ucvtf.f32i64(<vscale x 4 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_ucvtf_f32_i64_ptrue(double %z0,<vscale x 4 x float> %x, <vscale x 2 x i64> %y) {
+; CHECK-LABEL: test_ucvtf_f32_i64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: ucvtf z0.s, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_ucvtf_f32_i64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: ucvtf z0.s, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.ucvtf.f32i64(<vscale x 4 x float> %x, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %y)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 2 x double> @test_ucvtf_f64_i32_ptrue_u(double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_ucvtf_f64_i32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: ucvtf z0.d, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_ucvtf_f64_i32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: ucvtf z0.d, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.ucvtf.f64i32(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_ucvtf_f64_i32_ptrue(double %z0,<vscale x 2 x double> %x, <vscale x 4 x i32> %y) {
+; CHECK-LABEL: test_ucvtf_f64_i32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: ucvtf z0.d, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_ucvtf_f64_i32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: ucvtf z0.d, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.ucvtf.f64i32(<vscale x 2 x double> %x, <vscale x 2 x i1> %pg, <vscale x 4 x i32> %y)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 8 x half> @test_ucvtf_f16_i32_ptrue_u(double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_ucvtf_f16_i32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: ucvtf z0.h, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_ucvtf_f16_i32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: ucvtf z0.h, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.ucvtf.f16i32(<vscale x 8 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_ucvtf_f16_i32_ptrue(double %z0,<vscale x 8 x half> %x, <vscale x 4 x i32> %y) {
+; CHECK-LABEL: test_ucvtf_f16_i32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: ucvtf z0.h, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_ucvtf_f16_i32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: ucvtf z0.h, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.ucvtf.f16i32(<vscale x 8 x half> %x, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %y)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_ucvtf_f16_i64_ptrue_u(double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_ucvtf_f16_i64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: ucvtf z0.h, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_ucvtf_f16_i64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: ucvtf z0.h, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.ucvtf.f16i64(<vscale x 8 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_ucvtf_f16_i64_ptrue(double %z0,<vscale x 8 x half> %x, <vscale x 2 x i64> %y) {
+; CHECK-LABEL: test_ucvtf_f16_i64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: ucvtf z0.h, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_ucvtf_f16_i64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: ucvtf z0.h, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.ucvtf.f16i64(<vscale x 8 x half> %x, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %y)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_scvtf_nxv8f16_nxv8i16_ptrue_u(double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_scvtf_nxv8f16_nxv8i16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: scvtf z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_scvtf_nxv8f16_nxv8i16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: scvtf z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.scvtf.nxv8f16.nxv8i16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_scvtf_nxv8f16_nxv8i16_ptrue(double %z0,<vscale x 8 x half> %x, <vscale x 8 x i16> %y) {
+; CHECK-LABEL: test_scvtf_nxv8f16_nxv8i16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: scvtf z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_scvtf_nxv8f16_nxv8i16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: scvtf z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.scvtf.nxv8f16.nxv8i16(<vscale x 8 x half> %x, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %y)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_ucvtf_nxv8f16_nxv8i16_ptrue_u(double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_ucvtf_nxv8f16_nxv8i16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: ucvtf z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_ucvtf_nxv8f16_nxv8i16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: ucvtf z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.ucvtf.nxv8f16.nxv8i16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_ucvtf_nxv8f16_nxv8i16_ptrue(double %z0,<vscale x 8 x half> %x, <vscale x 8 x i16> %y) {
+; CHECK-LABEL: test_ucvtf_nxv8f16_nxv8i16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: ucvtf z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_ucvtf_nxv8f16_nxv8i16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: ucvtf z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.ucvtf.nxv8f16.nxv8i16(<vscale x 8 x half> %x, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %y)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 4 x float> @test_scvtf_nxv4f32_nxv4i32_ptrue_u(double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_scvtf_nxv4f32_nxv4i32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: scvtf z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_scvtf_nxv4f32_nxv4i32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: scvtf z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.scvtf.nxv4f32.nxv4i32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_scvtf_nxv4f32_nxv4i32_ptrue(double %z0,<vscale x 4 x float> %x, <vscale x 4 x i32> %y) {
+; CHECK-LABEL: test_scvtf_nxv4f32_nxv4i32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: scvtf z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_scvtf_nxv4f32_nxv4i32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: scvtf z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.scvtf.nxv4f32.nxv4i32(<vscale x 4 x float> %x, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %y)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_ucvtf_nxv4f32_nxv4i32_ptrue_u(double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_ucvtf_nxv4f32_nxv4i32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: ucvtf z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_ucvtf_nxv4f32_nxv4i32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: ucvtf z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.ucvtf.nxv4f32.nxv4i32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_ucvtf_nxv4f32_nxv4i32_ptrue(double %z0,<vscale x 4 x float> %x, <vscale x 4 x i32> %y) {
+; CHECK-LABEL: test_ucvtf_nxv4f32_nxv4i32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: ucvtf z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_ucvtf_nxv4f32_nxv4i32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: ucvtf z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.ucvtf.nxv4f32.nxv4i32(<vscale x 4 x float> %x, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %y)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 2 x double> @test_scvtf_nxvf64_nxv2i64_ptrue_u(double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_scvtf_nxvf64_nxv2i64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: scvtf z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_scvtf_nxvf64_nxv2i64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: scvtf z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.scvtf.nxvf64.nxv2i64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_scvtf_nxvf64_nxv2i64_ptrue(double %z0,<vscale x 2 x double> %x, <vscale x 2 x i64> %y) {
+; CHECK-LABEL: test_scvtf_nxvf64_nxv2i64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: scvtf z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_scvtf_nxvf64_nxv2i64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: scvtf z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.scvtf.nxvf64.nxv2i64(<vscale x 2 x double> %x, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %y)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_ucvtf_nxvf64_nxv2i64_ptrue_u(double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_ucvtf_nxvf64_nxv2i64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: ucvtf z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_ucvtf_nxvf64_nxv2i64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: ucvtf z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.ucvtf.nxvf64.nxv2i64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_ucvtf_nxvf64_nxv2i64_ptrue(double %z0,<vscale x 2 x double> %x, <vscale x 2 x i64> %y) {
+; CHECK-LABEL: test_ucvtf_nxvf64_nxv2i64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: ucvtf z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_ucvtf_nxvf64_nxv2i64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: ucvtf z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.ucvtf.nxvf64.nxv2i64(<vscale x 2 x double> %x, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %y)
+ ret <vscale x 2 x double> %0
+}
+
>From 8b5dba238b6d447557f425e6da01cec3fdb5f0c8 Mon Sep 17 00:00:00 2001
From: Momchil Velikov <momchil.velikov at arm.com>
Date: Fri, 20 Dec 2024 15:08:21 +0000
Subject: [PATCH 04/16] [fixup] Update tests, change undef to poison
---
.../CodeGen/AArch64/zeroing-forms-uscvtf.ll | 112 +++++++++---------
1 file changed, 56 insertions(+), 56 deletions(-)
diff --git a/llvm/test/CodeGen/AArch64/zeroing-forms-uscvtf.ll b/llvm/test/CodeGen/AArch64/zeroing-forms-uscvtf.ll
index 06c87a70c19192..fd0126f3166dd3 100644
--- a/llvm/test/CodeGen/AArch64/zeroing-forms-uscvtf.ll
+++ b/llvm/test/CodeGen/AArch64/zeroing-forms-uscvtf.ll
@@ -18,7 +18,7 @@ define <vscale x 4 x float> @test_scvtf_f32_s64_x_1(<vscale x 2 x i1> %pg, <vsca
; CHECK-2p2-NEXT: scvtf z0.s, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.scvtf.f32i64(<vscale x 4 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.scvtf.f32i64(<vscale x 4 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 4 x float> %0
}
@@ -33,7 +33,7 @@ define <vscale x 4 x float> @test_scvtf_f32_s64_x_2(<vscale x 2 x i1> %pg, doubl
; CHECK-2p2-NEXT: scvtf z0.s, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.scvtf.f32i64(<vscale x 4 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.scvtf.f32i64(<vscale x 4 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 4 x float> %0
}
@@ -64,7 +64,7 @@ define <vscale x 2 x double> @test_scvtf_f64_s32_x_1(<vscale x 2 x i1> %pg, <vsc
; CHECK-2p2-NEXT: scvtf z0.d, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.scvtf.f64i32(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.scvtf.f64i32(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 2 x double> %0
}
@@ -79,7 +79,7 @@ define <vscale x 2 x double> @test_scvtf_f64_s32_x_2(<vscale x 2 x i1> %pg, doub
; CHECK-2p2-NEXT: scvtf z0.d, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.scvtf.f64i32(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.scvtf.f64i32(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 2 x double> %0
}
@@ -110,7 +110,7 @@ define <vscale x 8 x half> @test_scvtf_f16_s32_x_1(<vscale x 4 x i1> %pg, <vscal
; CHECK-2p2-NEXT: scvtf z0.h, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.scvtf.f16i32(<vscale x 8 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.scvtf.f16i32(<vscale x 8 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 8 x half> %0
}
@@ -125,7 +125,7 @@ define <vscale x 8 x half> @test_scvtf_f16_s32_x_2(<vscale x 4 x i1> %pg, double
; CHECK-2p2-NEXT: scvtf z0.h, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.scvtf.f16i32(<vscale x 8 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.scvtf.f16i32(<vscale x 8 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 8 x half> %0
}
@@ -156,7 +156,7 @@ define <vscale x 8 x half> @test_scvtf_f16_s64_x_1(<vscale x 2 x i1> %pg, <vscal
; CHECK-2p2-NEXT: scvtf z0.h, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.scvtf.f16i64(<vscale x 8 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.scvtf.f16i64(<vscale x 8 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 8 x half> %0
}
@@ -171,7 +171,7 @@ define <vscale x 8 x half> @test_scvtf_f16_s64_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: scvtf z0.h, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.scvtf.f16i64(<vscale x 8 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.scvtf.f16i64(<vscale x 8 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 8 x half> %0
}
@@ -202,7 +202,7 @@ define <vscale x 4 x float> @test_ucvtf_f32_u64_x_1(<vscale x 2 x i1> %pg, <vsca
; CHECK-2p2-NEXT: ucvtf z0.s, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.ucvtf.f32i64(<vscale x 4 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.ucvtf.f32i64(<vscale x 4 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 4 x float> %0
}
@@ -217,7 +217,7 @@ define <vscale x 4 x float> @test_ucvtf_f32_u64_x_2(<vscale x 2 x i1> %pg, doubl
; CHECK-2p2-NEXT: ucvtf z0.s, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.ucvtf.f32i64(<vscale x 4 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.ucvtf.f32i64(<vscale x 4 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 4 x float> %0
}
@@ -248,7 +248,7 @@ define <vscale x 2 x double> @test_ucvtf_f64_u32_x_1(<vscale x 2 x i1> %pg, <vsc
; CHECK-2p2-NEXT: ucvtf z0.d, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.ucvtf.f64i32(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.ucvtf.f64i32(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 2 x double> %0
}
@@ -263,7 +263,7 @@ define <vscale x 2 x double> @test_ucvtf_f64_u32_x_2(<vscale x 2 x i1> %pg, doub
; CHECK-2p2-NEXT: ucvtf z0.d, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.ucvtf.f64i32(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.ucvtf.f64i32(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 2 x double> %0
}
@@ -294,7 +294,7 @@ define <vscale x 8 x half> @test_ucvtf_f16_u32_x_1(<vscale x 4 x i1> %pg, <vscal
; CHECK-2p2-NEXT: ucvtf z0.h, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.ucvtf.f16i32(<vscale x 8 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.ucvtf.f16i32(<vscale x 8 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 8 x half> %0
}
@@ -309,7 +309,7 @@ define <vscale x 8 x half> @test_ucvtf_f16_u32_x_2(<vscale x 4 x i1> %pg, double
; CHECK-2p2-NEXT: ucvtf z0.h, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.ucvtf.f16i32(<vscale x 8 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.ucvtf.f16i32(<vscale x 8 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 8 x half> %0
}
@@ -340,7 +340,7 @@ define <vscale x 8 x half> @test_ucvtf_f16_u64_x_1(<vscale x 2 x i1> %pg, <vscal
; CHECK-2p2-NEXT: ucvtf z0.h, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.ucvtf.f16i64(<vscale x 8 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.ucvtf.f16i64(<vscale x 8 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 8 x half> %0
}
@@ -355,7 +355,7 @@ define <vscale x 8 x half> @test_ucvtf_f16_u64_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: ucvtf z0.h, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.ucvtf.f16i64(<vscale x 8 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.ucvtf.f16i64(<vscale x 8 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 8 x half> %0
}
@@ -386,7 +386,7 @@ define <vscale x 8 x half> @test_svcvt_f16_s16_x_1(<vscale x 8 x i1> %pg, <vscal
; CHECK-2p2-NEXT: scvtf z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.scvtf.nxv8f16.nxv8i16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.scvtf.nxv8f16.nxv8i16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x half> %0
}
@@ -402,7 +402,7 @@ define <vscale x 8 x half> @test_svcvt_f16_s16_x_2(<vscale x 8 x i1> %pg, double
; CHECK-2p2-NEXT: scvtf z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.scvtf.nxv8f16.nxv8i16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.scvtf.nxv8f16.nxv8i16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x half> %0
}
@@ -433,7 +433,7 @@ define <vscale x 8 x half> @test_svcvt_f16_u16_x_1(<vscale x 8 x i1> %pg, <vscal
; CHECK-2p2-NEXT: ucvtf z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.ucvtf.nxv8f16.nxv8i16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.ucvtf.nxv8f16.nxv8i16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x half> %0
}
@@ -449,7 +449,7 @@ define <vscale x 8 x half> @test_svcvt_f16_u16_x_2(<vscale x 8 x i1> %pg, double
; CHECK-2p2-NEXT: ucvtf z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.ucvtf.nxv8f16.nxv8i16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.ucvtf.nxv8f16.nxv8i16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x half> %0
}
@@ -480,7 +480,7 @@ define <vscale x 4 x float> @test_svcvt_f32_s32_x_1(<vscale x 4 x i1> %pg, <vsca
; CHECK-2p2-NEXT: scvtf z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.scvtf.nxv4f32.nxv4i32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.scvtf.nxv4f32.nxv4i32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x float> %0
}
@@ -496,7 +496,7 @@ define <vscale x 4 x float> @test_svcvt_f32_s32_x_2(<vscale x 4 x i1> %pg, doubl
; CHECK-2p2-NEXT: scvtf z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.scvtf.nxv4f32.nxv4i32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.scvtf.nxv4f32.nxv4i32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x float> %0
}
@@ -527,7 +527,7 @@ define <vscale x 4 x float> @test_svcvt_f32_u32_x_1(<vscale x 4 x i1> %pg, <vsca
; CHECK-2p2-NEXT: ucvtf z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.ucvtf.nxv4f32.nxv4i32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.ucvtf.nxv4f32.nxv4i32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x float> %0
}
@@ -543,7 +543,7 @@ define <vscale x 4 x float> @test_svcvt_f32_u32_x_2(<vscale x 4 x i1> %pg, doubl
; CHECK-2p2-NEXT: ucvtf z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.ucvtf.nxv4f32.nxv4i32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.ucvtf.nxv4f32.nxv4i32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x float> %0
}
@@ -574,7 +574,7 @@ define <vscale x 2 x double> @test_svcvt_f64_s64_x_1(<vscale x 2 x i1> %pg, <vsc
; CHECK-2p2-NEXT: scvtf z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.scvtf.nxv2f64.nxv2i64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.scvtf.nxv2f64.nxv2i64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x double> %0
}
@@ -590,7 +590,7 @@ define <vscale x 2 x double> @test_svcvt_f64_s64_x_2(<vscale x 2 x i1> %pg, doub
; CHECK-2p2-NEXT: scvtf z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.scvtf.nxv2f64.nxv2i64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.scvtf.nxv2f64.nxv2i64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x double> %0
}
@@ -621,7 +621,7 @@ define <vscale x 2 x double> @test_svcvt_f64_u64_x_1(<vscale x 2 x i1> %pg, <vsc
; CHECK-2p2-NEXT: ucvtf z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.ucvtf.nxv2f64.nxv2i64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.ucvtf.nxv2f64.nxv2i64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x double> %0
}
@@ -637,7 +637,7 @@ define <vscale x 2 x double> @test_svcvt_f64_u64_x_2(<vscale x 2 x i1> %pg, doub
; CHECK-2p2-NEXT: ucvtf z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.ucvtf.nxv2f64.nxv2i64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.ucvtf.nxv2f64.nxv2i64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x double> %0
}
@@ -744,7 +744,7 @@ define <vscale x 8 x half> @test_scvtf_f16_i32_ptrue_u(double %z0, <vscale x 4 x
; CHECK-2p2-NEXT: scvtf z0.h, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
%0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.scvtf.f16i32(<vscale x 8 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 8 x half> %0
}
@@ -763,7 +763,7 @@ define <vscale x 8 x half> @test_scvtf_f16_i32_ptrue(double %z0,<vscale x 8 x ha
; CHECK-2p2-NEXT: scvtf z0.h, p0/z, z2.s
; CHECK-2p2-NEXT: ret
entry:
- %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
%0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.scvtf.f16i32(<vscale x 8 x half> %x, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %y)
ret <vscale x 8 x half> %0
}
@@ -892,7 +892,7 @@ define <vscale x 8 x half> @test_ucvtf_f16_i32_ptrue_u(double %z0, <vscale x 4 x
; CHECK-2p2-NEXT: ucvtf z0.h, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
%0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.ucvtf.f16i32(<vscale x 8 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 8 x half> %0
}
@@ -911,7 +911,7 @@ define <vscale x 8 x half> @test_ucvtf_f16_i32_ptrue(double %z0,<vscale x 8 x ha
; CHECK-2p2-NEXT: ucvtf z0.h, p0/z, z2.s
; CHECK-2p2-NEXT: ret
entry:
- %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
%0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.ucvtf.f16i32(<vscale x 8 x half> %x, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %y)
ret <vscale x 8 x half> %0
}
@@ -967,7 +967,7 @@ define <vscale x 8 x half> @test_scvtf_nxv8f16_nxv8i16_ptrue_u(double %z0, <vsca
; CHECK-2p2-NEXT: scvtf z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
%0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.scvtf.nxv8f16.nxv8i16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x half> %0
}
@@ -986,7 +986,7 @@ define <vscale x 8 x half> @test_scvtf_nxv8f16_nxv8i16_ptrue(double %z0,<vscale
; CHECK-2p2-NEXT: scvtf z0.h, p0/z, z2.h
; CHECK-2p2-NEXT: ret
entry:
- %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
%0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.scvtf.nxv8f16.nxv8i16(<vscale x 8 x half> %x, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %y)
ret <vscale x 8 x half> %0
}
@@ -1005,7 +1005,7 @@ define <vscale x 8 x half> @test_ucvtf_nxv8f16_nxv8i16_ptrue_u(double %z0, <vsca
; CHECK-2p2-NEXT: ucvtf z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
%0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.ucvtf.nxv8f16.nxv8i16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x half> %0
}
@@ -1024,7 +1024,7 @@ define <vscale x 8 x half> @test_ucvtf_nxv8f16_nxv8i16_ptrue(double %z0,<vscale
; CHECK-2p2-NEXT: ucvtf z0.h, p0/z, z2.h
; CHECK-2p2-NEXT: ret
entry:
- %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
%0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.ucvtf.nxv8f16.nxv8i16(<vscale x 8 x half> %x, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %y)
ret <vscale x 8 x half> %0
}
@@ -1043,7 +1043,7 @@ define <vscale x 4 x float> @test_scvtf_nxv4f32_nxv4i32_ptrue_u(double %z0, <vsc
; CHECK-2p2-NEXT: scvtf z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
%0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.scvtf.nxv4f32.nxv4i32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x float> %0
}
@@ -1062,7 +1062,7 @@ define <vscale x 4 x float> @test_scvtf_nxv4f32_nxv4i32_ptrue(double %z0,<vscale
; CHECK-2p2-NEXT: scvtf z0.s, p0/z, z2.s
; CHECK-2p2-NEXT: ret
entry:
- %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
%0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.scvtf.nxv4f32.nxv4i32(<vscale x 4 x float> %x, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %y)
ret <vscale x 4 x float> %0
}
@@ -1081,7 +1081,7 @@ define <vscale x 4 x float> @test_ucvtf_nxv4f32_nxv4i32_ptrue_u(double %z0, <vsc
; CHECK-2p2-NEXT: ucvtf z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
%0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.ucvtf.nxv4f32.nxv4i32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x float> %0
}
@@ -1100,84 +1100,84 @@ define <vscale x 4 x float> @test_ucvtf_nxv4f32_nxv4i32_ptrue(double %z0,<vscale
; CHECK-2p2-NEXT: ucvtf z0.s, p0/z, z2.s
; CHECK-2p2-NEXT: ret
entry:
- %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
%0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.ucvtf.nxv4f32.nxv4i32(<vscale x 4 x float> %x, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %y)
ret <vscale x 4 x float> %0
}
-define <vscale x 2 x double> @test_scvtf_nxvf64_nxv2i64_ptrue_u(double %z0, <vscale x 2 x i64> %x) {
-; CHECK-LABEL: test_scvtf_nxvf64_nxv2i64_ptrue_u:
+define <vscale x 2 x double> @test_scvtf_nxv2f64_nxv2i64_ptrue_u(double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_scvtf_nxv2f64_nxv2i64_ptrue_u:
; CHECK: // %bb.0: // %entry
; CHECK-NEXT: ptrue p0.d
; CHECK-NEXT: movprfx z0, z1
; CHECK-NEXT: scvtf z0.d, p0/m, z1.d
; CHECK-NEXT: ret
;
-; CHECK-2p2-LABEL: test_scvtf_nxvf64_nxv2i64_ptrue_u:
+; CHECK-2p2-LABEL: test_scvtf_nxv2f64_nxv2i64_ptrue_u:
; CHECK-2p2: // %bb.0: // %entry
; CHECK-2p2-NEXT: ptrue p0.d
; CHECK-2p2-NEXT: scvtf z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
%pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.scvtf.nxvf64.nxv2i64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.scvtf.nxv2f64.nxv2i64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x double> %0
}
-define <vscale x 2 x double> @test_scvtf_nxvf64_nxv2i64_ptrue(double %z0,<vscale x 2 x double> %x, <vscale x 2 x i64> %y) {
-; CHECK-LABEL: test_scvtf_nxvf64_nxv2i64_ptrue:
+define <vscale x 2 x double> @test_scvtf_nxv2f64_nxv2i64_ptrue(double %z0,<vscale x 2 x double> %x, <vscale x 2 x i64> %y) {
+; CHECK-LABEL: test_scvtf_nxv2f64_nxv2i64_ptrue:
; CHECK: // %bb.0: // %entry
; CHECK-NEXT: ptrue p0.d
; CHECK-NEXT: movprfx z0, z2
; CHECK-NEXT: scvtf z0.d, p0/m, z2.d
; CHECK-NEXT: ret
;
-; CHECK-2p2-LABEL: test_scvtf_nxvf64_nxv2i64_ptrue:
+; CHECK-2p2-LABEL: test_scvtf_nxv2f64_nxv2i64_ptrue:
; CHECK-2p2: // %bb.0: // %entry
; CHECK-2p2-NEXT: ptrue p0.d
; CHECK-2p2-NEXT: scvtf z0.d, p0/z, z2.d
; CHECK-2p2-NEXT: ret
entry:
%pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.scvtf.nxvf64.nxv2i64(<vscale x 2 x double> %x, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %y)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.scvtf.nxv2f64.nxv2i64(<vscale x 2 x double> %x, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %y)
ret <vscale x 2 x double> %0
}
-define <vscale x 2 x double> @test_ucvtf_nxvf64_nxv2i64_ptrue_u(double %z0, <vscale x 2 x i64> %x) {
-; CHECK-LABEL: test_ucvtf_nxvf64_nxv2i64_ptrue_u:
+define <vscale x 2 x double> @test_ucvtf_nxv2f64_nxv2i64_ptrue_u(double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_ucvtf_nxv2f64_nxv2i64_ptrue_u:
; CHECK: // %bb.0: // %entry
; CHECK-NEXT: ptrue p0.d
; CHECK-NEXT: movprfx z0, z1
; CHECK-NEXT: ucvtf z0.d, p0/m, z1.d
; CHECK-NEXT: ret
;
-; CHECK-2p2-LABEL: test_ucvtf_nxvf64_nxv2i64_ptrue_u:
+; CHECK-2p2-LABEL: test_ucvtf_nxv2f64_nxv2i64_ptrue_u:
; CHECK-2p2: // %bb.0: // %entry
; CHECK-2p2-NEXT: ptrue p0.d
; CHECK-2p2-NEXT: ucvtf z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
%pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.ucvtf.nxvf64.nxv2i64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.ucvtf.nxv2f64.nxv2i64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x double> %0
}
-define <vscale x 2 x double> @test_ucvtf_nxvf64_nxv2i64_ptrue(double %z0,<vscale x 2 x double> %x, <vscale x 2 x i64> %y) {
-; CHECK-LABEL: test_ucvtf_nxvf64_nxv2i64_ptrue:
+define <vscale x 2 x double> @test_ucvtf_nxv2f64_nxv2i64_ptrue(double %z0,<vscale x 2 x double> %x, <vscale x 2 x i64> %y) {
+; CHECK-LABEL: test_ucvtf_nxv2f64_nxv2i64_ptrue:
; CHECK: // %bb.0: // %entry
; CHECK-NEXT: ptrue p0.d
; CHECK-NEXT: movprfx z0, z2
; CHECK-NEXT: ucvtf z0.d, p0/m, z2.d
; CHECK-NEXT: ret
;
-; CHECK-2p2-LABEL: test_ucvtf_nxvf64_nxv2i64_ptrue:
+; CHECK-2p2-LABEL: test_ucvtf_nxv2f64_nxv2i64_ptrue:
; CHECK-2p2: // %bb.0: // %entry
; CHECK-2p2-NEXT: ptrue p0.d
; CHECK-2p2-NEXT: ucvtf z0.d, p0/z, z2.d
; CHECK-2p2-NEXT: ret
entry:
%pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.ucvtf.nxvf64.nxv2i64(<vscale x 2 x double> %x, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %y)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.ucvtf.nxv2f64.nxv2i64(<vscale x 2 x double> %x, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %y)
ret <vscale x 2 x double> %0
}
>From ac5b0926bf850fe4c23bb8f61dec3cc066f11b67 Mon Sep 17 00:00:00 2001
From: Momchil Velikov <momchil.velikov at arm.com>
Date: Tue, 19 Nov 2024 12:17:13 +0000
Subject: [PATCH 05/16] [AArch64] Generate zeroing forms of certain SVE2.2
instructions (6/11)
---
.../lib/Target/AArch64/AArch64SVEInstrInfo.td | 10 +-
llvm/lib/Target/AArch64/SVEInstrFormats.td | 7 +-
.../AArch64/zeroing-forms-counts-not.ll | 1152 +++++++++++++++++
3 files changed, 1163 insertions(+), 6 deletions(-)
create mode 100644 llvm/test/CodeGen/AArch64/zeroing-forms-counts-not.ll
diff --git a/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td b/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
index 23d626fe786630..5c257c436fbeac 100644
--- a/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
+++ b/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
@@ -4314,11 +4314,11 @@ let Predicates = [HasSVE2p2orSME2p2] in {
defm FSQRT_ZPZz : sve_fp_z2op_p_zd_hsd<0b01101, "fsqrt">;
// SVE2p2 integer unary arithmetic (bitwise), zeroing predicate
- defm CLS_ZPzZ : sve_int_un_pred_arit_bitwise_z<0b000, "cls">;
- defm CLZ_ZPzZ : sve_int_un_pred_arit_bitwise_z<0b001, "clz">;
- defm CNT_ZPzZ : sve_int_un_pred_arit_bitwise_z<0b010, "cnt">;
- defm CNOT_ZPzZ : sve_int_un_pred_arit_bitwise_z<0b011, "cnot">;
- defm NOT_ZPzZ : sve_int_un_pred_arit_bitwise_z<0b110, "not">;
+ defm CLS_ZPzZ : sve_int_un_pred_arit_bitwise_z<0b000, "cls", AArch64cls_mt>;
+ defm CLZ_ZPzZ : sve_int_un_pred_arit_bitwise_z<0b001, "clz", AArch64clz_mt>;
+ defm CNT_ZPzZ : sve_int_un_pred_arit_bitwise_z<0b010, "cnt", AArch64cnt_mt>;
+ defm CNOT_ZPzZ : sve_int_un_pred_arit_bitwise_z<0b011, "cnot", AArch64cnot_mt>;
+ defm NOT_ZPzZ : sve_int_un_pred_arit_bitwise_z<0b110, " not", AArch64not_mt>;
// floating point
defm FABS_ZPzZ : sve_int_un_pred_arit_bitwise_fp_z<0b100, "fabs", AArch64fabs_mt>;
diff --git a/llvm/lib/Target/AArch64/SVEInstrFormats.td b/llvm/lib/Target/AArch64/SVEInstrFormats.td
index 64d0d3749efdc4..a9a23574de5d1d 100644
--- a/llvm/lib/Target/AArch64/SVEInstrFormats.td
+++ b/llvm/lib/Target/AArch64/SVEInstrFormats.td
@@ -4992,11 +4992,16 @@ multiclass sve_int_un_pred_arit_bitwise<bits<3> opc, string asm,
defm : SVE_1_Op_PassthruUndef_Pat<nxv2i64, op, nxv2i1, nxv2i64, !cast<Pseudo>(NAME # _D_UNDEF)>;
}
-multiclass sve_int_un_pred_arit_bitwise_z<bits<3> opc, string asm> {
+multiclass sve_int_un_pred_arit_bitwise_z<bits<3> opc, string asm, SDPatternOperator op> {
def _B : sve_int_un_pred_arit_z<0b00, { opc, 0b1 }, asm, ZPR8>;
def _H : sve_int_un_pred_arit_z<0b01, { opc, 0b1 }, asm, ZPR16>;
def _S : sve_int_un_pred_arit_z<0b10, { opc, 0b1 }, asm, ZPR32>;
def _D : sve_int_un_pred_arit_z<0b11, { opc, 0b1 }, asm, ZPR64>;
+
+ def : SVE_1_Op_PassthruUndefZero_Pat<nxv16i8, op, nxv16i1, nxv16i8, !cast<Instruction>(NAME # _B)>;
+ def : SVE_1_Op_PassthruUndefZero_Pat<nxv8i16, op, nxv8i1, nxv8i16, !cast<Instruction>(NAME # _H)>;
+ def : SVE_1_Op_PassthruUndefZero_Pat<nxv4i32, op, nxv4i1, nxv4i32, !cast<Instruction>(NAME # _S)>;
+ def : SVE_1_Op_PassthruUndefZero_Pat<nxv2i64, op, nxv2i1, nxv2i64, !cast<Instruction>(NAME # _D)>;
}
multiclass sve_int_un_pred_arit_bitwise_fp<bits<3> opc, string asm,
diff --git a/llvm/test/CodeGen/AArch64/zeroing-forms-counts-not.ll b/llvm/test/CodeGen/AArch64/zeroing-forms-counts-not.ll
new file mode 100644
index 00000000000000..efdf4bd6a5c64b
--- /dev/null
+++ b/llvm/test/CodeGen/AArch64/zeroing-forms-counts-not.ll
@@ -0,0 +1,1152 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
+; RUN: llc -mattr=+bf16,+sve < %s | FileCheck %s
+; RUN: llc -mattr=+bf16,+sve2p2 < %s | FileCheck %s -check-prefix CHECK-2p2
+
+; RUN: llc -mattr=+bf16,+sme -force-streaming < %s | FileCheck %s
+; RUN: llc -mattr=+bf16,+sme2p2 -force-streaming < %s | FileCheck %s -check-prefix CHECK-2p2
+
+target triple = "aarch64-linux"
+
+define <vscale x 16 x i8> @test_svcls_s8_x_1(<vscale x 16 x i1> %pg, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svcls_s8_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: cls z0.b, p0/m, z0.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcls_s8_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cls z0.b, p0/z, z0.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.cls.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svcls_s8_x_2(<vscale x 16 x i1> %pg, double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svcls_s8_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cls z0.b, p0/m, z1.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcls_s8_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cls z0.b, p0/z, z1.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.cls.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svcls_s8_z(<vscale x 16 x i1> %pg, double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svcls_s8_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.b, #0 // =0x0
+; CHECK-NEXT: cls z0.b, p0/m, z1.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcls_s8_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cls z0.b, p0/z, z1.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.cls.nxv16i8(<vscale x 16 x i8> zeroinitializer, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 8 x i16> @test_svcls_s16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svcls_s16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: cls z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcls_s16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cls z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cls.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svcls_s16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svcls_s16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cls z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcls_s16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cls z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cls.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svcls_s16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svcls_s16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: cls z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcls_s16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cls z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cls.nxv8i16(<vscale x 8 x i16> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svcls_s32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svcls_s32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: cls z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcls_s32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cls z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cls.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svcls_s32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svcls_s32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cls z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcls_s32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cls z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cls.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svcls_s32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svcls_s32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: cls z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcls_s32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cls z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cls.nxv4i32(<vscale x 4 x i32> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svcls_s64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svcls_s64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: cls z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcls_s64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cls z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cls.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svcls_s64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svcls_s64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cls z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcls_s64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cls z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cls.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svcls_s64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svcls_s64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: cls z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcls_s64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cls z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cls.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 16 x i8> @test_svclz_s8_x_1(<vscale x 16 x i1> %pg, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svclz_s8_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: clz z0.b, p0/m, z0.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svclz_s8_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: clz z0.b, p0/z, z0.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.clz.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svclz_s8_x_2(<vscale x 16 x i1> %pg, double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svclz_s8_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: clz z0.b, p0/m, z1.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svclz_s8_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: clz z0.b, p0/z, z1.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.clz.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svclz_s8_z(<vscale x 16 x i1> %pg, double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svclz_s8_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.b, #0 // =0x0
+; CHECK-NEXT: clz z0.b, p0/m, z1.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svclz_s8_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: clz z0.b, p0/z, z1.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.clz.nxv16i8(<vscale x 16 x i8> zeroinitializer, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 8 x i16> @test_svclz_s16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svclz_s16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: clz z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svclz_s16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: clz z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.clz.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svclz_s16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svclz_s16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: clz z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svclz_s16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: clz z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.clz.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svclz_s16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svclz_s16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: clz z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svclz_s16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: clz z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.clz.nxv8i16(<vscale x 8 x i16> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svclz_s32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svclz_s32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: clz z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svclz_s32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: clz z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.clz.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svclz_s32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svclz_s32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: clz z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svclz_s32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: clz z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.clz.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svclz_s32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svclz_s32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: clz z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svclz_s32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: clz z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.clz.nxv4i32(<vscale x 4 x i32> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svclz_s64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svclz_s64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: clz z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svclz_s64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: clz z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.clz.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svclz_s64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svclz_s64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: clz z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svclz_s64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: clz z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.clz.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svclz_s64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svclz_s64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: clz z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svclz_s64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: clz z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.clz.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 16 x i8> @test_svcnt_s8_x_1(<vscale x 16 x i1> %pg, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svcnt_s8_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: cnt z0.b, p0/m, z0.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_s8_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnt z0.b, p0/z, z0.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.cnt.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svcnt_s8_x_2(<vscale x 16 x i1> %pg, double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svcnt_s8_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cnt z0.b, p0/m, z1.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_s8_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnt z0.b, p0/z, z1.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.cnt.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svcnt_s8_z(<vscale x 16 x i1> %pg, double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svcnt_s8_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.b, #0 // =0x0
+; CHECK-NEXT: cnt z0.b, p0/m, z1.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_s8_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnt z0.b, p0/z, z1.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.cnt.nxv16i8(<vscale x 16 x i8> zeroinitializer, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 8 x i16> @test_svcnt_s16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svcnt_s16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: cnt z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_s16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnt z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnt.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svcnt_s16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svcnt_s16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cnt z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_s16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnt z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnt.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svcnt_s16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svcnt_s16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: cnt z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_s16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnt z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnt.nxv8i16(<vscale x 8 x i16> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svcnt_s32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svcnt_s32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: cnt z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_s32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnt z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cnt.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svcnt_s32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svcnt_s32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cnt z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_s32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnt z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cnt.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svcnt_s32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svcnt_s32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: cnt z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_s32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnt z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cnt.nxv4i32(<vscale x 4 x i32> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svcnt_s64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svcnt_s64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: cnt z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_s64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnt z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cnt.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svcnt_s64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svcnt_s64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cnt z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_s64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnt z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cnt.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svcnt_s64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svcnt_s64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: cnt z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_s64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnt z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cnt.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 8 x i16> @test_svcnt_f16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svcnt_f16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: cnt z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_f16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnt z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnt.nxv8f16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svcnt_f16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svcnt_f16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cnt z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_f16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnt z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnt.nxv8f16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svcnt_f16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svcnt_f16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: cnt z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_f16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnt z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnt.nxv8f16(<vscale x 8 x i16> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svcnt_bf16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x bfloat> %x) {
+; CHECK-LABEL: test_svcnt_bf16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: cnt z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_bf16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnt z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnt.nxv8bf16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x bfloat> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svcnt_bf16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x bfloat> %x) {
+; CHECK-LABEL: test_svcnt_bf16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cnt z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_bf16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnt z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnt.nxv8bf16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x bfloat> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svcnt_bf16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x bfloat> %x) {
+; CHECK-LABEL: test_svcnt_bf16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: cnt z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_bf16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnt z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnt.nxv8bf16(<vscale x 8 x i16> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x bfloat> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svcnt_f32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svcnt_f32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: cnt z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_f32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnt z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cnt.nxv4f32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svcnt_f32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svcnt_f32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cnt z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_f32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnt z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cnt.nxv4f32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svcnt_f32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svcnt_f32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: cnt z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_f32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnt z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cnt.nxv4f32(<vscale x 4 x i32> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svcnt_f64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svcnt_f64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: cnt z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_f64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnt z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cnt.nxv2f64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svcnt_f64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svcnt_f64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cnt z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_f64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnt z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cnt.nxv2f64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svcnt_f64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svcnt_f64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: cnt z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_f64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnt z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cnt.nxv2f64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 16 x i8> @test_svcnot_s8_x_1(<vscale x 16 x i1> %pg, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svcnot_s8_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: cnot z0.b, p0/m, z0.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnot_s8_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnot z0.b, p0/z, z0.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.cnot.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svcnot_s8_x_2(<vscale x 16 x i1> %pg, double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svcnot_s8_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cnot z0.b, p0/m, z1.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnot_s8_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnot z0.b, p0/z, z1.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.cnot.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svcnot_s8_z(<vscale x 16 x i1> %pg, double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svcnot_s8_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.b, #0 // =0x0
+; CHECK-NEXT: cnot z0.b, p0/m, z1.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnot_s8_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnot z0.b, p0/z, z1.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.cnot.nxv16i8(<vscale x 16 x i8> zeroinitializer, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 8 x i16> @test_svcnot_s16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svcnot_s16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: cnot z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnot_s16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnot z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnot.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svcnot_s16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svcnot_s16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cnot z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnot_s16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnot z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnot.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svcnot_s16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svcnot_s16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: cnot z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnot_s16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnot z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnot.nxv8i16(<vscale x 8 x i16> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svcnot_s32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svcnot_s32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: cnot z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnot_s32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnot z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cnot.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svcnot_s32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svcnot_s32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cnot z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnot_s32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnot z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cnot.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svcnot_s32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svcnot_s32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: cnot z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnot_s32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnot z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cnot.nxv4i32(<vscale x 4 x i32> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svcnot_s64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svcnot_s64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: cnot z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnot_s64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnot z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cnot.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svcnot_s64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svcnot_s64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cnot z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnot_s64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnot z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cnot.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svcnot_s64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svcnot_s64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: cnot z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnot_s64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: cnot z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cnot.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 16 x i8> @test_svnot_s8_x_1(<vscale x 16 x i1> %pg, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svnot_s8_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: not z0.b, p0/m, z0.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svnot_s8_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: not z0.b, p0/z, z0.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.not.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svnot_s8_x_2(<vscale x 16 x i1> %pg, double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svnot_s8_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: not z0.b, p0/m, z1.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svnot_s8_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: not z0.b, p0/z, z1.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.not.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svnot_s8_z(<vscale x 16 x i1> %pg, double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svnot_s8_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.b, #0 // =0x0
+; CHECK-NEXT: not z0.b, p0/m, z1.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svnot_s8_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: not z0.b, p0/z, z1.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.not.nxv16i8(<vscale x 16 x i8> zeroinitializer, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 8 x i16> @test_svnot_s16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svnot_s16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: not z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svnot_s16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: not z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.not.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svnot_s16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svnot_s16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: not z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svnot_s16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: not z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.not.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svnot_s16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svnot_s16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: not z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svnot_s16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: not z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.not.nxv8i16(<vscale x 8 x i16> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svnot_s32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svnot_s32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: not z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svnot_s32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: not z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.not.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svnot_s32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svnot_s32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: not z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svnot_s32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: not z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.not.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svnot_s32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svnot_s32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: not z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svnot_s32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: not z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.not.nxv4i32(<vscale x 4 x i32> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svnot_s64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svnot_s64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: not z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svnot_s64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: not z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.not.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svnot_s64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svnot_s64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: not z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svnot_s64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: not z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.not.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svnot_s64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svnot_s64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: not z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svnot_s64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: not z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.not.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svnot_u64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svnot_u64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: not z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svnot_u64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: not z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.not.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
>From 34142d3b71dd58a235918471436e2155732cccaf Mon Sep 17 00:00:00 2001
From: Momchil Velikov <momchil.velikov at arm.com>
Date: Fri, 20 Dec 2024 16:34:33 +0000
Subject: [PATCH 06/16] [fixup] Rebase, update tests, add tests for the new
all-true patterns, change undef to poison
---
.../lib/Target/AArch64/AArch64SVEInstrInfo.td | 2 +-
llvm/lib/Target/AArch64/SVEInstrFormats.td | 8 +-
.../AArch64/zeroing-forms-counts-not.ll | 1002 ++++++++++++++++-
3 files changed, 954 insertions(+), 58 deletions(-)
diff --git a/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td b/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
index 5c257c436fbeac..c43e9e4a46f36e 100644
--- a/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
+++ b/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
@@ -4318,7 +4318,7 @@ let Predicates = [HasSVE2p2orSME2p2] in {
defm CLZ_ZPzZ : sve_int_un_pred_arit_bitwise_z<0b001, "clz", AArch64clz_mt>;
defm CNT_ZPzZ : sve_int_un_pred_arit_bitwise_z<0b010, "cnt", AArch64cnt_mt>;
defm CNOT_ZPzZ : sve_int_un_pred_arit_bitwise_z<0b011, "cnot", AArch64cnot_mt>;
- defm NOT_ZPzZ : sve_int_un_pred_arit_bitwise_z<0b110, " not", AArch64not_mt>;
+ defm NOT_ZPzZ : sve_int_un_pred_arit_bitwise_z<0b110, "not", AArch64not_mt>;
// floating point
defm FABS_ZPzZ : sve_int_un_pred_arit_bitwise_fp_z<0b100, "fabs", AArch64fabs_mt>;
diff --git a/llvm/lib/Target/AArch64/SVEInstrFormats.td b/llvm/lib/Target/AArch64/SVEInstrFormats.td
index a9a23574de5d1d..959f98555466be 100644
--- a/llvm/lib/Target/AArch64/SVEInstrFormats.td
+++ b/llvm/lib/Target/AArch64/SVEInstrFormats.td
@@ -4998,10 +4998,10 @@ multiclass sve_int_un_pred_arit_bitwise_z<bits<3> opc, string asm, SDPatternOper
def _S : sve_int_un_pred_arit_z<0b10, { opc, 0b1 }, asm, ZPR32>;
def _D : sve_int_un_pred_arit_z<0b11, { opc, 0b1 }, asm, ZPR64>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv16i8, op, nxv16i1, nxv16i8, !cast<Instruction>(NAME # _B)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv8i16, op, nxv8i1, nxv8i16, !cast<Instruction>(NAME # _H)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv4i32, op, nxv4i1, nxv4i32, !cast<Instruction>(NAME # _S)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv2i64, op, nxv2i1, nxv2i64, !cast<Instruction>(NAME # _D)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv16i8, op, nxv16i1, nxv16i8, !cast<Instruction>(NAME # _B)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv8i16, op, nxv8i1, nxv8i16, !cast<Instruction>(NAME # _H)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv4i32, op, nxv4i1, nxv4i32, !cast<Instruction>(NAME # _S)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv2i64, op, nxv2i1, nxv2i64, !cast<Instruction>(NAME # _D)>;
}
multiclass sve_int_un_pred_arit_bitwise_fp<bits<3> opc, string asm,
diff --git a/llvm/test/CodeGen/AArch64/zeroing-forms-counts-not.ll b/llvm/test/CodeGen/AArch64/zeroing-forms-counts-not.ll
index efdf4bd6a5c64b..f7970ca81f6084 100644
--- a/llvm/test/CodeGen/AArch64/zeroing-forms-counts-not.ll
+++ b/llvm/test/CodeGen/AArch64/zeroing-forms-counts-not.ll
@@ -18,7 +18,7 @@ define <vscale x 16 x i8> @test_svcls_s8_x_1(<vscale x 16 x i1> %pg, <vscale x 1
; CHECK-2p2-NEXT: cls z0.b, p0/z, z0.b
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.cls.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.cls.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
ret <vscale x 16 x i8> %0
}
@@ -34,7 +34,7 @@ define <vscale x 16 x i8> @test_svcls_s8_x_2(<vscale x 16 x i1> %pg, double %z0,
; CHECK-2p2-NEXT: cls z0.b, p0/z, z1.b
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.cls.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.cls.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
ret <vscale x 16 x i8> %0
}
@@ -65,7 +65,7 @@ define <vscale x 8 x i16> @test_svcls_s16_x_1(<vscale x 8 x i1> %pg, <vscale x 8
; CHECK-2p2-NEXT: cls z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cls.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cls.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x i16> %0
}
@@ -81,7 +81,7 @@ define <vscale x 8 x i16> @test_svcls_s16_x_2(<vscale x 8 x i1> %pg, double %z0,
; CHECK-2p2-NEXT: cls z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cls.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cls.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x i16> %0
}
@@ -112,7 +112,7 @@ define <vscale x 4 x i32> @test_svcls_s32_x_1(<vscale x 4 x i1> %pg, <vscale x 4
; CHECK-2p2-NEXT: cls z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cls.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cls.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -128,7 +128,7 @@ define <vscale x 4 x i32> @test_svcls_s32_x_2(<vscale x 4 x i1> %pg, double %z0,
; CHECK-2p2-NEXT: cls z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cls.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cls.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -159,7 +159,7 @@ define <vscale x 2 x i64> @test_svcls_s64_x_1(<vscale x 2 x i1> %pg, <vscale x 2
; CHECK-2p2-NEXT: cls z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cls.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cls.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -175,7 +175,7 @@ define <vscale x 2 x i64> @test_svcls_s64_x_2(<vscale x 2 x i1> %pg, double %z0,
; CHECK-2p2-NEXT: cls z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cls.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cls.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -206,7 +206,7 @@ define <vscale x 16 x i8> @test_svclz_s8_x_1(<vscale x 16 x i1> %pg, <vscale x 1
; CHECK-2p2-NEXT: clz z0.b, p0/z, z0.b
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.clz.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.clz.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
ret <vscale x 16 x i8> %0
}
@@ -222,7 +222,7 @@ define <vscale x 16 x i8> @test_svclz_s8_x_2(<vscale x 16 x i1> %pg, double %z0,
; CHECK-2p2-NEXT: clz z0.b, p0/z, z1.b
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.clz.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.clz.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
ret <vscale x 16 x i8> %0
}
@@ -253,7 +253,7 @@ define <vscale x 8 x i16> @test_svclz_s16_x_1(<vscale x 8 x i1> %pg, <vscale x 8
; CHECK-2p2-NEXT: clz z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.clz.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.clz.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x i16> %0
}
@@ -269,7 +269,7 @@ define <vscale x 8 x i16> @test_svclz_s16_x_2(<vscale x 8 x i1> %pg, double %z0,
; CHECK-2p2-NEXT: clz z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.clz.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.clz.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x i16> %0
}
@@ -300,7 +300,7 @@ define <vscale x 4 x i32> @test_svclz_s32_x_1(<vscale x 4 x i1> %pg, <vscale x 4
; CHECK-2p2-NEXT: clz z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.clz.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.clz.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -316,7 +316,7 @@ define <vscale x 4 x i32> @test_svclz_s32_x_2(<vscale x 4 x i1> %pg, double %z0,
; CHECK-2p2-NEXT: clz z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.clz.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.clz.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -347,7 +347,7 @@ define <vscale x 2 x i64> @test_svclz_s64_x_1(<vscale x 2 x i1> %pg, <vscale x 2
; CHECK-2p2-NEXT: clz z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.clz.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.clz.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -363,7 +363,7 @@ define <vscale x 2 x i64> @test_svclz_s64_x_2(<vscale x 2 x i1> %pg, double %z0,
; CHECK-2p2-NEXT: clz z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.clz.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.clz.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -394,7 +394,7 @@ define <vscale x 16 x i8> @test_svcnt_s8_x_1(<vscale x 16 x i1> %pg, <vscale x 1
; CHECK-2p2-NEXT: cnt z0.b, p0/z, z0.b
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.cnt.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.cnt.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
ret <vscale x 16 x i8> %0
}
@@ -410,7 +410,7 @@ define <vscale x 16 x i8> @test_svcnt_s8_x_2(<vscale x 16 x i1> %pg, double %z0,
; CHECK-2p2-NEXT: cnt z0.b, p0/z, z1.b
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.cnt.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.cnt.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
ret <vscale x 16 x i8> %0
}
@@ -441,7 +441,7 @@ define <vscale x 8 x i16> @test_svcnt_s16_x_1(<vscale x 8 x i1> %pg, <vscale x 8
; CHECK-2p2-NEXT: cnt z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnt.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnt.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x i16> %0
}
@@ -457,7 +457,7 @@ define <vscale x 8 x i16> @test_svcnt_s16_x_2(<vscale x 8 x i1> %pg, double %z0,
; CHECK-2p2-NEXT: cnt z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnt.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnt.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x i16> %0
}
@@ -488,7 +488,7 @@ define <vscale x 4 x i32> @test_svcnt_s32_x_1(<vscale x 4 x i1> %pg, <vscale x 4
; CHECK-2p2-NEXT: cnt z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cnt.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cnt.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -504,7 +504,7 @@ define <vscale x 4 x i32> @test_svcnt_s32_x_2(<vscale x 4 x i1> %pg, double %z0,
; CHECK-2p2-NEXT: cnt z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cnt.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cnt.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -535,7 +535,7 @@ define <vscale x 2 x i64> @test_svcnt_s64_x_1(<vscale x 2 x i1> %pg, <vscale x 2
; CHECK-2p2-NEXT: cnt z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cnt.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cnt.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -551,7 +551,7 @@ define <vscale x 2 x i64> @test_svcnt_s64_x_2(<vscale x 2 x i1> %pg, double %z0,
; CHECK-2p2-NEXT: cnt z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cnt.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cnt.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -582,7 +582,7 @@ define <vscale x 8 x i16> @test_svcnt_f16_x_1(<vscale x 8 x i1> %pg, <vscale x 8
; CHECK-2p2-NEXT: cnt z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnt.nxv8f16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnt.nxv8f16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x i16> %0
}
@@ -598,7 +598,7 @@ define <vscale x 8 x i16> @test_svcnt_f16_x_2(<vscale x 8 x i1> %pg, double %z0,
; CHECK-2p2-NEXT: cnt z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnt.nxv8f16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnt.nxv8f16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x i16> %0
}
@@ -629,7 +629,7 @@ define <vscale x 8 x i16> @test_svcnt_bf16_x_1(<vscale x 8 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: cnt z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnt.nxv8bf16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x bfloat> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnt.nxv8bf16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x bfloat> %x)
ret <vscale x 8 x i16> %0
}
@@ -645,7 +645,7 @@ define <vscale x 8 x i16> @test_svcnt_bf16_x_2(<vscale x 8 x i1> %pg, double %z0
; CHECK-2p2-NEXT: cnt z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnt.nxv8bf16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x bfloat> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnt.nxv8bf16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x bfloat> %x)
ret <vscale x 8 x i16> %0
}
@@ -676,7 +676,7 @@ define <vscale x 4 x i32> @test_svcnt_f32_x_1(<vscale x 4 x i1> %pg, <vscale x 4
; CHECK-2p2-NEXT: cnt z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cnt.nxv4f32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cnt.nxv4f32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x i32> %0
}
@@ -692,7 +692,7 @@ define <vscale x 4 x i32> @test_svcnt_f32_x_2(<vscale x 4 x i1> %pg, double %z0,
; CHECK-2p2-NEXT: cnt z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cnt.nxv4f32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cnt.nxv4f32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x i32> %0
}
@@ -723,7 +723,7 @@ define <vscale x 2 x i64> @test_svcnt_f64_x_1(<vscale x 2 x i1> %pg, <vscale x 2
; CHECK-2p2-NEXT: cnt z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cnt.nxv2f64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cnt.nxv2f64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x i64> %0
}
@@ -739,7 +739,7 @@ define <vscale x 2 x i64> @test_svcnt_f64_x_2(<vscale x 2 x i1> %pg, double %z0,
; CHECK-2p2-NEXT: cnt z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cnt.nxv2f64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cnt.nxv2f64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x i64> %0
}
@@ -770,7 +770,7 @@ define <vscale x 16 x i8> @test_svcnot_s8_x_1(<vscale x 16 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: cnot z0.b, p0/z, z0.b
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.cnot.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.cnot.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
ret <vscale x 16 x i8> %0
}
@@ -786,7 +786,7 @@ define <vscale x 16 x i8> @test_svcnot_s8_x_2(<vscale x 16 x i1> %pg, double %z0
; CHECK-2p2-NEXT: cnot z0.b, p0/z, z1.b
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.cnot.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.cnot.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
ret <vscale x 16 x i8> %0
}
@@ -817,7 +817,7 @@ define <vscale x 8 x i16> @test_svcnot_s16_x_1(<vscale x 8 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: cnot z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnot.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnot.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x i16> %0
}
@@ -833,7 +833,7 @@ define <vscale x 8 x i16> @test_svcnot_s16_x_2(<vscale x 8 x i1> %pg, double %z0
; CHECK-2p2-NEXT: cnot z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnot.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnot.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x i16> %0
}
@@ -864,7 +864,7 @@ define <vscale x 4 x i32> @test_svcnot_s32_x_1(<vscale x 4 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: cnot z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cnot.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cnot.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -880,7 +880,7 @@ define <vscale x 4 x i32> @test_svcnot_s32_x_2(<vscale x 4 x i1> %pg, double %z0
; CHECK-2p2-NEXT: cnot z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cnot.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cnot.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -911,7 +911,7 @@ define <vscale x 2 x i64> @test_svcnot_s64_x_1(<vscale x 2 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: cnot z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cnot.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cnot.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -927,7 +927,7 @@ define <vscale x 2 x i64> @test_svcnot_s64_x_2(<vscale x 2 x i1> %pg, double %z0
; CHECK-2p2-NEXT: cnot z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cnot.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cnot.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -958,7 +958,7 @@ define <vscale x 16 x i8> @test_svnot_s8_x_1(<vscale x 16 x i1> %pg, <vscale x 1
; CHECK-2p2-NEXT: not z0.b, p0/z, z0.b
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.not.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.not.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
ret <vscale x 16 x i8> %0
}
@@ -974,7 +974,7 @@ define <vscale x 16 x i8> @test_svnot_s8_x_2(<vscale x 16 x i1> %pg, double %z0,
; CHECK-2p2-NEXT: not z0.b, p0/z, z1.b
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.not.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.not.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
ret <vscale x 16 x i8> %0
}
@@ -1005,7 +1005,7 @@ define <vscale x 8 x i16> @test_svnot_s16_x_1(<vscale x 8 x i1> %pg, <vscale x 8
; CHECK-2p2-NEXT: not z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.not.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.not.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x i16> %0
}
@@ -1021,7 +1021,7 @@ define <vscale x 8 x i16> @test_svnot_s16_x_2(<vscale x 8 x i1> %pg, double %z0,
; CHECK-2p2-NEXT: not z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.not.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.not.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x i16> %0
}
@@ -1052,7 +1052,7 @@ define <vscale x 4 x i32> @test_svnot_s32_x_1(<vscale x 4 x i1> %pg, <vscale x 4
; CHECK-2p2-NEXT: not z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.not.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.not.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -1068,7 +1068,7 @@ define <vscale x 4 x i32> @test_svnot_s32_x_2(<vscale x 4 x i1> %pg, double %z0,
; CHECK-2p2-NEXT: not z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.not.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.not.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -1099,7 +1099,7 @@ define <vscale x 2 x i64> @test_svnot_s64_x_1(<vscale x 2 x i1> %pg, <vscale x 2
; CHECK-2p2-NEXT: not z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.not.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.not.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -1115,7 +1115,7 @@ define <vscale x 2 x i64> @test_svnot_s64_x_2(<vscale x 2 x i1> %pg, double %z0,
; CHECK-2p2-NEXT: not z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.not.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.not.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -1135,18 +1135,914 @@ entry:
ret <vscale x 2 x i64> %0
}
-define <vscale x 2 x i64> @test_svnot_u64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
-; CHECK-LABEL: test_svnot_u64_z:
+define <vscale x 16 x i8> @test_svcls_nxv16i8_ptrue_u(double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svcls_nxv16i8_ptrue_u:
; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: ptrue p0.b
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cls z0.b, p0/m, z1.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcls_nxv16i8_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.b
+; CHECK-2p2-NEXT: cls z0.b, p0/z, z1.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.cls.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svcls_nxv16i8_ptrue(double %z0, <vscale x 16 x i8> %x, <vscale x 16 x i8> %y) {
+; CHECK-LABEL: test_svcls_nxv16i8_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.b
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: cls z0.b, p0/m, z2.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcls_nxv16i8_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.b
+; CHECK-2p2-NEXT: cls z0.b, p0/z, z2.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.cls.nxv16i8(<vscale x 16 x i8> %x, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %y)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 8 x i16> @test_svcls_nxv8i16_ptrue_u(double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svcls_nxv8i16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cls z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcls_nxv8i16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: cls z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cls.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svcls_nxv8i16_ptrue(double %z0, <vscale x 8 x i16> %x, <vscale x 8 x i16> %y) {
+; CHECK-LABEL: test_svcls_nxv8i16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: cls z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcls_nxv8i16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: cls z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cls.nxv8i16(<vscale x 8 x i16> %x, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %y)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svcls_nxv4i32_ptrue_u(double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svcls_nxv4i32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cls z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcls_nxv4i32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: cls z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cls.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svcls_nxv4i32_ptrue(double %z0, <vscale x 4 x i32> %x, <vscale x 4 x i32> %y) {
+; CHECK-LABEL: test_svcls_nxv4i32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: cls z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcls_nxv4i32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: cls z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cls.nxv4i32(<vscale x 4 x i32> %x, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %y)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svcls_nxv2i64_ptrue_u(double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svcls_nxv2i64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cls z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcls_nxv2i64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: cls z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cls.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svcls_nxv2i64_ptrue(double %z0, <vscale x 2 x i64> %x, <vscale x 2 x i64> %y) {
+; CHECK-LABEL: test_svcls_nxv2i64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: cls z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcls_nxv2i64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: cls z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cls.nxv2i64(<vscale x 2 x i64> %x, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %y)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 16 x i8> @test_svclz_nxv16i8_ptrue_u(double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svclz_nxv16i8_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.b
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: clz z0.b, p0/m, z1.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svclz_nxv16i8_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.b
+; CHECK-2p2-NEXT: clz z0.b, p0/z, z1.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.clz.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svclz_nxv16i8_ptrue(double %z0, <vscale x 16 x i8> %x, <vscale x 16 x i8> %y) {
+; CHECK-LABEL: test_svclz_nxv16i8_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.b
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: clz z0.b, p0/m, z2.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svclz_nxv16i8_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.b
+; CHECK-2p2-NEXT: clz z0.b, p0/z, z2.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.clz.nxv16i8(<vscale x 16 x i8> %x, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %y)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 8 x i16> @test_svclz_nxv8i16_ptrue_u(double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svclz_nxv8i16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: clz z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svclz_nxv8i16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: clz z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.clz.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svclz_nxv8i16_ptrue(double %z0, <vscale x 8 x i16> %x, <vscale x 8 x i16> %y) {
+; CHECK-LABEL: test_svclz_nxv8i16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: clz z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svclz_nxv8i16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: clz z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.clz.nxv8i16(<vscale x 8 x i16> %x, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %y)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svclz_nxv4i32_ptrue_u(double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svclz_nxv4i32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: clz z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svclz_nxv4i32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: clz z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.clz.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svclz_nxv4i32_ptrue(double %z0, <vscale x 4 x i32> %x, <vscale x 4 x i32> %y) {
+; CHECK-LABEL: test_svclz_nxv4i32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: clz z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svclz_nxv4i32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: clz z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.clz.nxv4i32(<vscale x 4 x i32> %x, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %y)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svclz_nxv2i64_ptrue_u(double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svclz_nxv2i64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: clz z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svclz_nxv2i64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: clz z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.clz.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svclz_nxv2i64_ptrue(double %z0, <vscale x 2 x i64> %x, <vscale x 2 x i64> %y) {
+; CHECK-LABEL: test_svclz_nxv2i64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: clz z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svclz_nxv2i64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: clz z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.clz.nxv2i64(<vscale x 2 x i64> %x, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %y)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 16 x i8> @test_svcnt_nxv16i8_ptrue_u(double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svcnt_nxv16i8_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.b
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cnt z0.b, p0/m, z1.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_nxv16i8_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.b
+; CHECK-2p2-NEXT: cnt z0.b, p0/z, z1.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.cnt.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svcnt_nxv16i8_ptrue(double %z0, <vscale x 16 x i8> %x, <vscale x 16 x i8> %y) {
+; CHECK-LABEL: test_svcnt_nxv16i8_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.b
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: cnt z0.b, p0/m, z2.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_nxv16i8_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.b
+; CHECK-2p2-NEXT: cnt z0.b, p0/z, z2.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.cnt.nxv16i8(<vscale x 16 x i8> %x, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %y)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 8 x i16> @test_svcnt_nxv8i16_ptrue_u(double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svcnt_nxv8i16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cnt z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_nxv8i16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: cnt z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnt.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svcnt_nxv8i16_ptrue(double %z0, <vscale x 8 x i16> %x, <vscale x 8 x i16> %y) {
+; CHECK-LABEL: test_svcnt_nxv8i16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: cnt z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_nxv8i16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: cnt z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnt.nxv8i16(<vscale x 8 x i16> %x, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %y)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svcnt_nxv4i32_ptrue_u(double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svcnt_nxv4i32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cnt z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_nxv4i32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: cnt z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cnt.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svcnt_nxv4i32_ptrue(double %z0, <vscale x 4 x i32> %x, <vscale x 4 x i32> %y) {
+; CHECK-LABEL: test_svcnt_nxv4i32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: cnt z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_nxv4i32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: cnt z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cnt.nxv4i32(<vscale x 4 x i32> %x, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %y)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svcnt_nxv2i64_ptrue_u(double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svcnt_nxv2i64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cnt z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_nxv2i64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: cnt z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cnt.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svcnt_nxv2i64_ptrue(double %z0, <vscale x 2 x i64> %x, <vscale x 2 x i64> %y) {
+; CHECK-LABEL: test_svcnt_nxv2i64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: cnt z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_nxv2i64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: cnt z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cnt.nxv2i64(<vscale x 2 x i64> %x, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %y)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 8 x i16> @test_svcnt_nxv8f16_ptrue_u(double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svcnt_nxv8f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cnt z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_nxv8f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: cnt z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnt.nxv8f16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svcnt_nxv8f16_ptrue(double %z0, <vscale x 8 x i16> %x, <vscale x 8 x half> %y) {
+; CHECK-LABEL: test_svcnt_nxv8f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: cnt z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_nxv8f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: cnt z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnt.nxv8f16(<vscale x 8 x i16> %x, <vscale x 8 x i1> %pg, <vscale x 8 x half> %y)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svcnt_nxv8bf16_ptrue_u(double %z0, <vscale x 8 x bfloat> %x) {
+; CHECK-LABEL: test_svcnt_nxv8bf16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cnt z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_nxv8bf16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: cnt z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnt.nxv8bf16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x bfloat> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svcnt_nxv8bf16_ptrue(double %z0, <vscale x 8 x i16> %x, <vscale x 8 x bfloat> %y) {
+; CHECK-LABEL: test_svcnt_nxv8bf16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: cnt z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_nxv8bf16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: cnt z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnt.nxv8bf16(<vscale x 8 x i16> %x, <vscale x 8 x i1> %pg, <vscale x 8 x bfloat> %y)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svcnt_nxv4f32_ptrue_u(double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svcnt_nxv4f32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cnt z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_nxv4f32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: cnt z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cnt.nxv4f32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svcnt_nxv4f32_ptrue(double %z0, <vscale x 4 x i32> %x, <vscale x 4 x float> %y) {
+; CHECK-LABEL: test_svcnt_nxv4f32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: cnt z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_nxv4f32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: cnt z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cnt.nxv4f32(<vscale x 4 x i32> %x, <vscale x 4 x i1> %pg, <vscale x 4 x float> %y)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svcnt_nxv2f64_ptrue_u(double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svcnt_nxv2f64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cnt z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_nxv2f64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: cnt z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cnt.nxv2f64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svcnt_nxv2f64_ptrue(double %z0, <vscale x 2 x i64> %x, <vscale x 2 x double> %y) {
+; CHECK-LABEL: test_svcnt_nxv2f64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: cnt z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnt_nxv2f64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: cnt z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cnt.nxv2f64(<vscale x 2 x i64> %x, <vscale x 2 x i1> %pg, <vscale x 2 x double> %y)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 16 x i8> @test_svcnot_nxv16i8_ptrue_u(double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svcnot_nxv16i8_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.b
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cnot z0.b, p0/m, z1.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnot_nxv16i8_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.b
+; CHECK-2p2-NEXT: cnot z0.b, p0/z, z1.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.cnot.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svcnot_nxv16i8_ptrue(double %z0, <vscale x 16 x i8> %x, <vscale x 16 x i8> %y) {
+; CHECK-LABEL: test_svcnot_nxv16i8_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.b
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: cnot z0.b, p0/m, z2.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnot_nxv16i8_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.b
+; CHECK-2p2-NEXT: cnot z0.b, p0/z, z2.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.cnot.nxv16i8(<vscale x 16 x i8> %x, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %y)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 8 x i16> @test_svcnot_nxv8i16_ptrue_u(double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svcnot_nxv8i16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cnot z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnot_nxv8i16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: cnot z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnot.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svcnot_nxv8i16_ptrue(double %z0, <vscale x 8 x i16> %x, <vscale x 8 x i16> %y) {
+; CHECK-LABEL: test_svcnot_nxv8i16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: cnot z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnot_nxv8i16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: cnot z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.cnot.nxv8i16(<vscale x 8 x i16> %x, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %y)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svcnot_nxv4i32_ptrue_u(double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svcnot_nxv4i32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cnot z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnot_nxv4i32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: cnot z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cnot.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svcnot_nxv4i32_ptrue(double %z0, <vscale x 4 x i32> %x, <vscale x 4 x i32> %y) {
+; CHECK-LABEL: test_svcnot_nxv4i32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: cnot z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnot_nxv4i32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: cnot z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.cnot.nxv4i32(<vscale x 4 x i32> %x, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %y)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svcnot_nxv2i64_ptrue_u(double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svcnot_nxv2i64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: cnot z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnot_nxv2i64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: cnot z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cnot.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svcnot_nxv2i64_ptrue(double %z0, <vscale x 2 x i64> %x, <vscale x 2 x i64> %y) {
+; CHECK-LABEL: test_svcnot_nxv2i64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: cnot z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svcnot_nxv2i64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: cnot z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.cnot.nxv2i64(<vscale x 2 x i64> %x, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %y)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 16 x i8> @test_svnot_nxv16i8_ptrue_u(double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svnot_nxv16i8_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.b
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: not z0.b, p0/m, z1.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svnot_nxv16i8_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.b
+; CHECK-2p2-NEXT: not z0.b, p0/z, z1.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.not.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svnot_nxv16i8_ptrue(double %z0, <vscale x 16 x i8> %x, <vscale x 16 x i8> %y) {
+; CHECK-LABEL: test_svnot_nxv16i8_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.b
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: not z0.b, p0/m, z2.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svnot_nxv16i8_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.b
+; CHECK-2p2-NEXT: not z0.b, p0/z, z2.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.not.nxv16i8(<vscale x 16 x i8> %x, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %y)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 8 x i16> @test_svnot_nxv8i16_ptrue_u(double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svnot_nxv8i16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: not z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svnot_nxv8i16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: not z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.not.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svnot_nxv8i16_ptrue(double %z0, <vscale x 8 x i16> %x, <vscale x 8 x i16> %y) {
+; CHECK-LABEL: test_svnot_nxv8i16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: not z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svnot_nxv8i16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: not z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.not.nxv8i16(<vscale x 8 x i16> %x, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %y)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svnot_nxv4i32_ptrue_u(double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svnot_nxv4i32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: not z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svnot_nxv4i32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: not z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.not.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svnot_nxv4i32_ptrue(double %z0, <vscale x 4 x i32> %x, <vscale x 4 x i32> %y) {
+; CHECK-LABEL: test_svnot_nxv4i32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: not z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svnot_nxv4i32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: not z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.not.nxv4i32(<vscale x 4 x i32> %x, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %y)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svnot_nxv2i64_ptrue_u(double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svnot_nxv2i64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
; CHECK-NEXT: not z0.d, p0/m, z1.d
; CHECK-NEXT: ret
;
-; CHECK-2p2-LABEL: test_svnot_u64_z:
+; CHECK-2p2-LABEL: test_svnot_nxv2i64_ptrue_u:
; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
; CHECK-2p2-NEXT: not z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.not.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.not.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svnot_nxv2i64_ptrue(double %z0, <vscale x 2 x i64> %x, <vscale x 2 x i64> %y) {
+; CHECK-LABEL: test_svnot_nxv2i64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: not z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svnot_nxv2i64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: not z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.not.nxv2i64(<vscale x 2 x i64> %x, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %y)
ret <vscale x 2 x i64> %0
}
>From 1965e53776672e5358213626decb90764fff8868 Mon Sep 17 00:00:00 2001
From: Momchil Velikov <momchil.velikov at arm.com>
Date: Tue, 19 Nov 2024 12:23:46 +0000
Subject: [PATCH 07/16] [AArch64] Generate zeroing forms of certain SVE2.2
instructions (7/11)
---
.../lib/Target/AArch64/AArch64SVEInstrInfo.td | 2 +-
llvm/lib/Target/AArch64/SVEInstrFormats.td | 6 +-
.../CodeGen/AArch64/zeroing-forms-flogb.ll | 146 ++++++++++++++++++
3 files changed, 152 insertions(+), 2 deletions(-)
create mode 100644 llvm/test/CodeGen/AArch64/zeroing-forms-flogb.ll
diff --git a/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td b/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
index c43e9e4a46f36e..7e57480f2e39bb 100644
--- a/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
+++ b/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
@@ -4280,7 +4280,7 @@ let Predicates = [HasSVE2p2orSME2p2] in {
defm SCVTF_ZPzZ : sve_fp_z2op_p_zd_c<0b0, "scvtf", "int_aarch64_sve_scvtf", AArch64scvtf_mt>;
defm UCVTF_ZPzZ : sve_fp_z2op_p_zd_c<0b1, "ucvtf", "int_aarch64_sve_ucvtf", AArch64ucvtf_mt>;
// Signed integer base 2 logarithm of fp value, zeroing predicate
- defm FLOGB_ZPzZ : sve_fp_z2op_p_zd_d_flogb<"flogb">;
+ defm FLOGB_ZPzZ : sve_fp_z2op_p_zd_d_flogb<"flogb", int_aarch64_sve_flogb>;
// SVE2 integer unary operations, zeroing predicate
def URECPE_ZPzZ : sve2_int_un_pred_arit_z<0b10, 0b00, "urecpe", ZPR32>;
diff --git a/llvm/lib/Target/AArch64/SVEInstrFormats.td b/llvm/lib/Target/AArch64/SVEInstrFormats.td
index 959f98555466be..0e501a8ca254e5 100644
--- a/llvm/lib/Target/AArch64/SVEInstrFormats.td
+++ b/llvm/lib/Target/AArch64/SVEInstrFormats.td
@@ -3342,10 +3342,14 @@ multiclass sve_fp_z2op_p_zd_c<bit U, string asm, string int_op, SDPatternOperato
defm : SVE_1_Op_PassthruUndefZero_Pat<nxv2f64, ir_op, nxv2i1,nxv2i64, !cast<Instruction>(NAME # _DtoD)>;
}
-multiclass sve_fp_z2op_p_zd_d_flogb<string asm> {
+multiclass sve_fp_z2op_p_zd_d_flogb<string asm, SDPatternOperator op> {
def _H : sve_fp_z2op_p_zd<0b0011001, asm, ZPR16, ZPR16>;
def _S : sve_fp_z2op_p_zd<0b0011010, asm, ZPR32, ZPR32>;
def _D : sve_fp_z2op_p_zd<0b0011011, asm, ZPR64, ZPR64>;
+
+ def : SVE_3_Op_UndefZero_Pat<nxv8i16, op, nxv8i16, nxv8i1, nxv8f16, !cast<Instruction>(NAME # _H)>;
+ def : SVE_3_Op_UndefZero_Pat<nxv4i32, op, nxv4i32, nxv4i1, nxv4f32, !cast<Instruction>(NAME # _S)>;
+ def : SVE_3_Op_UndefZero_Pat<nxv2i64, op, nxv2i64, nxv2i1, nxv2f64, !cast<Instruction>(NAME # _D)>;
}
multiclass sve_fp_z2op_p_zd_b_0<string asm, string op> {
diff --git a/llvm/test/CodeGen/AArch64/zeroing-forms-flogb.ll b/llvm/test/CodeGen/AArch64/zeroing-forms-flogb.ll
new file mode 100644
index 00000000000000..8d799cd3224212
--- /dev/null
+++ b/llvm/test/CodeGen/AArch64/zeroing-forms-flogb.ll
@@ -0,0 +1,146 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
+; RUN: llc -mattr=+sve2 < %s | FileCheck %s
+; RUN: llc -mattr=+sve2p2 < %s | FileCheck %s -check-prefix CHECK-2p2
+
+; RUN: llc -mattr=+sme2 -force-streaming < %s | FileCheck %s
+; RUN: llc -mattr=+sme2p2 -force-streaming < %s | FileCheck %s -check-prefix CHECK-2p2
+
+target triple = "aarch64-linux"
+
+define <vscale x 8 x i16> @test_svlogb_f16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svlogb_f16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: flogb z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svlogb_f16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: flogb z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.flogb.nxv8f16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svlogb_f16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svlogb_f16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: flogb z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svlogb_f16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: flogb z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.flogb.nxv8f16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svlogb_f16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svlogb_f16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: flogb z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svlogb_f16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: flogb z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.flogb.nxv8f16(<vscale x 8 x i16> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svlogb_f32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svlogb_f32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: flogb z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svlogb_f32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: flogb z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.flogb.nxv4f32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svlogb_f32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svlogb_f32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: flogb z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svlogb_f32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: flogb z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.flogb.nxv4f32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svlogb_f32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svlogb_f32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: flogb z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svlogb_f32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: flogb z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.flogb.nxv4f32(<vscale x 4 x i32> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svlogb_f64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svlogb_f64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: flogb z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svlogb_f64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: flogb z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.flogb.nxv2f64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svlogb_f64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svlogb_f64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: flogb z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svlogb_f64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: flogb z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.flogb.nxv2f64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svlogb_f64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svlogb_f64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: flogb z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svlogb_f64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: flogb z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.flogb.nxv2f64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x i64> %0
+}
>From e22e4caf7eb011ccf1c0aa98e1b2b9c43b477c84 Mon Sep 17 00:00:00 2001
From: Momchil Velikov <momchil.velikov at arm.com>
Date: Fri, 20 Dec 2024 17:10:54 +0000
Subject: [PATCH 08/16] [fixup] Rebase, update tests, add tests for the new
all-true patterns, change undef to poison
---
llvm/lib/Target/AArch64/SVEInstrFormats.td | 6 +-
.../CodeGen/AArch64/zeroing-forms-flogb.ll | 124 +++++++++++++++++-
2 files changed, 121 insertions(+), 9 deletions(-)
diff --git a/llvm/lib/Target/AArch64/SVEInstrFormats.td b/llvm/lib/Target/AArch64/SVEInstrFormats.td
index 0e501a8ca254e5..a985d3a8b252c8 100644
--- a/llvm/lib/Target/AArch64/SVEInstrFormats.td
+++ b/llvm/lib/Target/AArch64/SVEInstrFormats.td
@@ -3347,9 +3347,9 @@ multiclass sve_fp_z2op_p_zd_d_flogb<string asm, SDPatternOperator op> {
def _S : sve_fp_z2op_p_zd<0b0011010, asm, ZPR32, ZPR32>;
def _D : sve_fp_z2op_p_zd<0b0011011, asm, ZPR64, ZPR64>;
- def : SVE_3_Op_UndefZero_Pat<nxv8i16, op, nxv8i16, nxv8i1, nxv8f16, !cast<Instruction>(NAME # _H)>;
- def : SVE_3_Op_UndefZero_Pat<nxv4i32, op, nxv4i32, nxv4i1, nxv4f32, !cast<Instruction>(NAME # _S)>;
- def : SVE_3_Op_UndefZero_Pat<nxv2i64, op, nxv2i64, nxv2i1, nxv2f64, !cast<Instruction>(NAME # _D)>;
+ defm : SVE_3_Op_UndefZero_Pat<nxv8i16, op, nxv8i16, nxv8i1, nxv8f16, !cast<Instruction>(NAME # _H)>;
+ defm : SVE_3_Op_UndefZero_Pat<nxv4i32, op, nxv4i32, nxv4i1, nxv4f32, !cast<Instruction>(NAME # _S)>;
+ defm : SVE_3_Op_UndefZero_Pat<nxv2i64, op, nxv2i64, nxv2i1, nxv2f64, !cast<Instruction>(NAME # _D)>;
}
multiclass sve_fp_z2op_p_zd_b_0<string asm, string op> {
diff --git a/llvm/test/CodeGen/AArch64/zeroing-forms-flogb.ll b/llvm/test/CodeGen/AArch64/zeroing-forms-flogb.ll
index 8d799cd3224212..23620a3419b999 100644
--- a/llvm/test/CodeGen/AArch64/zeroing-forms-flogb.ll
+++ b/llvm/test/CodeGen/AArch64/zeroing-forms-flogb.ll
@@ -18,7 +18,7 @@ define <vscale x 8 x i16> @test_svlogb_f16_x_1(<vscale x 8 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: flogb z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.flogb.nxv8f16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.flogb.nxv8f16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x i16> %0
}
@@ -33,7 +33,7 @@ define <vscale x 8 x i16> @test_svlogb_f16_x_2(<vscale x 8 x i1> %pg, double %z0
; CHECK-2p2-NEXT: flogb z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.flogb.nxv8f16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.flogb.nxv8f16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x i16> %0
}
@@ -64,7 +64,7 @@ define <vscale x 4 x i32> @test_svlogb_f32_x_1(<vscale x 4 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: flogb z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.flogb.nxv4f32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.flogb.nxv4f32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x i32> %0
}
@@ -79,7 +79,7 @@ define <vscale x 4 x i32> @test_svlogb_f32_x_2(<vscale x 4 x i1> %pg, double %z0
; CHECK-2p2-NEXT: flogb z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.flogb.nxv4f32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.flogb.nxv4f32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x i32> %0
}
@@ -110,7 +110,7 @@ define <vscale x 2 x i64> @test_svlogb_f64_x_1(<vscale x 2 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: flogb z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.flogb.nxv2f64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.flogb.nxv2f64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x i64> %0
}
@@ -125,7 +125,7 @@ define <vscale x 2 x i64> @test_svlogb_f64_x_2(<vscale x 2 x i1> %pg, double %z0
; CHECK-2p2-NEXT: flogb z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.flogb.nxv2f64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.flogb.nxv2f64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x i64> %0
}
@@ -144,3 +144,115 @@ entry:
%0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.flogb.nxv2f64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x i64> %0
}
+
+define <vscale x 8 x i16> @test_svlogb_nxv8f16_ptrue_u(double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svlogb_nxv8f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: flogb z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svlogb_nxv8f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: flogb z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.flogb.nxv8f16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svlogb_nxv8f16_ptrue(double %z0, <vscale x 8 x i16> %x, <vscale x 8 x half> %y) {
+; CHECK-LABEL: test_svlogb_nxv8f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: flogb z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svlogb_nxv8f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: flogb z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.flogb.nxv8f16(<vscale x 8 x i16> %x, <vscale x 8 x i1> %pg, <vscale x 8 x half> %y)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svlogb_nxv4f32_ptrue_u(double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svlogb_nxv4f32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: flogb z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svlogb_nxv4f32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: flogb z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.flogb.nxv4f32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svlogb_nxv4f32_ptrue(double %z0, <vscale x 4 x i32> %x, <vscale x 4 x float> %y) {
+; CHECK-LABEL: test_svlogb_nxv4f32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: flogb z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svlogb_nxv4f32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: flogb z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.flogb.nxv4f32(<vscale x 4 x i32> %x, <vscale x 4 x i1> %pg, <vscale x 4 x float> %y)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svlogb_nxv2f64_ptrue_u(double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svlogb_nxv2f64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: flogb z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svlogb_nxv2f64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: flogb z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.flogb.nxv2f64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svlogb_nxv2f64_ptrue(double %z0, <vscale x 2 x i64> %x, <vscale x 2 x double> %y) {
+; CHECK-LABEL: test_svlogb_nxv2f64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: flogb z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svlogb_nxv2f64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: flogb z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.flogb.nxv2f64(<vscale x 2 x i64> %x, <vscale x 2 x i1> %pg, <vscale x 2 x double> %y)
+ ret <vscale x 2 x i64> %0
+}
+
>From f4f246a75e253c51e66d4ec829ad70630645d8d9 Mon Sep 17 00:00:00 2001
From: Momchil Velikov <momchil.velikov at arm.com>
Date: Tue, 19 Nov 2024 12:33:33 +0000
Subject: [PATCH 09/16] [AArch64] Generate zeroing forms of certain SVE2.2
instructions (8/11)
---
.../lib/Target/AArch64/AArch64SVEInstrInfo.td | 18 +-
llvm/lib/Target/AArch64/SVEInstrFormats.td | 9 +-
.../zeroing-forms-frint-frecpx-fsqrt.ll | 2550 +++++++++++++++++
3 files changed, 2567 insertions(+), 10 deletions(-)
create mode 100644 llvm/test/CodeGen/AArch64/zeroing-forms-frint-frecpx-fsqrt.ll
diff --git a/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td b/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
index 7e57480f2e39bb..b289c5bd3247a4 100644
--- a/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
+++ b/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
@@ -4301,17 +4301,17 @@ let Predicates = [HasSVE2p2orSME2p2] in {
defm FRINT64X_ZPzZ : sve_fp_z2op_p_zd_frint<0b11, "frint64x">;
// Floating-point round to integral fp value, zeroing predicate
- defm FRINTN_ZPzZ : sve_fp_z2op_p_zd_hsd<0b00000, "frintn">;
- defm FRINTP_ZPzZ : sve_fp_z2op_p_zd_hsd<0b00001, "frintp">;
- defm FRINTM_ZPzZ : sve_fp_z2op_p_zd_hsd<0b00010, "frintm">;
- defm FRINTZ_ZPzZ : sve_fp_z2op_p_zd_hsd<0b00011, "frintz">;
- defm FRINTA_ZPzZ : sve_fp_z2op_p_zd_hsd<0b00100, "frinta">;
- defm FRINTX_ZPzZ : sve_fp_z2op_p_zd_hsd<0b00110, "frintx">;
- defm FRINTI_ZPzZ : sve_fp_z2op_p_zd_hsd<0b00111, "frinti">;
+ defm FRINTN_ZPzZ : sve_fp_z2op_p_zd_hsd<0b00000, "frintn", AArch64frintn_mt>;
+ defm FRINTP_ZPzZ : sve_fp_z2op_p_zd_hsd<0b00001, "frintp", AArch64frintp_mt>;
+ defm FRINTM_ZPzZ : sve_fp_z2op_p_zd_hsd<0b00010, "frintm", AArch64frintm_mt>;
+ defm FRINTZ_ZPzZ : sve_fp_z2op_p_zd_hsd<0b00011, "frintz", AArch64frintz_mt>;
+ defm FRINTA_ZPzZ : sve_fp_z2op_p_zd_hsd<0b00100, "frinta", AArch64frinta_mt>;
+ defm FRINTX_ZPzZ : sve_fp_z2op_p_zd_hsd<0b00110, "frintx", AArch64frintx_mt>;
+ defm FRINTI_ZPzZ : sve_fp_z2op_p_zd_hsd<0b00111, "frinti", AArch64frinti_mt>;
// Floating-point invert exponent, zeroing predicate
- defm FRECPX_ZPzZ : sve_fp_z2op_p_zd_hsd<0b01100, "frecpx">;
+ defm FRECPX_ZPzZ : sve_fp_z2op_p_zd_hsd<0b01100, "frecpx", AArch64frecpx_mt>;
// Floating-point square root, zeroing predicate
- defm FSQRT_ZPZz : sve_fp_z2op_p_zd_hsd<0b01101, "fsqrt">;
+ defm FSQRT_ZPZz : sve_fp_z2op_p_zd_hsd<0b01101, "fsqrt", AArch64fsqrt_mt>;
// SVE2p2 integer unary arithmetic (bitwise), zeroing predicate
defm CLS_ZPzZ : sve_int_un_pred_arit_bitwise_z<0b000, "cls", AArch64cls_mt>;
diff --git a/llvm/lib/Target/AArch64/SVEInstrFormats.td b/llvm/lib/Target/AArch64/SVEInstrFormats.td
index a985d3a8b252c8..87c882817e9117 100644
--- a/llvm/lib/Target/AArch64/SVEInstrFormats.td
+++ b/llvm/lib/Target/AArch64/SVEInstrFormats.td
@@ -3287,10 +3287,17 @@ multiclass sve_fp_z2op_p_zd<string asm, SDPatternOperator op> {
defm : SVE_3_Op_UndefZero_Pat<nxv4f32, op, nxv4f32, nxv2i1, nxv2f64, !cast<Instruction>(NAME # _DtoS)>;
}
-multiclass sve_fp_z2op_p_zd_hsd<bits<5> opc, string asm> {
+multiclass sve_fp_z2op_p_zd_hsd<bits<5> opc, string asm, SDPatternOperator op> {
def _H : sve_fp_z2op_p_zd<{ 0b01, opc }, asm, ZPR16, ZPR16>;
def _S : sve_fp_z2op_p_zd<{ 0b10, opc }, asm, ZPR32, ZPR32>;
def _D : sve_fp_z2op_p_zd<{ 0b11, opc }, asm, ZPR64, ZPR64>;
+
+ def : SVE_1_Op_PassthruUndefZero_Pat<nxv8f16, op, nxv8i1, nxv8f16, !cast<Instruction>(NAME # _H)>;
+ def : SVE_1_Op_PassthruUndefZero_Pat<nxv4f16, op, nxv4i1, nxv4f16, !cast<Instruction>(NAME # _H)>;
+ def : SVE_1_Op_PassthruUndefZero_Pat<nxv2f16, op, nxv2i1, nxv2f16, !cast<Instruction>(NAME # _H)>;
+ def : SVE_1_Op_PassthruUndefZero_Pat<nxv4f32, op, nxv4i1, nxv4f32, !cast<Instruction>(NAME # _S)>;
+ def : SVE_1_Op_PassthruUndefZero_Pat<nxv2f32, op, nxv2i1, nxv2f32, !cast<Instruction>(NAME # _S)>;
+ def : SVE_1_Op_PassthruUndefZero_Pat<nxv2f64, op, nxv2i1, nxv2f64, !cast<Instruction>(NAME # _D)>;
}
multiclass sve_fp_z2op_p_zd_frint<bits<2> opc, string asm> {
diff --git a/llvm/test/CodeGen/AArch64/zeroing-forms-frint-frecpx-fsqrt.ll b/llvm/test/CodeGen/AArch64/zeroing-forms-frint-frecpx-fsqrt.ll
new file mode 100644
index 00000000000000..8de7f3db29674c
--- /dev/null
+++ b/llvm/test/CodeGen/AArch64/zeroing-forms-frint-frecpx-fsqrt.ll
@@ -0,0 +1,2550 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
+; RUN: llc -mattr=+sve < %s | FileCheck %s
+; RUN: llc -mattr=+sve2p2 < %s | FileCheck %s -check-prefix CHECK-2p2
+
+; RUN: llc -mattr=+sme -force-streaming < %s | FileCheck %s
+; RUN: llc -mattr=+sme2p2 -force-streaming < %s | FileCheck %s -check-prefix CHECK-2p2
+
+target triple = "aarch64-linux"
+
+define <vscale x 8 x half> @test_svrinta_f16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svrinta_f16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frinta z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinta_f16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinta z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frinta.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svrinta_f16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svrinta_f16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frinta z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinta_f16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinta z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frinta.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svrinta_f16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svrinta_f16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: frinta z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinta_f16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinta z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frinta.nxv8f16(<vscale x 8 x half> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 4 x half> @test_svrinta_4f16_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svrinta_4f16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frinta z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinta_4f16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinta z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frinta.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 4 x half> @test_svrinta_4f16_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svrinta_4f16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frinta z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinta_4f16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinta z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frinta.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 4 x half> @test_svrinta_4f16_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svrinta_4f16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: frinta z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinta_4f16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinta z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frinta.nxv4f16(<vscale x 4 x half> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 2 x half> @test_svrinta_2f16_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svrinta_2f16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frinta z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinta_2f16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinta z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frinta.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x half> @test_svrinta_2f16_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svrinta_2f16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frinta z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinta_2f16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinta z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frinta.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x half> @test_svrinta_2f16_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svrinta_2f16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: frinta z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinta_2f16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinta z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frinta.nxv2f16(<vscale x 2 x half> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x float> @test_svrinta_2f32_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svrinta_2f32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frinta z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinta_2f32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinta z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frinta.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 2 x float> @test_svrinta_2f32_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svrinta_2f32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frinta z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinta_2f32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinta z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frinta.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 2 x float> @test_svrinta_2f32_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svrinta_2f32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: frinta z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinta_2f32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinta z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frinta.nxv2f32(<vscale x 2 x float> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 4 x float> @test_svrinta_f32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svrinta_f32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frinta z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinta_f32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinta z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frinta.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svrinta_f32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svrinta_f32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frinta z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinta_f32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinta z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frinta.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svrinta_f32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svrinta_f32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: frinta z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinta_f32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinta z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frinta.nxv4f32(<vscale x 4 x float> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 2 x double> @test_svrinta_f64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svrinta_f64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frinta z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinta_f64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinta z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frinta.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svrinta_f64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svrinta_f64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frinta z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinta_f64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinta z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frinta.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svrinta_f64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svrinta_f64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: frinta z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinta_f64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinta z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frinta.nxv2f64(<vscale x 2 x double> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 8 x half> @test_svrinti_f16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svrinti_f16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frinti z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinti_f16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinti z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frinti.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svrinti_f16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svrinti_f16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frinti z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinti_f16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinti z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frinti.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svrinti_f16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svrinti_f16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: frinti z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinti_f16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinti z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frinti.nxv8f16(<vscale x 8 x half> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 4 x half> @test_svrinti_4f16_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svrinti_4f16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frinti z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinti_4f16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinti z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frinti.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 4 x half> @test_svrinti_4f16_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svrinti_4f16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frinti z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinti_4f16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinti z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frinti.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 4 x half> @test_svrinti_4f16_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svrinti_4f16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: frinti z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinti_4f16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinti z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frinti.nxv4f16(<vscale x 4 x half> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 2 x half> @test_svrinti_2f16_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svrinti_2f16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frinti z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinti_2f16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinti z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frinti.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x half> @test_svrinti_2f16_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svrinti_2f16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frinti z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinti_2f16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinti z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frinti.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x half> @test_svrinti_2f16_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svrinti_2f16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: frinti z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinti_2f16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinti z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frinti.nxv2f16(<vscale x 2 x half> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x float> @test_svrinti_2f32_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svrinti_2f32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frinti z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinti_2f32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinti z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frinti.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 2 x float> @test_svrinti_2f32_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svrinti_2f32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frinti z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinti_2f32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinti z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frinti.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 2 x float> @test_svrinti_2f32_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svrinti_2f32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: frinti z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinti_2f32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinti z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frinti.nxv2f32(<vscale x 2 x float> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 4 x float> @test_svrinti_f32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svrinti_f32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frinti z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinti_f32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinti z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frinti.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svrinti_f32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svrinti_f32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frinti z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinti_f32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinti z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frinti.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svrinti_f32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svrinti_f32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: frinti z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinti_f32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinti z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frinti.nxv4f32(<vscale x 4 x float> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 2 x double> @test_svrinti_f64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svrinti_f64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frinti z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinti_f64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinti z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frinti.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svrinti_f64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svrinti_f64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frinti z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinti_f64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinti z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frinti.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+
+define <vscale x 2 x double> @test_svrinti_f64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svrinti_f64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: frinti z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrinti_f64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frinti z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frinti.nxv2f64(<vscale x 2 x double> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+
+define <vscale x 8 x half> @test_svrintm_f16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svrintm_f16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frintm z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintm_f16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintm z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintm.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svrintm_f16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svrintm_f16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintm z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintm_f16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintm z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintm.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+
+define <vscale x 8 x half> @test_svrintm_f16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svrintm_f16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: frintm z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintm_f16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintm z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintm.nxv8f16(<vscale x 8 x half> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 4 x half> @test_svrintm_4f16_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svrintm_4f16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frintm z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintm_4f16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintm z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintm.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 4 x half> @test_svrintm_4f16_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svrintm_4f16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintm z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintm_4f16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintm z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintm.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 4 x half> @test_svrintm_4f16_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svrintm_4f16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: frintm z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintm_4f16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintm z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintm.nxv4f16(<vscale x 4 x half> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 2 x half> @test_svrintm_2f16_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svrintm_2f16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frintm z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintm_2f16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintm z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintm.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x half> @test_svrintm_2f16_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svrintm_2f16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintm z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintm_2f16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintm z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintm.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x half> @test_svrintm_2f16_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svrintm_2f16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: frintm z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintm_2f16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintm z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintm.nxv2f16(<vscale x 2 x half> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x float> @test_svrintm_2f32_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svrintm_2f32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frintm z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintm_2f32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintm z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintm.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 2 x float> @test_svrintm_2f32_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svrintm_2f32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintm z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintm_2f32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintm z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintm.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 2 x float> @test_svrintm_2f32_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svrintm_2f32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: frintm z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintm_2f32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintm z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintm.nxv2f32(<vscale x 2 x float> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 4 x float> @test_svrintm_f32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svrintm_f32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frintm z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintm_f32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintm z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintm.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svrintm_f32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svrintm_f32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintm z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintm_f32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintm z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintm.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svrintm_f32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svrintm_f32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: frintm z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintm_f32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintm z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintm.nxv4f32(<vscale x 4 x float> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 2 x double> @test_svrintm_f64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svrintm_f64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frintm z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintm_f64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintm z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintm.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svrintm_f64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svrintm_f64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintm z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintm_f64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintm z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintm.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svrintm_f64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svrintm_f64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: frintm z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintm_f64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintm z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintm.nxv2f64(<vscale x 2 x double> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 8 x half> @test_svrintn_f16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svrintn_f16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frintn z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintn_f16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintn z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintn.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svrintn_f16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svrintn_f16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintn z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintn_f16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintn z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintn.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svrintn_f16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svrintn_f16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: frintn z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintn_f16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintn z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintn.nxv8f16(<vscale x 8 x half> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 4 x half> @test_svrintn_4f16_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svrintn_4f16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frintn z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintn_4f16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintn z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintn.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 4 x half> @test_svrintn_4f16_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svrintn_4f16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintn z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintn_4f16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintn z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintn.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 4 x half> @test_svrintn_4f16_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svrintn_4f16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: frintn z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintn_4f16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintn z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintn.nxv4f16(<vscale x 4 x half> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 2 x half> @test_svrintn_2f16_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svrintn_2f16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frintn z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintn_2f16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintn z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintn.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x half> @test_svrintn_2f16_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svrintn_2f16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintn z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintn_2f16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintn z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintn.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x half> @test_svrintn_2f16_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svrintn_2f16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: frintn z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintn_2f16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintn z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintn.nxv2f16(<vscale x 2 x half> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x float> @test_svrintn_2f32_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svrintn_2f32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frintn z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintn_2f32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintn z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintn.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 2 x float> @test_svrintn_2f32_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svrintn_2f32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintn z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintn_2f32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintn z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintn.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 2 x float> @test_svrintn_2f32_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svrintn_2f32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: frintn z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintn_2f32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintn z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintn.nxv2f32(<vscale x 2 x float> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 4 x float> @test_svrintn_f32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svrintn_f32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frintn z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintn_f32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintn z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintn.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svrintn_f32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svrintn_f32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintn z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintn_f32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintn z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintn.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svrintn_f32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svrintn_f32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: frintn z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintn_f32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintn z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintn.nxv4f32(<vscale x 4 x float> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 2 x double> @test_svrintn_f64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svrintn_f64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frintn z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintn_f64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintn z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintn.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svrintn_f64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svrintn_f64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintn z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintn_f64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintn z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintn.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svrintn_f64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svrintn_f64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: frintn z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintn_f64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintn z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintn.nxv2f64(<vscale x 2 x double> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 8 x half> @test_svrintp_f16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svrintp_f16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frintp z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintp_f16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintp z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintp.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svrintp_f16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svrintp_f16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintp z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintp_f16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintp z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintp.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svrintp_f16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svrintp_f16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: frintp z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintp_f16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintp z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintp.nxv8f16(<vscale x 8 x half> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 4 x half> @test_svrintp_4f16_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svrintp_4f16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frintp z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintp_4f16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintp z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintp.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 4 x half> @test_svrintp_4f16_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svrintp_4f16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintp z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintp_4f16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintp z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintp.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 4 x half> @test_svrintp_4f16_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svrintp_4f16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: frintp z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintp_4f16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintp z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintp.nxv4f16(<vscale x 4 x half> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 2 x half> @test_svrintp_2f16_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svrintp_2f16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frintp z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintp_2f16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintp z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintp.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x half> @test_svrintp_2f16_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svrintp_2f16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintp z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintp_2f16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintp z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintp.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x half> @test_svrintp_2f16_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svrintp_2f16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: frintp z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintp_2f16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintp z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintp.nxv2f16(<vscale x 2 x half> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x float> @test_svrintp_2f32_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svrintp_2f32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frintp z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintp_2f32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintp z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintp.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 2 x float> @test_svrintp_2f32_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svrintp_2f32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintp z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintp_2f32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintp z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintp.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 2 x float> @test_svrintp_2f32_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svrintp_2f32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: frintp z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintp_2f32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintp z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintp.nxv2f32(<vscale x 2 x float> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 4 x float> @test_svrintp_f32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svrintp_f32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frintp z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintp_f32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintp z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintp.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svrintp_f32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svrintp_f32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintp z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintp_f32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintp z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintp.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svrintp_f32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svrintp_f32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: frintp z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintp_f32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintp z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintp.nxv4f32(<vscale x 4 x float> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 2 x double> @test_svrintp_f64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svrintp_f64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frintp z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintp_f64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintp z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintp.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svrintp_f64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svrintp_f64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintp z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintp_f64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintp z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintp.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+
+define <vscale x 2 x double> @test_svrintp_f64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svrintp_f64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: frintp z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintp_f64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintp z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintp.nxv2f64(<vscale x 2 x double> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 8 x half> @test_svrintx_f16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svrintx_f16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frintx z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintx_f16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintx z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintx.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svrintx_f16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svrintx_f16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintx z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintx_f16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintx z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintx.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svrintx_f16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svrintx_f16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: frintx z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintx_f16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintx z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintx.nxv8f16(<vscale x 8 x half> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 4 x half> @test_svrintx_4f16_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svrintx_4f16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frintx z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintx_4f16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintx z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintx.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 4 x half> @test_svrintx_4f16_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svrintx_4f16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintx z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintx_4f16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintx z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintx.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 4 x half> @test_svrintx_4f16_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svrintx_4f16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: frintx z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintx_4f16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintx z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintx.nxv4f16(<vscale x 4 x half> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 2 x half> @test_svrintx_2f16_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svrintx_2f16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frintx z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintx_2f16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintx z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintx.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x half> @test_svrintx_2f16_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svrintx_2f16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintx z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintx_2f16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintx z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintx.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x half> @test_svrintx_2f16_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svrintx_2f16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: frintx z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintx_2f16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintx z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintx.nxv2f16(<vscale x 2 x half> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x float> @test_svrintx_2f32_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svrintx_2f32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frintx z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintx_2f32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintx z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintx.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 2 x float> @test_svrintx_2f32_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svrintx_2f32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintx z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintx_2f32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintx z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintx.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 2 x float> @test_svrintx_2f32_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svrintx_2f32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: frintx z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintx_2f32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintx z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintx.nxv2f32(<vscale x 2 x float> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 4 x float> @test_svrintx_f32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svrintx_f32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frintx z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintx_f32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintx z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintx.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svrintx_f32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svrintx_f32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintx z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintx_f32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintx z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintx.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svrintx_f32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svrintx_f32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: frintx z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintx_f32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintx z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintx.nxv4f32(<vscale x 4 x float> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 2 x double> @test_svrintx_f64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svrintx_f64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frintx z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintx_f64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintx z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintx.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svrintx_f64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svrintx_f64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintx z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintx_f64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintx z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintx.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svrintx_f64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svrintx_f64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: frintx z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintx_f64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintx z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintx.nxv2f64(<vscale x 2 x double> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 8 x half> @test_svrintz_f16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svrintz_f16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frintz z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintz_f16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintz z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintz.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svrintz_f16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svrintz_f16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintz z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintz_f16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintz z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintz.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svrintz_f16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svrintz_f16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: frintz z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintz_f16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintz z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintz.nxv8f16(<vscale x 8 x half> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 4 x half> @test_svrintz_4f16_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svrintz_4f16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frintz z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintz_4f16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintz z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintz.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 4 x half> @test_svrintz_4f16_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svrintz_4f16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintz z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintz_4f16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintz z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintz.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 4 x half> @test_svrintz_4f16_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svrintz_4f16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: frintz z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintz_4f16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintz z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintz.nxv4f16(<vscale x 4 x half> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 2 x half> @test_svrintz_2f16_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svrintz_2f16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frintz z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintz_2f16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintz z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintz.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x half> @test_svrintz_2f16_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svrintz_2f16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintz z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintz_2f16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintz z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintz.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x half> @test_svrintz_2f16_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svrintz_2f16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: frintz z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintz_2f16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintz z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintz.nxv2f16(<vscale x 2 x half> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x float> @test_svrintz_2f32_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svrintz_2f32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frintz z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintz_2f32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintz z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintz.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 2 x float> @test_svrintz_2f32_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svrintz_2f32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintz z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintz_2f32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintz z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintz.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 2 x float> @test_svrintz_2f32_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svrintz_2f32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: frintz z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintz_2f32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintz z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintz.nxv2f32(<vscale x 2 x float> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 4 x float> @test_svrintz_f32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svrintz_f32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frintz z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintz_f32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintz z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintz.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svrintz_f32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svrintz_f32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintz z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintz_f32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintz z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintz.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svrintz_f32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svrintz_f32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: frintz z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintz_f32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintz z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintz.nxv4f32(<vscale x 4 x float> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 2 x double> @test_svrintz_f64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svrintz_f64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frintz z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintz_f64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintz z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintz.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svrintz_f64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svrintz_f64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintz z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintz_f64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintz z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintz.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svrintz_f64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svrintz_f64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: frintz z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrintz_f64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frintz z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintz.nxv2f64(<vscale x 2 x double> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 8 x half> @test_svrecpx_f16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svrecpx_f16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frecpx z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrecpx_f16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frecpx z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frecpx.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svrecpx_f16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svrecpx_f16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frecpx z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrecpx_f16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frecpx z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frecpx.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svrecpx_f16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svrecpx_f16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: frecpx z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrecpx_f16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frecpx z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frecpx.nxv8f16(<vscale x 8 x half> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 4 x half> @test_svrecpx_4f16_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svrecpx_4f16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frecpx z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrecpx_4f16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frecpx z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frecpx.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 4 x half> @test_svrecpx_4f16_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svrecpx_4f16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frecpx z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrecpx_4f16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frecpx z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frecpx.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 4 x half> @test_svrecpx_4f16_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svrecpx_4f16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: frecpx z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrecpx_4f16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frecpx z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frecpx.nxv4f16(<vscale x 4 x half> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 2 x half> @test_svrecpx_2f16_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svrecpx_2f16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frecpx z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrecpx_2f16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frecpx z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frecpx.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x half> @test_svrecpx_2f16_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svrecpx_2f16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frecpx z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrecpx_2f16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frecpx z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frecpx.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x half> @test_svrecpx_2f16_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svrecpx_2f16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: frecpx z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrecpx_2f16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frecpx z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frecpx.nxv2f16(<vscale x 2 x half> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x float> @test_svrecpx_2f32_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svrecpx_2f32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frecpx z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrecpx_2f32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frecpx z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frecpx.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 2 x float> @test_svrecpx_2f32_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svrecpx_2f32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frecpx z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrecpx_2f32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frecpx z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frecpx.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 2 x float> @test_svrecpx_2f32_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svrecpx_2f32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: frecpx z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrecpx_2f32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frecpx z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frecpx.nxv2f32(<vscale x 2 x float> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 4 x float> @test_svrecpx_f32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svrecpx_f32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frecpx z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrecpx_f32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frecpx z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frecpx.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svrecpx_f32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svrecpx_f32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frecpx z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrecpx_f32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frecpx z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frecpx.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svrecpx_f32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svrecpx_f32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: frecpx z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrecpx_f32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frecpx z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frecpx.nxv4f32(<vscale x 4 x float> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 2 x double> @test_svrecpx_f64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svrecpx_f64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: frecpx z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrecpx_f64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frecpx z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frecpx.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svrecpx_f64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svrecpx_f64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frecpx z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrecpx_f64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frecpx z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frecpx.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svrecpx_f64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svrecpx_f64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: frecpx z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrecpx_f64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: frecpx z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frecpx.nxv2f64(<vscale x 2 x double> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 8 x half> @test_svsqrt_f16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svsqrt_f16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: fsqrt z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqrt_f16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: fsqrt z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fsqrt.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svsqrt_f16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svsqrt_f16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: fsqrt z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqrt_f16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: fsqrt z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fsqrt.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svsqrt_f16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svsqrt_f16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: fsqrt z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqrt_f16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: fsqrt z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fsqrt.nxv8f16(<vscale x 8 x half> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 4 x half> @test_svsqrt_4f16_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svsqrt_4f16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: fsqrt z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqrt_4f16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: fsqrt z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.fsqrt.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 4 x half> @test_svsqrt_4f16_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svsqrt_4f16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: fsqrt z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqrt_4f16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: fsqrt z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.fsqrt.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 4 x half> @test_svsqrt_4f16_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svsqrt_4f16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: fsqrt z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqrt_4f16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: fsqrt z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.fsqrt.nxv4f16(<vscale x 4 x half> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 2 x half> @test_svsqrt_2f16_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svsqrt_2f16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: fsqrt z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqrt_2f16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: fsqrt z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.fsqrt.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x half> @test_svsqrt_2f16_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svsqrt_2f16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: fsqrt z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqrt_2f16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: fsqrt z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.fsqrt.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x half> @test_svsqrt_2f16_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svsqrt_2f16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: fsqrt z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqrt_2f16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: fsqrt z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.fsqrt.nxv2f16(<vscale x 2 x half> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x float> @test_svsqrt_2f32_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svsqrt_2f32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: fsqrt z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqrt_2f32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: fsqrt z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.fsqrt.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 2 x float> @test_svsqrt_2f32_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svsqrt_2f32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: fsqrt z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqrt_2f32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: fsqrt z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.fsqrt.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 2 x float> @test_svsqrt_2f32_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svsqrt_2f32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: fsqrt z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqrt_2f32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: fsqrt z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.fsqrt.nxv2f32(<vscale x 2 x float> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 4 x float> @test_svsqrt_f32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svsqrt_f32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: fsqrt z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqrt_f32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: fsqrt z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fsqrt.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svsqrt_f32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svsqrt_f32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: fsqrt z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqrt_f32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: fsqrt z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fsqrt.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svsqrt_f32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svsqrt_f32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: fsqrt z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqrt_f32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: fsqrt z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fsqrt.nxv4f32(<vscale x 4 x float> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 2 x double> @test_svsqrt_f64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svsqrt_f64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: fsqrt z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqrt_f64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: fsqrt z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fsqrt.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svsqrt_f64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svsqrt_f64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: fsqrt z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqrt_f64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: fsqrt z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fsqrt.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svsqrt_f64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svsqrt_f64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: fsqrt z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqrt_f64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: fsqrt z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fsqrt.nxv2f64(<vscale x 2 x double> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
>From 933ea3dae3f14528463fa91b47fcaec0162a2a94 Mon Sep 17 00:00:00 2001
From: Momchil Velikov <momchil.velikov at arm.com>
Date: Fri, 20 Dec 2024 17:53:46 +0000
Subject: [PATCH 10/16] [fixup] Rebase, update tests, add tests for the new
all-true patterns, change undef to poison
---
llvm/lib/Target/AArch64/SVEInstrFormats.td | 12 +-
.../zeroing-forms-frint-frecpx-fsqrt.ll | 2322 ++++++++++++++++-
2 files changed, 2220 insertions(+), 114 deletions(-)
diff --git a/llvm/lib/Target/AArch64/SVEInstrFormats.td b/llvm/lib/Target/AArch64/SVEInstrFormats.td
index 87c882817e9117..99994a50150183 100644
--- a/llvm/lib/Target/AArch64/SVEInstrFormats.td
+++ b/llvm/lib/Target/AArch64/SVEInstrFormats.td
@@ -3292,12 +3292,12 @@ multiclass sve_fp_z2op_p_zd_hsd<bits<5> opc, string asm, SDPatternOperator op> {
def _S : sve_fp_z2op_p_zd<{ 0b10, opc }, asm, ZPR32, ZPR32>;
def _D : sve_fp_z2op_p_zd<{ 0b11, opc }, asm, ZPR64, ZPR64>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv8f16, op, nxv8i1, nxv8f16, !cast<Instruction>(NAME # _H)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv4f16, op, nxv4i1, nxv4f16, !cast<Instruction>(NAME # _H)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv2f16, op, nxv2i1, nxv2f16, !cast<Instruction>(NAME # _H)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv4f32, op, nxv4i1, nxv4f32, !cast<Instruction>(NAME # _S)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv2f32, op, nxv2i1, nxv2f32, !cast<Instruction>(NAME # _S)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv2f64, op, nxv2i1, nxv2f64, !cast<Instruction>(NAME # _D)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv8f16, op, nxv8i1, nxv8f16, !cast<Instruction>(NAME # _H)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv4f16, op, nxv4i1, nxv4f16, !cast<Instruction>(NAME # _H)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv2f16, op, nxv2i1, nxv2f16, !cast<Instruction>(NAME # _H)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv4f32, op, nxv4i1, nxv4f32, !cast<Instruction>(NAME # _S)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv2f32, op, nxv2i1, nxv2f32, !cast<Instruction>(NAME # _S)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv2f64, op, nxv2i1, nxv2f64, !cast<Instruction>(NAME # _D)>;
}
multiclass sve_fp_z2op_p_zd_frint<bits<2> opc, string asm> {
diff --git a/llvm/test/CodeGen/AArch64/zeroing-forms-frint-frecpx-fsqrt.ll b/llvm/test/CodeGen/AArch64/zeroing-forms-frint-frecpx-fsqrt.ll
index 8de7f3db29674c..c493ec2dcc95d8 100644
--- a/llvm/test/CodeGen/AArch64/zeroing-forms-frint-frecpx-fsqrt.ll
+++ b/llvm/test/CodeGen/AArch64/zeroing-forms-frint-frecpx-fsqrt.ll
@@ -18,7 +18,7 @@ define <vscale x 8 x half> @test_svrinta_f16_x_1(<vscale x 8 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frinta z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frinta.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frinta.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x half> %0
}
@@ -34,7 +34,7 @@ define <vscale x 8 x half> @test_svrinta_f16_x_2(<vscale x 8 x i1> %pg, double %
; CHECK-2p2-NEXT: frinta z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frinta.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frinta.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x half> %0
}
@@ -65,7 +65,7 @@ define <vscale x 4 x half> @test_svrinta_4f16_x_1(<vscale x 4 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frinta z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frinta.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frinta.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
ret <vscale x 4 x half> %0
}
@@ -81,7 +81,7 @@ define <vscale x 4 x half> @test_svrinta_4f16_x_2(<vscale x 4 x i1> %pg, double
; CHECK-2p2-NEXT: frinta z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frinta.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frinta.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
ret <vscale x 4 x half> %0
}
@@ -112,7 +112,7 @@ define <vscale x 2 x half> @test_svrinta_2f16_x_1(<vscale x 2 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frinta z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frinta.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frinta.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
ret <vscale x 2 x half> %0
}
@@ -128,7 +128,7 @@ define <vscale x 2 x half> @test_svrinta_2f16_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: frinta z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frinta.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frinta.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
ret <vscale x 2 x half> %0
}
@@ -159,7 +159,7 @@ define <vscale x 2 x float> @test_svrinta_2f32_x_1(<vscale x 2 x i1> %pg, <vscal
; CHECK-2p2-NEXT: frinta z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frinta.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frinta.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
ret <vscale x 2 x float> %0
}
@@ -175,7 +175,7 @@ define <vscale x 2 x float> @test_svrinta_2f32_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: frinta z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frinta.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frinta.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
ret <vscale x 2 x float> %0
}
@@ -206,7 +206,7 @@ define <vscale x 4 x float> @test_svrinta_f32_x_1(<vscale x 4 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frinta z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frinta.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frinta.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x float> %0
}
@@ -222,7 +222,7 @@ define <vscale x 4 x float> @test_svrinta_f32_x_2(<vscale x 4 x i1> %pg, double
; CHECK-2p2-NEXT: frinta z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frinta.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frinta.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x float> %0
}
@@ -253,7 +253,7 @@ define <vscale x 2 x double> @test_svrinta_f64_x_1(<vscale x 2 x i1> %pg, <vscal
; CHECK-2p2-NEXT: frinta z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frinta.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frinta.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x double> %0
}
@@ -269,7 +269,7 @@ define <vscale x 2 x double> @test_svrinta_f64_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: frinta z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frinta.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frinta.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x double> %0
}
@@ -300,7 +300,7 @@ define <vscale x 8 x half> @test_svrinti_f16_x_1(<vscale x 8 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frinti z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frinti.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frinti.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x half> %0
}
@@ -316,7 +316,7 @@ define <vscale x 8 x half> @test_svrinti_f16_x_2(<vscale x 8 x i1> %pg, double %
; CHECK-2p2-NEXT: frinti z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frinti.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frinti.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x half> %0
}
@@ -347,7 +347,7 @@ define <vscale x 4 x half> @test_svrinti_4f16_x_1(<vscale x 4 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frinti z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frinti.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frinti.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
ret <vscale x 4 x half> %0
}
@@ -363,7 +363,7 @@ define <vscale x 4 x half> @test_svrinti_4f16_x_2(<vscale x 4 x i1> %pg, double
; CHECK-2p2-NEXT: frinti z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frinti.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frinti.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
ret <vscale x 4 x half> %0
}
@@ -394,7 +394,7 @@ define <vscale x 2 x half> @test_svrinti_2f16_x_1(<vscale x 2 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frinti z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frinti.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frinti.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
ret <vscale x 2 x half> %0
}
@@ -410,7 +410,7 @@ define <vscale x 2 x half> @test_svrinti_2f16_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: frinti z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frinti.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frinti.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
ret <vscale x 2 x half> %0
}
@@ -441,7 +441,7 @@ define <vscale x 2 x float> @test_svrinti_2f32_x_1(<vscale x 2 x i1> %pg, <vscal
; CHECK-2p2-NEXT: frinti z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frinti.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frinti.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
ret <vscale x 2 x float> %0
}
@@ -457,7 +457,7 @@ define <vscale x 2 x float> @test_svrinti_2f32_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: frinti z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frinti.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frinti.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
ret <vscale x 2 x float> %0
}
@@ -488,7 +488,7 @@ define <vscale x 4 x float> @test_svrinti_f32_x_1(<vscale x 4 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frinti z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frinti.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frinti.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x float> %0
}
@@ -504,7 +504,7 @@ define <vscale x 4 x float> @test_svrinti_f32_x_2(<vscale x 4 x i1> %pg, double
; CHECK-2p2-NEXT: frinti z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frinti.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frinti.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x float> %0
}
@@ -535,7 +535,7 @@ define <vscale x 2 x double> @test_svrinti_f64_x_1(<vscale x 2 x i1> %pg, <vscal
; CHECK-2p2-NEXT: frinti z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frinti.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frinti.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x double> %0
}
@@ -551,7 +551,7 @@ define <vscale x 2 x double> @test_svrinti_f64_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: frinti z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frinti.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frinti.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x double> %0
}
@@ -584,7 +584,7 @@ define <vscale x 8 x half> @test_svrintm_f16_x_1(<vscale x 8 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frintm z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintm.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintm.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x half> %0
}
@@ -600,7 +600,7 @@ define <vscale x 8 x half> @test_svrintm_f16_x_2(<vscale x 8 x i1> %pg, double %
; CHECK-2p2-NEXT: frintm z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintm.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintm.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x half> %0
}
@@ -632,7 +632,7 @@ define <vscale x 4 x half> @test_svrintm_4f16_x_1(<vscale x 4 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frintm z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintm.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintm.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
ret <vscale x 4 x half> %0
}
@@ -648,7 +648,7 @@ define <vscale x 4 x half> @test_svrintm_4f16_x_2(<vscale x 4 x i1> %pg, double
; CHECK-2p2-NEXT: frintm z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintm.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintm.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
ret <vscale x 4 x half> %0
}
@@ -679,7 +679,7 @@ define <vscale x 2 x half> @test_svrintm_2f16_x_1(<vscale x 2 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frintm z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintm.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintm.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
ret <vscale x 2 x half> %0
}
@@ -695,7 +695,7 @@ define <vscale x 2 x half> @test_svrintm_2f16_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: frintm z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintm.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintm.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
ret <vscale x 2 x half> %0
}
@@ -726,7 +726,7 @@ define <vscale x 2 x float> @test_svrintm_2f32_x_1(<vscale x 2 x i1> %pg, <vscal
; CHECK-2p2-NEXT: frintm z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintm.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintm.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
ret <vscale x 2 x float> %0
}
@@ -742,7 +742,7 @@ define <vscale x 2 x float> @test_svrintm_2f32_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: frintm z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintm.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintm.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
ret <vscale x 2 x float> %0
}
@@ -773,7 +773,7 @@ define <vscale x 4 x float> @test_svrintm_f32_x_1(<vscale x 4 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frintm z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintm.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintm.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x float> %0
}
@@ -789,7 +789,7 @@ define <vscale x 4 x float> @test_svrintm_f32_x_2(<vscale x 4 x i1> %pg, double
; CHECK-2p2-NEXT: frintm z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintm.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintm.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x float> %0
}
@@ -820,7 +820,7 @@ define <vscale x 2 x double> @test_svrintm_f64_x_1(<vscale x 2 x i1> %pg, <vscal
; CHECK-2p2-NEXT: frintm z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintm.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintm.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x double> %0
}
@@ -836,7 +836,7 @@ define <vscale x 2 x double> @test_svrintm_f64_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: frintm z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintm.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintm.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x double> %0
}
@@ -867,7 +867,7 @@ define <vscale x 8 x half> @test_svrintn_f16_x_1(<vscale x 8 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frintn z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintn.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintn.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x half> %0
}
@@ -883,7 +883,7 @@ define <vscale x 8 x half> @test_svrintn_f16_x_2(<vscale x 8 x i1> %pg, double %
; CHECK-2p2-NEXT: frintn z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintn.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintn.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x half> %0
}
@@ -914,7 +914,7 @@ define <vscale x 4 x half> @test_svrintn_4f16_x_1(<vscale x 4 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frintn z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintn.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintn.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
ret <vscale x 4 x half> %0
}
@@ -930,7 +930,7 @@ define <vscale x 4 x half> @test_svrintn_4f16_x_2(<vscale x 4 x i1> %pg, double
; CHECK-2p2-NEXT: frintn z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintn.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintn.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
ret <vscale x 4 x half> %0
}
@@ -961,7 +961,7 @@ define <vscale x 2 x half> @test_svrintn_2f16_x_1(<vscale x 2 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frintn z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintn.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintn.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
ret <vscale x 2 x half> %0
}
@@ -977,7 +977,7 @@ define <vscale x 2 x half> @test_svrintn_2f16_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: frintn z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintn.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintn.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
ret <vscale x 2 x half> %0
}
@@ -1008,7 +1008,7 @@ define <vscale x 2 x float> @test_svrintn_2f32_x_1(<vscale x 2 x i1> %pg, <vscal
; CHECK-2p2-NEXT: frintn z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintn.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintn.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
ret <vscale x 2 x float> %0
}
@@ -1024,7 +1024,7 @@ define <vscale x 2 x float> @test_svrintn_2f32_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: frintn z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintn.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintn.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
ret <vscale x 2 x float> %0
}
@@ -1055,7 +1055,7 @@ define <vscale x 4 x float> @test_svrintn_f32_x_1(<vscale x 4 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frintn z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintn.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintn.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x float> %0
}
@@ -1071,7 +1071,7 @@ define <vscale x 4 x float> @test_svrintn_f32_x_2(<vscale x 4 x i1> %pg, double
; CHECK-2p2-NEXT: frintn z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintn.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintn.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x float> %0
}
@@ -1102,7 +1102,7 @@ define <vscale x 2 x double> @test_svrintn_f64_x_1(<vscale x 2 x i1> %pg, <vscal
; CHECK-2p2-NEXT: frintn z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintn.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintn.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x double> %0
}
@@ -1118,7 +1118,7 @@ define <vscale x 2 x double> @test_svrintn_f64_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: frintn z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintn.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintn.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x double> %0
}
@@ -1149,7 +1149,7 @@ define <vscale x 8 x half> @test_svrintp_f16_x_1(<vscale x 8 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frintp z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintp.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintp.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x half> %0
}
@@ -1165,7 +1165,7 @@ define <vscale x 8 x half> @test_svrintp_f16_x_2(<vscale x 8 x i1> %pg, double %
; CHECK-2p2-NEXT: frintp z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintp.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintp.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x half> %0
}
@@ -1196,7 +1196,7 @@ define <vscale x 4 x half> @test_svrintp_4f16_x_1(<vscale x 4 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frintp z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintp.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintp.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
ret <vscale x 4 x half> %0
}
@@ -1212,7 +1212,7 @@ define <vscale x 4 x half> @test_svrintp_4f16_x_2(<vscale x 4 x i1> %pg, double
; CHECK-2p2-NEXT: frintp z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintp.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintp.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
ret <vscale x 4 x half> %0
}
@@ -1243,7 +1243,7 @@ define <vscale x 2 x half> @test_svrintp_2f16_x_1(<vscale x 2 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frintp z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintp.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintp.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
ret <vscale x 2 x half> %0
}
@@ -1259,7 +1259,7 @@ define <vscale x 2 x half> @test_svrintp_2f16_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: frintp z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintp.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintp.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
ret <vscale x 2 x half> %0
}
@@ -1290,7 +1290,7 @@ define <vscale x 2 x float> @test_svrintp_2f32_x_1(<vscale x 2 x i1> %pg, <vscal
; CHECK-2p2-NEXT: frintp z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintp.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintp.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
ret <vscale x 2 x float> %0
}
@@ -1306,7 +1306,7 @@ define <vscale x 2 x float> @test_svrintp_2f32_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: frintp z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintp.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintp.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
ret <vscale x 2 x float> %0
}
@@ -1337,7 +1337,7 @@ define <vscale x 4 x float> @test_svrintp_f32_x_1(<vscale x 4 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frintp z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintp.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintp.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x float> %0
}
@@ -1353,7 +1353,7 @@ define <vscale x 4 x float> @test_svrintp_f32_x_2(<vscale x 4 x i1> %pg, double
; CHECK-2p2-NEXT: frintp z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintp.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintp.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x float> %0
}
@@ -1384,7 +1384,7 @@ define <vscale x 2 x double> @test_svrintp_f64_x_1(<vscale x 2 x i1> %pg, <vscal
; CHECK-2p2-NEXT: frintp z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintp.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintp.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x double> %0
}
@@ -1400,7 +1400,7 @@ define <vscale x 2 x double> @test_svrintp_f64_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: frintp z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintp.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintp.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x double> %0
}
@@ -1432,7 +1432,7 @@ define <vscale x 8 x half> @test_svrintx_f16_x_1(<vscale x 8 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frintx z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintx.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintx.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x half> %0
}
@@ -1448,7 +1448,7 @@ define <vscale x 8 x half> @test_svrintx_f16_x_2(<vscale x 8 x i1> %pg, double %
; CHECK-2p2-NEXT: frintx z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintx.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintx.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x half> %0
}
@@ -1479,7 +1479,7 @@ define <vscale x 4 x half> @test_svrintx_4f16_x_1(<vscale x 4 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frintx z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintx.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintx.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
ret <vscale x 4 x half> %0
}
@@ -1495,7 +1495,7 @@ define <vscale x 4 x half> @test_svrintx_4f16_x_2(<vscale x 4 x i1> %pg, double
; CHECK-2p2-NEXT: frintx z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintx.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintx.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
ret <vscale x 4 x half> %0
}
@@ -1526,7 +1526,7 @@ define <vscale x 2 x half> @test_svrintx_2f16_x_1(<vscale x 2 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frintx z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintx.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintx.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
ret <vscale x 2 x half> %0
}
@@ -1542,7 +1542,7 @@ define <vscale x 2 x half> @test_svrintx_2f16_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: frintx z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintx.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintx.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
ret <vscale x 2 x half> %0
}
@@ -1573,7 +1573,7 @@ define <vscale x 2 x float> @test_svrintx_2f32_x_1(<vscale x 2 x i1> %pg, <vscal
; CHECK-2p2-NEXT: frintx z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintx.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintx.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
ret <vscale x 2 x float> %0
}
@@ -1589,7 +1589,7 @@ define <vscale x 2 x float> @test_svrintx_2f32_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: frintx z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintx.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintx.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
ret <vscale x 2 x float> %0
}
@@ -1620,7 +1620,7 @@ define <vscale x 4 x float> @test_svrintx_f32_x_1(<vscale x 4 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frintx z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintx.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintx.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x float> %0
}
@@ -1636,7 +1636,7 @@ define <vscale x 4 x float> @test_svrintx_f32_x_2(<vscale x 4 x i1> %pg, double
; CHECK-2p2-NEXT: frintx z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintx.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintx.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x float> %0
}
@@ -1667,7 +1667,7 @@ define <vscale x 2 x double> @test_svrintx_f64_x_1(<vscale x 2 x i1> %pg, <vscal
; CHECK-2p2-NEXT: frintx z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintx.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintx.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x double> %0
}
@@ -1683,7 +1683,7 @@ define <vscale x 2 x double> @test_svrintx_f64_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: frintx z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintx.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintx.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x double> %0
}
@@ -1714,7 +1714,7 @@ define <vscale x 8 x half> @test_svrintz_f16_x_1(<vscale x 8 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frintz z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintz.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintz.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x half> %0
}
@@ -1730,7 +1730,7 @@ define <vscale x 8 x half> @test_svrintz_f16_x_2(<vscale x 8 x i1> %pg, double %
; CHECK-2p2-NEXT: frintz z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintz.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintz.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x half> %0
}
@@ -1761,7 +1761,7 @@ define <vscale x 4 x half> @test_svrintz_4f16_x_1(<vscale x 4 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frintz z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintz.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintz.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
ret <vscale x 4 x half> %0
}
@@ -1777,7 +1777,7 @@ define <vscale x 4 x half> @test_svrintz_4f16_x_2(<vscale x 4 x i1> %pg, double
; CHECK-2p2-NEXT: frintz z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintz.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintz.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
ret <vscale x 4 x half> %0
}
@@ -1808,7 +1808,7 @@ define <vscale x 2 x half> @test_svrintz_2f16_x_1(<vscale x 2 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frintz z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintz.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintz.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
ret <vscale x 2 x half> %0
}
@@ -1824,7 +1824,7 @@ define <vscale x 2 x half> @test_svrintz_2f16_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: frintz z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintz.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintz.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
ret <vscale x 2 x half> %0
}
@@ -1855,7 +1855,7 @@ define <vscale x 2 x float> @test_svrintz_2f32_x_1(<vscale x 2 x i1> %pg, <vscal
; CHECK-2p2-NEXT: frintz z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintz.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintz.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
ret <vscale x 2 x float> %0
}
@@ -1871,7 +1871,7 @@ define <vscale x 2 x float> @test_svrintz_2f32_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: frintz z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintz.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintz.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
ret <vscale x 2 x float> %0
}
@@ -1902,7 +1902,7 @@ define <vscale x 4 x float> @test_svrintz_f32_x_1(<vscale x 4 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frintz z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintz.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintz.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x float> %0
}
@@ -1918,7 +1918,7 @@ define <vscale x 4 x float> @test_svrintz_f32_x_2(<vscale x 4 x i1> %pg, double
; CHECK-2p2-NEXT: frintz z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintz.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintz.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x float> %0
}
@@ -1949,7 +1949,7 @@ define <vscale x 2 x double> @test_svrintz_f64_x_1(<vscale x 2 x i1> %pg, <vscal
; CHECK-2p2-NEXT: frintz z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintz.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintz.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x double> %0
}
@@ -1965,7 +1965,7 @@ define <vscale x 2 x double> @test_svrintz_f64_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: frintz z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintz.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintz.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x double> %0
}
@@ -1996,7 +1996,7 @@ define <vscale x 8 x half> @test_svrecpx_f16_x_1(<vscale x 8 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frecpx z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frecpx.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frecpx.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x half> %0
}
@@ -2012,7 +2012,7 @@ define <vscale x 8 x half> @test_svrecpx_f16_x_2(<vscale x 8 x i1> %pg, double %
; CHECK-2p2-NEXT: frecpx z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frecpx.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frecpx.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x half> %0
}
@@ -2043,7 +2043,7 @@ define <vscale x 4 x half> @test_svrecpx_4f16_x_1(<vscale x 4 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frecpx z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frecpx.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frecpx.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
ret <vscale x 4 x half> %0
}
@@ -2059,7 +2059,7 @@ define <vscale x 4 x half> @test_svrecpx_4f16_x_2(<vscale x 4 x i1> %pg, double
; CHECK-2p2-NEXT: frecpx z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frecpx.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frecpx.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
ret <vscale x 4 x half> %0
}
@@ -2090,7 +2090,7 @@ define <vscale x 2 x half> @test_svrecpx_2f16_x_1(<vscale x 2 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frecpx z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frecpx.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frecpx.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
ret <vscale x 2 x half> %0
}
@@ -2106,7 +2106,7 @@ define <vscale x 2 x half> @test_svrecpx_2f16_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: frecpx z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frecpx.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frecpx.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
ret <vscale x 2 x half> %0
}
@@ -2137,7 +2137,7 @@ define <vscale x 2 x float> @test_svrecpx_2f32_x_1(<vscale x 2 x i1> %pg, <vscal
; CHECK-2p2-NEXT: frecpx z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frecpx.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frecpx.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
ret <vscale x 2 x float> %0
}
@@ -2153,7 +2153,7 @@ define <vscale x 2 x float> @test_svrecpx_2f32_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: frecpx z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frecpx.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frecpx.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
ret <vscale x 2 x float> %0
}
@@ -2184,7 +2184,7 @@ define <vscale x 4 x float> @test_svrecpx_f32_x_1(<vscale x 4 x i1> %pg, <vscale
; CHECK-2p2-NEXT: frecpx z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frecpx.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frecpx.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x float> %0
}
@@ -2200,7 +2200,7 @@ define <vscale x 4 x float> @test_svrecpx_f32_x_2(<vscale x 4 x i1> %pg, double
; CHECK-2p2-NEXT: frecpx z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frecpx.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frecpx.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x float> %0
}
@@ -2231,7 +2231,7 @@ define <vscale x 2 x double> @test_svrecpx_f64_x_1(<vscale x 2 x i1> %pg, <vscal
; CHECK-2p2-NEXT: frecpx z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frecpx.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frecpx.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x double> %0
}
@@ -2247,7 +2247,7 @@ define <vscale x 2 x double> @test_svrecpx_f64_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: frecpx z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frecpx.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frecpx.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x double> %0
}
@@ -2278,7 +2278,7 @@ define <vscale x 8 x half> @test_svsqrt_f16_x_1(<vscale x 8 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: fsqrt z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fsqrt.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fsqrt.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x half> %0
}
@@ -2294,7 +2294,7 @@ define <vscale x 8 x half> @test_svsqrt_f16_x_2(<vscale x 8 x i1> %pg, double %z
; CHECK-2p2-NEXT: fsqrt z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fsqrt.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fsqrt.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x half> %0
}
@@ -2325,7 +2325,7 @@ define <vscale x 4 x half> @test_svsqrt_4f16_x_1(<vscale x 4 x i1> %pg, <vscale
; CHECK-2p2-NEXT: fsqrt z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.fsqrt.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.fsqrt.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
ret <vscale x 4 x half> %0
}
@@ -2341,7 +2341,7 @@ define <vscale x 4 x half> @test_svsqrt_4f16_x_2(<vscale x 4 x i1> %pg, double %
; CHECK-2p2-NEXT: fsqrt z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.fsqrt.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.fsqrt.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
ret <vscale x 4 x half> %0
}
@@ -2372,7 +2372,7 @@ define <vscale x 2 x half> @test_svsqrt_2f16_x_1(<vscale x 2 x i1> %pg, <vscale
; CHECK-2p2-NEXT: fsqrt z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.fsqrt.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.fsqrt.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
ret <vscale x 2 x half> %0
}
@@ -2388,7 +2388,7 @@ define <vscale x 2 x half> @test_svsqrt_2f16_x_2(<vscale x 2 x i1> %pg, double %
; CHECK-2p2-NEXT: fsqrt z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.fsqrt.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.fsqrt.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
ret <vscale x 2 x half> %0
}
@@ -2419,7 +2419,7 @@ define <vscale x 2 x float> @test_svsqrt_2f32_x_1(<vscale x 2 x i1> %pg, <vscale
; CHECK-2p2-NEXT: fsqrt z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.fsqrt.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.fsqrt.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
ret <vscale x 2 x float> %0
}
@@ -2435,7 +2435,7 @@ define <vscale x 2 x float> @test_svsqrt_2f32_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: fsqrt z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.fsqrt.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.fsqrt.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
ret <vscale x 2 x float> %0
}
@@ -2466,7 +2466,7 @@ define <vscale x 4 x float> @test_svsqrt_f32_x_1(<vscale x 4 x i1> %pg, <vscale
; CHECK-2p2-NEXT: fsqrt z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fsqrt.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fsqrt.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x float> %0
}
@@ -2482,7 +2482,7 @@ define <vscale x 4 x float> @test_svsqrt_f32_x_2(<vscale x 4 x i1> %pg, double %
; CHECK-2p2-NEXT: fsqrt z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fsqrt.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fsqrt.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x float> %0
}
@@ -2513,7 +2513,7 @@ define <vscale x 2 x double> @test_svsqrt_f64_x_1(<vscale x 2 x i1> %pg, <vscale
; CHECK-2p2-NEXT: fsqrt z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fsqrt.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fsqrt.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x double> %0
}
@@ -2529,7 +2529,7 @@ define <vscale x 2 x double> @test_svsqrt_f64_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: fsqrt z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fsqrt.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fsqrt.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x double> %0
}
@@ -2548,3 +2548,2109 @@ entry:
%0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fsqrt.nxv2f64(<vscale x 2 x double> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x double> %0
}
+
+define <vscale x 8 x half> @test_svfrinta_nxv8f16_ptrue_u(double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svfrinta_nxv8f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frinta z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrinta_nxv8f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: frinta z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frinta.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svfrinta_nxv8f16_ptrue(double %z0, <vscale x 8 x half> %x, <vscale x 8 x half> %y) {
+; CHECK-LABEL: test_svfrinta_nxv8f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frinta z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrinta_nxv8f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: frinta z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frinta.nxv8f16(<vscale x 8 x half> %x, <vscale x 8 x i1> %pg, <vscale x 8 x half> %y)
+ ret <vscale x 8 x half> %0
+}
+
+
+define <vscale x 4 x half> @test_svfrinta_nxv4f16_ptrue_u(double %z0, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svfrinta_nxv4f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frinta z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrinta_nxv4f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frinta z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frinta.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 4 x half> @test_svfrinta_nxv4f16_ptrue(double %z0, <vscale x 4 x half> %x, <vscale x 4 x half> %y) {
+; CHECK-LABEL: test_svfrinta_nxv4f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frinta z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrinta_nxv4f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frinta z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frinta.nxv4f16(<vscale x 4 x half> %x, <vscale x 4 x i1> %pg, <vscale x 4 x half> %y)
+ ret <vscale x 4 x half> %0
+}
+
+
+define <vscale x 2 x half> @test_svfrinta_nxv2f16_ptrue_u(double %z0, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svfrinta_nxv2f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frinta z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrinta_nxv2f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frinta z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frinta.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x half> @test_svfrinta_nxv2f16_ptrue(double %z0, <vscale x 2 x half> %x, <vscale x 2 x half> %y) {
+; CHECK-LABEL: test_svfrinta_nxv2f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frinta z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrinta_nxv2f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frinta z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frinta.nxv2f16(<vscale x 2 x half> %x, <vscale x 2 x i1> %pg, <vscale x 2 x half> %y)
+ ret <vscale x 2 x half> %0
+}
+
+
+define <vscale x 2 x float> @test_svfrinta_nxv2f32_ptrue_u(double %z0, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svfrinta_nxv2f32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frinta z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrinta_nxv2f32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frinta z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frinta.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 2 x float> @test_svfrinta_nxv2f32_ptrue(double %z0, <vscale x 2 x float> %x, <vscale x 2 x float> %y) {
+; CHECK-LABEL: test_svfrinta_nxv2f32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frinta z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrinta_nxv2f32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frinta z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frinta.nxv2f32(<vscale x 2 x float> %x, <vscale x 2 x i1> %pg, <vscale x 2 x float> %y)
+ ret <vscale x 2 x float> %0
+}
+
+
+define <vscale x 4 x float> @test_svfrinta_nxv4f32_ptrue_u(double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svfrinta_nxv4f32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frinta z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrinta_nxv4f32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frinta z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frinta.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svfrinta_nxv4f32_ptrue(double %z0, <vscale x 4 x float> %x, <vscale x 4 x float> %y) {
+; CHECK-LABEL: test_svfrinta_nxv4f32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frinta z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrinta_nxv4f32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frinta z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frinta.nxv4f32(<vscale x 4 x float> %x, <vscale x 4 x i1> %pg, <vscale x 4 x float> %y)
+ ret <vscale x 4 x float> %0
+}
+
+
+define <vscale x 2 x double> @test_svfrinta_nxv2f64_ptrue_u(double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svfrinta_nxv2f64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frinta z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrinta_nxv2f64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frinta z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frinta.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svfrinta_nxv2f64_ptrue(double %z0, <vscale x 2 x double> %x, <vscale x 2 x double> %y) {
+; CHECK-LABEL: test_svfrinta_nxv2f64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frinta z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrinta_nxv2f64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frinta z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frinta.nxv2f64(<vscale x 2 x double> %x, <vscale x 2 x i1> %pg, <vscale x 2 x double> %y)
+ ret <vscale x 2 x double> %0
+}
+
+
+define <vscale x 8 x half> @test_svfrinti_nxv8f16_ptrue_u(double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svfrinti_nxv8f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frinti z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrinti_nxv8f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: frinti z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frinti.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svfrinti_nxv8f16_ptrue(double %z0, <vscale x 8 x half> %x, <vscale x 8 x half> %y) {
+; CHECK-LABEL: test_svfrinti_nxv8f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frinti z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrinti_nxv8f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: frinti z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frinti.nxv8f16(<vscale x 8 x half> %x, <vscale x 8 x i1> %pg, <vscale x 8 x half> %y)
+ ret <vscale x 8 x half> %0
+}
+
+
+define <vscale x 4 x half> @test_svfrinti_nxv4f16_ptrue_u(double %z0, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svfrinti_nxv4f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frinti z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrinti_nxv4f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frinti z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frinti.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 4 x half> @test_svfrinti_nxv4f16_ptrue(double %z0, <vscale x 4 x half> %x, <vscale x 4 x half> %y) {
+; CHECK-LABEL: test_svfrinti_nxv4f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frinti z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrinti_nxv4f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frinti z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frinti.nxv4f16(<vscale x 4 x half> %x, <vscale x 4 x i1> %pg, <vscale x 4 x half> %y)
+ ret <vscale x 4 x half> %0
+}
+
+
+define <vscale x 2 x half> @test_svfrinti_nxv2f16_ptrue_u(double %z0, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svfrinti_nxv2f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frinti z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrinti_nxv2f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frinti z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frinti.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x half> @test_svfrinti_nxv2f16_ptrue(double %z0, <vscale x 2 x half> %x, <vscale x 2 x half> %y) {
+; CHECK-LABEL: test_svfrinti_nxv2f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frinti z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrinti_nxv2f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frinti z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frinti.nxv2f16(<vscale x 2 x half> %x, <vscale x 2 x i1> %pg, <vscale x 2 x half> %y)
+ ret <vscale x 2 x half> %0
+}
+
+
+define <vscale x 2 x float> @test_svfrinti_nxv2f32_ptrue_u(double %z0, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svfrinti_nxv2f32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frinti z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrinti_nxv2f32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frinti z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frinti.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 2 x float> @test_svfrinti_nxv2f32_ptrue(double %z0, <vscale x 2 x float> %x, <vscale x 2 x float> %y) {
+; CHECK-LABEL: test_svfrinti_nxv2f32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frinti z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrinti_nxv2f32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frinti z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frinti.nxv2f32(<vscale x 2 x float> %x, <vscale x 2 x i1> %pg, <vscale x 2 x float> %y)
+ ret <vscale x 2 x float> %0
+}
+
+
+define <vscale x 4 x float> @test_svfrinti_nxv4f32_ptrue_u(double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svfrinti_nxv4f32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frinti z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrinti_nxv4f32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frinti z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frinti.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svfrinti_nxv4f32_ptrue(double %z0, <vscale x 4 x float> %x, <vscale x 4 x float> %y) {
+; CHECK-LABEL: test_svfrinti_nxv4f32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frinti z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrinti_nxv4f32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frinti z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frinti.nxv4f32(<vscale x 4 x float> %x, <vscale x 4 x i1> %pg, <vscale x 4 x float> %y)
+ ret <vscale x 4 x float> %0
+}
+
+
+define <vscale x 2 x double> @test_svfrinti_nxv2f64_ptrue_u(double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svfrinti_nxv2f64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frinti z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrinti_nxv2f64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frinti z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frinti.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svfrinti_nxv2f64_ptrue(double %z0, <vscale x 2 x double> %x, <vscale x 2 x double> %y) {
+; CHECK-LABEL: test_svfrinti_nxv2f64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frinti z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrinti_nxv2f64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frinti z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frinti.nxv2f64(<vscale x 2 x double> %x, <vscale x 2 x i1> %pg, <vscale x 2 x double> %y)
+ ret <vscale x 2 x double> %0
+}
+
+
+define <vscale x 8 x half> @test_svfrintm_nxv8f16_ptrue_u(double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svfrintm_nxv8f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintm z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintm_nxv8f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: frintm z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintm.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svfrintm_nxv8f16_ptrue(double %z0, <vscale x 8 x half> %x, <vscale x 8 x half> %y) {
+; CHECK-LABEL: test_svfrintm_nxv8f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frintm z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintm_nxv8f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: frintm z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintm.nxv8f16(<vscale x 8 x half> %x, <vscale x 8 x i1> %pg, <vscale x 8 x half> %y)
+ ret <vscale x 8 x half> %0
+}
+
+
+define <vscale x 4 x half> @test_svfrintm_nxv4f16_ptrue_u(double %z0, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svfrintm_nxv4f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintm z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintm_nxv4f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frintm z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintm.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 4 x half> @test_svfrintm_nxv4f16_ptrue(double %z0, <vscale x 4 x half> %x, <vscale x 4 x half> %y) {
+; CHECK-LABEL: test_svfrintm_nxv4f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frintm z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintm_nxv4f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frintm z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintm.nxv4f16(<vscale x 4 x half> %x, <vscale x 4 x i1> %pg, <vscale x 4 x half> %y)
+ ret <vscale x 4 x half> %0
+}
+
+
+define <vscale x 2 x half> @test_svfrintm_nxv2f16_ptrue_u(double %z0, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svfrintm_nxv2f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintm z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintm_nxv2f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frintm z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintm.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x half> @test_svfrintm_nxv2f16_ptrue(double %z0, <vscale x 2 x half> %x, <vscale x 2 x half> %y) {
+; CHECK-LABEL: test_svfrintm_nxv2f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frintm z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintm_nxv2f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frintm z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintm.nxv2f16(<vscale x 2 x half> %x, <vscale x 2 x i1> %pg, <vscale x 2 x half> %y)
+ ret <vscale x 2 x half> %0
+}
+
+
+define <vscale x 2 x float> @test_svfrintm_nxv2f32_ptrue_u(double %z0, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svfrintm_nxv2f32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintm z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintm_nxv2f32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frintm z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintm.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 2 x float> @test_svfrintm_nxv2f32_ptrue(double %z0, <vscale x 2 x float> %x, <vscale x 2 x float> %y) {
+; CHECK-LABEL: test_svfrintm_nxv2f32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frintm z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintm_nxv2f32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frintm z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintm.nxv2f32(<vscale x 2 x float> %x, <vscale x 2 x i1> %pg, <vscale x 2 x float> %y)
+ ret <vscale x 2 x float> %0
+}
+
+
+define <vscale x 4 x float> @test_svfrintm_nxv4f32_ptrue_u(double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svfrintm_nxv4f32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintm z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintm_nxv4f32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frintm z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintm.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svfrintm_nxv4f32_ptrue(double %z0, <vscale x 4 x float> %x, <vscale x 4 x float> %y) {
+; CHECK-LABEL: test_svfrintm_nxv4f32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frintm z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintm_nxv4f32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frintm z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintm.nxv4f32(<vscale x 4 x float> %x, <vscale x 4 x i1> %pg, <vscale x 4 x float> %y)
+ ret <vscale x 4 x float> %0
+}
+
+
+define <vscale x 2 x double> @test_svfrintm_nxv2f64_ptrue_u(double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svfrintm_nxv2f64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintm z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintm_nxv2f64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frintm z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintm.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svfrintm_nxv2f64_ptrue(double %z0, <vscale x 2 x double> %x, <vscale x 2 x double> %y) {
+; CHECK-LABEL: test_svfrintm_nxv2f64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frintm z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintm_nxv2f64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frintm z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintm.nxv2f64(<vscale x 2 x double> %x, <vscale x 2 x i1> %pg, <vscale x 2 x double> %y)
+ ret <vscale x 2 x double> %0
+}
+
+
+define <vscale x 8 x half> @test_svfrintn_nxv8f16_ptrue_u(double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svfrintn_nxv8f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintn z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintn_nxv8f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: frintn z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintn.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svfrintn_nxv8f16_ptrue(double %z0, <vscale x 8 x half> %x, <vscale x 8 x half> %y) {
+; CHECK-LABEL: test_svfrintn_nxv8f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frintn z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintn_nxv8f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: frintn z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintn.nxv8f16(<vscale x 8 x half> %x, <vscale x 8 x i1> %pg, <vscale x 8 x half> %y)
+ ret <vscale x 8 x half> %0
+}
+
+
+define <vscale x 4 x half> @test_svfrintn_nxv4f16_ptrue_u(double %z0, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svfrintn_nxv4f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintn z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintn_nxv4f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frintn z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintn.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 4 x half> @test_svfrintn_nxv4f16_ptrue(double %z0, <vscale x 4 x half> %x, <vscale x 4 x half> %y) {
+; CHECK-LABEL: test_svfrintn_nxv4f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frintn z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintn_nxv4f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frintn z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintn.nxv4f16(<vscale x 4 x half> %x, <vscale x 4 x i1> %pg, <vscale x 4 x half> %y)
+ ret <vscale x 4 x half> %0
+}
+
+
+define <vscale x 2 x half> @test_svfrintn_nxv2f16_ptrue_u(double %z0, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svfrintn_nxv2f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintn z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintn_nxv2f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frintn z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintn.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x half> @test_svfrintn_nxv2f16_ptrue(double %z0, <vscale x 2 x half> %x, <vscale x 2 x half> %y) {
+; CHECK-LABEL: test_svfrintn_nxv2f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frintn z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintn_nxv2f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frintn z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintn.nxv2f16(<vscale x 2 x half> %x, <vscale x 2 x i1> %pg, <vscale x 2 x half> %y)
+ ret <vscale x 2 x half> %0
+}
+
+
+define <vscale x 2 x float> @test_svfrintn_nxv2f32_ptrue_u(double %z0, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svfrintn_nxv2f32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintn z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintn_nxv2f32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frintn z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintn.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 2 x float> @test_svfrintn_nxv2f32_ptrue(double %z0, <vscale x 2 x float> %x, <vscale x 2 x float> %y) {
+; CHECK-LABEL: test_svfrintn_nxv2f32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frintn z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintn_nxv2f32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frintn z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintn.nxv2f32(<vscale x 2 x float> %x, <vscale x 2 x i1> %pg, <vscale x 2 x float> %y)
+ ret <vscale x 2 x float> %0
+}
+
+
+define <vscale x 4 x float> @test_svfrintn_nxv4f32_ptrue_u(double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svfrintn_nxv4f32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintn z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintn_nxv4f32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frintn z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintn.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svfrintn_nxv4f32_ptrue(double %z0, <vscale x 4 x float> %x, <vscale x 4 x float> %y) {
+; CHECK-LABEL: test_svfrintn_nxv4f32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frintn z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintn_nxv4f32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frintn z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintn.nxv4f32(<vscale x 4 x float> %x, <vscale x 4 x i1> %pg, <vscale x 4 x float> %y)
+ ret <vscale x 4 x float> %0
+}
+
+
+define <vscale x 2 x double> @test_svfrintn_nxv2f64_ptrue_u(double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svfrintn_nxv2f64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintn z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintn_nxv2f64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frintn z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintn.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svfrintn_nxv2f64_ptrue(double %z0, <vscale x 2 x double> %x, <vscale x 2 x double> %y) {
+; CHECK-LABEL: test_svfrintn_nxv2f64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frintn z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintn_nxv2f64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frintn z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintn.nxv2f64(<vscale x 2 x double> %x, <vscale x 2 x i1> %pg, <vscale x 2 x double> %y)
+ ret <vscale x 2 x double> %0
+}
+
+
+define <vscale x 8 x half> @test_svfrintp_nxv8f16_ptrue_u(double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svfrintp_nxv8f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintp z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintp_nxv8f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: frintp z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintp.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svfrintp_nxv8f16_ptrue(double %z0, <vscale x 8 x half> %x, <vscale x 8 x half> %y) {
+; CHECK-LABEL: test_svfrintp_nxv8f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frintp z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintp_nxv8f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: frintp z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintp.nxv8f16(<vscale x 8 x half> %x, <vscale x 8 x i1> %pg, <vscale x 8 x half> %y)
+ ret <vscale x 8 x half> %0
+}
+
+
+define <vscale x 4 x half> @test_svfrintp_nxv4f16_ptrue_u(double %z0, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svfrintp_nxv4f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintp z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintp_nxv4f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frintp z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintp.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 4 x half> @test_svfrintp_nxv4f16_ptrue(double %z0, <vscale x 4 x half> %x, <vscale x 4 x half> %y) {
+; CHECK-LABEL: test_svfrintp_nxv4f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frintp z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintp_nxv4f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frintp z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintp.nxv4f16(<vscale x 4 x half> %x, <vscale x 4 x i1> %pg, <vscale x 4 x half> %y)
+ ret <vscale x 4 x half> %0
+}
+
+
+define <vscale x 2 x half> @test_svfrintp_nxv2f16_ptrue_u(double %z0, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svfrintp_nxv2f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintp z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintp_nxv2f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frintp z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintp.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x half> @test_svfrintp_nxv2f16_ptrue(double %z0, <vscale x 2 x half> %x, <vscale x 2 x half> %y) {
+; CHECK-LABEL: test_svfrintp_nxv2f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frintp z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintp_nxv2f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frintp z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintp.nxv2f16(<vscale x 2 x half> %x, <vscale x 2 x i1> %pg, <vscale x 2 x half> %y)
+ ret <vscale x 2 x half> %0
+}
+
+
+define <vscale x 2 x float> @test_svfrintp_nxv2f32_ptrue_u(double %z0, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svfrintp_nxv2f32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintp z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintp_nxv2f32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frintp z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintp.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 2 x float> @test_svfrintp_nxv2f32_ptrue(double %z0, <vscale x 2 x float> %x, <vscale x 2 x float> %y) {
+; CHECK-LABEL: test_svfrintp_nxv2f32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frintp z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintp_nxv2f32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frintp z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintp.nxv2f32(<vscale x 2 x float> %x, <vscale x 2 x i1> %pg, <vscale x 2 x float> %y)
+ ret <vscale x 2 x float> %0
+}
+
+
+define <vscale x 4 x float> @test_svfrintp_nxv4f32_ptrue_u(double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svfrintp_nxv4f32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintp z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintp_nxv4f32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frintp z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintp.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svfrintp_nxv4f32_ptrue(double %z0, <vscale x 4 x float> %x, <vscale x 4 x float> %y) {
+; CHECK-LABEL: test_svfrintp_nxv4f32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frintp z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintp_nxv4f32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frintp z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintp.nxv4f32(<vscale x 4 x float> %x, <vscale x 4 x i1> %pg, <vscale x 4 x float> %y)
+ ret <vscale x 4 x float> %0
+}
+
+
+define <vscale x 2 x double> @test_svfrintp_nxv2f64_ptrue_u(double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svfrintp_nxv2f64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintp z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintp_nxv2f64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frintp z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintp.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svfrintp_nxv2f64_ptrue(double %z0, <vscale x 2 x double> %x, <vscale x 2 x double> %y) {
+; CHECK-LABEL: test_svfrintp_nxv2f64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frintp z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintp_nxv2f64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frintp z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintp.nxv2f64(<vscale x 2 x double> %x, <vscale x 2 x i1> %pg, <vscale x 2 x double> %y)
+ ret <vscale x 2 x double> %0
+}
+
+
+define <vscale x 8 x half> @test_svfrintx_nxv8f16_ptrue_u(double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svfrintx_nxv8f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintx z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintx_nxv8f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: frintx z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintx.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svfrintx_nxv8f16_ptrue(double %z0, <vscale x 8 x half> %x, <vscale x 8 x half> %y) {
+; CHECK-LABEL: test_svfrintx_nxv8f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frintx z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintx_nxv8f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: frintx z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintx.nxv8f16(<vscale x 8 x half> %x, <vscale x 8 x i1> %pg, <vscale x 8 x half> %y)
+ ret <vscale x 8 x half> %0
+}
+
+
+define <vscale x 4 x half> @test_svfrintx_nxv4f16_ptrue_u(double %z0, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svfrintx_nxv4f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintx z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintx_nxv4f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frintx z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintx.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 4 x half> @test_svfrintx_nxv4f16_ptrue(double %z0, <vscale x 4 x half> %x, <vscale x 4 x half> %y) {
+; CHECK-LABEL: test_svfrintx_nxv4f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frintx z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintx_nxv4f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frintx z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintx.nxv4f16(<vscale x 4 x half> %x, <vscale x 4 x i1> %pg, <vscale x 4 x half> %y)
+ ret <vscale x 4 x half> %0
+}
+
+
+define <vscale x 2 x half> @test_svfrintx_nxv2f16_ptrue_u(double %z0, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svfrintx_nxv2f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintx z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintx_nxv2f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frintx z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintx.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x half> @test_svfrintx_nxv2f16_ptrue(double %z0, <vscale x 2 x half> %x, <vscale x 2 x half> %y) {
+; CHECK-LABEL: test_svfrintx_nxv2f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frintx z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintx_nxv2f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frintx z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintx.nxv2f16(<vscale x 2 x half> %x, <vscale x 2 x i1> %pg, <vscale x 2 x half> %y)
+ ret <vscale x 2 x half> %0
+}
+
+
+define <vscale x 2 x float> @test_svfrintx_nxv2f32_ptrue_u(double %z0, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svfrintx_nxv2f32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintx z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintx_nxv2f32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frintx z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintx.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 2 x float> @test_svfrintx_nxv2f32_ptrue(double %z0, <vscale x 2 x float> %x, <vscale x 2 x float> %y) {
+; CHECK-LABEL: test_svfrintx_nxv2f32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frintx z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintx_nxv2f32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frintx z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintx.nxv2f32(<vscale x 2 x float> %x, <vscale x 2 x i1> %pg, <vscale x 2 x float> %y)
+ ret <vscale x 2 x float> %0
+}
+
+
+define <vscale x 4 x float> @test_svfrintx_nxv4f32_ptrue_u(double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svfrintx_nxv4f32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintx z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintx_nxv4f32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frintx z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintx.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svfrintx_nxv4f32_ptrue(double %z0, <vscale x 4 x float> %x, <vscale x 4 x float> %y) {
+; CHECK-LABEL: test_svfrintx_nxv4f32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frintx z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintx_nxv4f32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frintx z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintx.nxv4f32(<vscale x 4 x float> %x, <vscale x 4 x i1> %pg, <vscale x 4 x float> %y)
+ ret <vscale x 4 x float> %0
+}
+
+
+define <vscale x 2 x double> @test_svfrintx_nxv2f64_ptrue_u(double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svfrintx_nxv2f64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintx z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintx_nxv2f64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frintx z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintx.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svfrintx_nxv2f64_ptrue(double %z0, <vscale x 2 x double> %x, <vscale x 2 x double> %y) {
+; CHECK-LABEL: test_svfrintx_nxv2f64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frintx z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintx_nxv2f64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frintx z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintx.nxv2f64(<vscale x 2 x double> %x, <vscale x 2 x i1> %pg, <vscale x 2 x double> %y)
+ ret <vscale x 2 x double> %0
+}
+
+
+define <vscale x 8 x half> @test_svfrintz_nxv8f16_ptrue_u(double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svfrintz_nxv8f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintz z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintz_nxv8f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: frintz z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintz.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svfrintz_nxv8f16_ptrue(double %z0, <vscale x 8 x half> %x, <vscale x 8 x half> %y) {
+; CHECK-LABEL: test_svfrintz_nxv8f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frintz z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintz_nxv8f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: frintz z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frintz.nxv8f16(<vscale x 8 x half> %x, <vscale x 8 x i1> %pg, <vscale x 8 x half> %y)
+ ret <vscale x 8 x half> %0
+}
+
+
+define <vscale x 4 x half> @test_svfrintz_nxv4f16_ptrue_u(double %z0, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svfrintz_nxv4f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintz z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintz_nxv4f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frintz z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintz.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 4 x half> @test_svfrintz_nxv4f16_ptrue(double %z0, <vscale x 4 x half> %x, <vscale x 4 x half> %y) {
+; CHECK-LABEL: test_svfrintz_nxv4f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frintz z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintz_nxv4f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frintz z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frintz.nxv4f16(<vscale x 4 x half> %x, <vscale x 4 x i1> %pg, <vscale x 4 x half> %y)
+ ret <vscale x 4 x half> %0
+}
+
+
+define <vscale x 2 x half> @test_svfrintz_nxv2f16_ptrue_u(double %z0, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svfrintz_nxv2f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintz z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintz_nxv2f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frintz z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintz.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x half> @test_svfrintz_nxv2f16_ptrue(double %z0, <vscale x 2 x half> %x, <vscale x 2 x half> %y) {
+; CHECK-LABEL: test_svfrintz_nxv2f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frintz z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintz_nxv2f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frintz z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frintz.nxv2f16(<vscale x 2 x half> %x, <vscale x 2 x i1> %pg, <vscale x 2 x half> %y)
+ ret <vscale x 2 x half> %0
+}
+
+
+define <vscale x 2 x float> @test_svfrintz_nxv2f32_ptrue_u(double %z0, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svfrintz_nxv2f32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintz z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintz_nxv2f32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frintz z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintz.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 2 x float> @test_svfrintz_nxv2f32_ptrue(double %z0, <vscale x 2 x float> %x, <vscale x 2 x float> %y) {
+; CHECK-LABEL: test_svfrintz_nxv2f32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frintz z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintz_nxv2f32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frintz z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frintz.nxv2f32(<vscale x 2 x float> %x, <vscale x 2 x i1> %pg, <vscale x 2 x float> %y)
+ ret <vscale x 2 x float> %0
+}
+
+
+define <vscale x 4 x float> @test_svfrintz_nxv4f32_ptrue_u(double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svfrintz_nxv4f32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintz z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintz_nxv4f32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frintz z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintz.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svfrintz_nxv4f32_ptrue(double %z0, <vscale x 4 x float> %x, <vscale x 4 x float> %y) {
+; CHECK-LABEL: test_svfrintz_nxv4f32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frintz z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintz_nxv4f32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frintz z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frintz.nxv4f32(<vscale x 4 x float> %x, <vscale x 4 x i1> %pg, <vscale x 4 x float> %y)
+ ret <vscale x 4 x float> %0
+}
+
+
+define <vscale x 2 x double> @test_svfrintz_nxv2f64_ptrue_u(double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svfrintz_nxv2f64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frintz z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintz_nxv2f64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frintz z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintz.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svfrintz_nxv2f64_ptrue(double %z0, <vscale x 2 x double> %x, <vscale x 2 x double> %y) {
+; CHECK-LABEL: test_svfrintz_nxv2f64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frintz z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrintz_nxv2f64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frintz z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frintz.nxv2f64(<vscale x 2 x double> %x, <vscale x 2 x i1> %pg, <vscale x 2 x double> %y)
+ ret <vscale x 2 x double> %0
+}
+
+
+define <vscale x 8 x half> @test_svfrecpx_nxv8f16_ptrue_u(double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svfrecpx_nxv8f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frecpx z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrecpx_nxv8f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: frecpx z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frecpx.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svfrecpx_nxv8f16_ptrue(double %z0, <vscale x 8 x half> %x, <vscale x 8 x half> %y) {
+; CHECK-LABEL: test_svfrecpx_nxv8f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frecpx z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrecpx_nxv8f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: frecpx z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.frecpx.nxv8f16(<vscale x 8 x half> %x, <vscale x 8 x i1> %pg, <vscale x 8 x half> %y)
+ ret <vscale x 8 x half> %0
+}
+
+
+define <vscale x 4 x half> @test_svfrecpx_nxv4f16_ptrue_u(double %z0, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svfrecpx_nxv4f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frecpx z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrecpx_nxv4f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frecpx z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frecpx.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 4 x half> @test_svfrecpx_nxv4f16_ptrue(double %z0, <vscale x 4 x half> %x, <vscale x 4 x half> %y) {
+; CHECK-LABEL: test_svfrecpx_nxv4f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frecpx z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrecpx_nxv4f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frecpx z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.frecpx.nxv4f16(<vscale x 4 x half> %x, <vscale x 4 x i1> %pg, <vscale x 4 x half> %y)
+ ret <vscale x 4 x half> %0
+}
+
+
+define <vscale x 2 x half> @test_svfrecpx_nxv2f16_ptrue_u(double %z0, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svfrecpx_nxv2f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frecpx z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrecpx_nxv2f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frecpx z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frecpx.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x half> @test_svfrecpx_nxv2f16_ptrue(double %z0, <vscale x 2 x half> %x, <vscale x 2 x half> %y) {
+; CHECK-LABEL: test_svfrecpx_nxv2f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frecpx z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrecpx_nxv2f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frecpx z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.frecpx.nxv2f16(<vscale x 2 x half> %x, <vscale x 2 x i1> %pg, <vscale x 2 x half> %y)
+ ret <vscale x 2 x half> %0
+}
+
+
+define <vscale x 2 x float> @test_svfrecpx_nxv2f32_ptrue_u(double %z0, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svfrecpx_nxv2f32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frecpx z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrecpx_nxv2f32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frecpx z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frecpx.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 2 x float> @test_svfrecpx_nxv2f32_ptrue(double %z0, <vscale x 2 x float> %x, <vscale x 2 x float> %y) {
+; CHECK-LABEL: test_svfrecpx_nxv2f32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frecpx z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrecpx_nxv2f32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frecpx z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.frecpx.nxv2f32(<vscale x 2 x float> %x, <vscale x 2 x i1> %pg, <vscale x 2 x float> %y)
+ ret <vscale x 2 x float> %0
+}
+
+
+define <vscale x 4 x float> @test_svfrecpx_nxv4f32_ptrue_u(double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svfrecpx_nxv4f32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frecpx z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrecpx_nxv4f32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frecpx z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frecpx.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svfrecpx_nxv4f32_ptrue(double %z0, <vscale x 4 x float> %x, <vscale x 4 x float> %y) {
+; CHECK-LABEL: test_svfrecpx_nxv4f32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frecpx z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrecpx_nxv4f32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: frecpx z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.frecpx.nxv4f32(<vscale x 4 x float> %x, <vscale x 4 x i1> %pg, <vscale x 4 x float> %y)
+ ret <vscale x 4 x float> %0
+}
+
+
+define <vscale x 2 x double> @test_svfrecpx_nxv2f64_ptrue_u(double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svfrecpx_nxv2f64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: frecpx z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrecpx_nxv2f64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frecpx z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frecpx.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svfrecpx_nxv2f64_ptrue(double %z0, <vscale x 2 x double> %x, <vscale x 2 x double> %y) {
+; CHECK-LABEL: test_svfrecpx_nxv2f64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: frecpx z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfrecpx_nxv2f64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: frecpx z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.frecpx.nxv2f64(<vscale x 2 x double> %x, <vscale x 2 x i1> %pg, <vscale x 2 x double> %y)
+ ret <vscale x 2 x double> %0
+}
+
+
+define <vscale x 8 x half> @test_svfsqrt_nxv8f16_ptrue_u(double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svfsqrt_nxv8f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: fsqrt z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfsqrt_nxv8f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: fsqrt z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fsqrt.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svfsqrt_nxv8f16_ptrue(double %z0, <vscale x 8 x half> %x, <vscale x 8 x half> %y) {
+; CHECK-LABEL: test_svfsqrt_nxv8f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: fsqrt z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfsqrt_nxv8f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: fsqrt z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.fsqrt.nxv8f16(<vscale x 8 x half> %x, <vscale x 8 x i1> %pg, <vscale x 8 x half> %y)
+ ret <vscale x 8 x half> %0
+}
+
+
+define <vscale x 4 x half> @test_svfsqrt_nxv4f16_ptrue_u(double %z0, <vscale x 4 x half> %x) {
+; CHECK-LABEL: test_svfsqrt_nxv4f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: fsqrt z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfsqrt_nxv4f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: fsqrt z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.fsqrt.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> %pg, <vscale x 4 x half> %x)
+ ret <vscale x 4 x half> %0
+}
+
+define <vscale x 4 x half> @test_svfsqrt_nxv4f16_ptrue(double %z0, <vscale x 4 x half> %x, <vscale x 4 x half> %y) {
+; CHECK-LABEL: test_svfsqrt_nxv4f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: fsqrt z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfsqrt_nxv4f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: fsqrt z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x half> @llvm.aarch64.sve.fsqrt.nxv4f16(<vscale x 4 x half> %x, <vscale x 4 x i1> %pg, <vscale x 4 x half> %y)
+ ret <vscale x 4 x half> %0
+}
+
+
+define <vscale x 2 x half> @test_svfsqrt_nxv2f16_ptrue_u(double %z0, <vscale x 2 x half> %x) {
+; CHECK-LABEL: test_svfsqrt_nxv2f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: fsqrt z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfsqrt_nxv2f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fsqrt z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.fsqrt.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> %pg, <vscale x 2 x half> %x)
+ ret <vscale x 2 x half> %0
+}
+
+define <vscale x 2 x half> @test_svfsqrt_nxv2f16_ptrue(double %z0, <vscale x 2 x half> %x, <vscale x 2 x half> %y) {
+; CHECK-LABEL: test_svfsqrt_nxv2f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: fsqrt z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfsqrt_nxv2f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fsqrt z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x half> @llvm.aarch64.sve.fsqrt.nxv2f16(<vscale x 2 x half> %x, <vscale x 2 x i1> %pg, <vscale x 2 x half> %y)
+ ret <vscale x 2 x half> %0
+}
+
+
+define <vscale x 2 x float> @test_svfsqrt_nxv2f32_ptrue_u(double %z0, <vscale x 2 x float> %x) {
+; CHECK-LABEL: test_svfsqrt_nxv2f32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: fsqrt z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfsqrt_nxv2f32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fsqrt z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.fsqrt.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> %pg, <vscale x 2 x float> %x)
+ ret <vscale x 2 x float> %0
+}
+
+define <vscale x 2 x float> @test_svfsqrt_nxv2f32_ptrue(double %z0, <vscale x 2 x float> %x, <vscale x 2 x float> %y) {
+; CHECK-LABEL: test_svfsqrt_nxv2f32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: fsqrt z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfsqrt_nxv2f32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fsqrt z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x float> @llvm.aarch64.sve.fsqrt.nxv2f32(<vscale x 2 x float> %x, <vscale x 2 x i1> %pg, <vscale x 2 x float> %y)
+ ret <vscale x 2 x float> %0
+}
+
+
+define <vscale x 4 x float> @test_svfsqrt_nxv4f32_ptrue_u(double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svfsqrt_nxv4f32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: fsqrt z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfsqrt_nxv4f32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: fsqrt z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fsqrt.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svfsqrt_nxv4f32_ptrue(double %z0, <vscale x 4 x float> %x, <vscale x 4 x float> %y) {
+; CHECK-LABEL: test_svfsqrt_nxv4f32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: fsqrt z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfsqrt_nxv4f32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: fsqrt z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.fsqrt.nxv4f32(<vscale x 4 x float> %x, <vscale x 4 x i1> %pg, <vscale x 4 x float> %y)
+ ret <vscale x 4 x float> %0
+}
+
+
+define <vscale x 2 x double> @test_svfsqrt_nxv2f64_ptrue_u(double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svfsqrt_nxv2f64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: fsqrt z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfsqrt_nxv2f64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fsqrt z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fsqrt.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svfsqrt_nxv2f64_ptrue(double %z0, <vscale x 2 x double> %x, <vscale x 2 x double> %y) {
+; CHECK-LABEL: test_svfsqrt_nxv2f64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: fsqrt z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svfsqrt_nxv2f64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: fsqrt z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.fsqrt.nxv2f64(<vscale x 2 x double> %x, <vscale x 2 x i1> %pg, <vscale x 2 x double> %y)
+ ret <vscale x 2 x double> %0
+}
+
>From ec81793081ec775735cad58fde7187266349942b Mon Sep 17 00:00:00 2001
From: Momchil Velikov <momchil.velikov at arm.com>
Date: Tue, 19 Nov 2024 13:52:06 +0000
Subject: [PATCH 11/16] [AArch64] Generate zeroing forms of certain SVE2.2
instructions (9/11)
---
.../lib/Target/AArch64/AArch64SVEInstrInfo.td | 8 +-
llvm/lib/Target/AArch64/SVEInstrFormats.td | 13 +-
...eroing-forms-urecpe-ursqrte-sqabs-sqneg.ll | 478 ++++++++++++++++++
3 files changed, 494 insertions(+), 5 deletions(-)
create mode 100644 llvm/test/CodeGen/AArch64/zeroing-forms-urecpe-ursqrte-sqabs-sqneg.ll
diff --git a/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td b/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
index b289c5bd3247a4..81180eb69b20f1 100644
--- a/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
+++ b/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
@@ -4283,10 +4283,10 @@ let Predicates = [HasSVE2p2orSME2p2] in {
defm FLOGB_ZPzZ : sve_fp_z2op_p_zd_d_flogb<"flogb", int_aarch64_sve_flogb>;
// SVE2 integer unary operations, zeroing predicate
- def URECPE_ZPzZ : sve2_int_un_pred_arit_z<0b10, 0b00, "urecpe", ZPR32>;
- def URSQRTE_ZPzZ : sve2_int_un_pred_arit_z<0b10, 0b01, "ursqrte", ZPR32>;
- defm SQABS_ZPzZ : sve2_int_un_pred_arit_z<0b10, "sqabs">;
- defm SQNEG_ZPzZ : sve2_int_un_pred_arit_z<0b11, "sqneg">;
+ defm URECPE_ZPzZ : sve2_int_un_pred_arit_z_S<0b00, "urecpe", int_aarch64_sve_urecpe>;
+ defm URSQRTE_ZPzZ : sve2_int_un_pred_arit_z_S<0b01, "ursqrte", int_aarch64_sve_ursqrte>;
+ defm SQABS_ZPzZ : sve2_int_un_pred_arit_z< 0b10, "sqabs", int_aarch64_sve_sqabs>;
+ defm SQNEG_ZPzZ : sve2_int_un_pred_arit_z< 0b11, "sqneg", int_aarch64_sve_sqneg>;
// Floating point round to integral fp value in integer size range
// Merging
diff --git a/llvm/lib/Target/AArch64/SVEInstrFormats.td b/llvm/lib/Target/AArch64/SVEInstrFormats.td
index 99994a50150183..b38c06976d66d5 100644
--- a/llvm/lib/Target/AArch64/SVEInstrFormats.td
+++ b/llvm/lib/Target/AArch64/SVEInstrFormats.td
@@ -4217,11 +4217,22 @@ multiclass sve2_int_un_pred_arit<bits<2> opc, string asm, SDPatternOperator op>
defm : SVE_3_Op_Undef_Pat<nxv2i64, op, nxv2i64, nxv2i1, nxv2i64, !cast<Pseudo>(NAME # _D_UNDEF)>;
}
-multiclass sve2_int_un_pred_arit_z<bits<2> opc, string asm> {
+multiclass sve2_int_un_pred_arit_z_S<bits<2> opc, string asm, SDPatternOperator op> {
+ def _S : sve2_int_un_pred_arit_z<0b10, opc, asm, ZPR32>;
+
+ def : SVE_3_Op_UndefZero_Pat<nxv4i32, op, nxv4i32, nxv4i1, nxv4i32, !cast<Instruction>(NAME # _S)>;
+}
+
+multiclass sve2_int_un_pred_arit_z<bits<2> opc, string asm, SDPatternOperator op> {
def _B : sve2_int_un_pred_arit_z<0b00, opc, asm, ZPR8>;
def _H : sve2_int_un_pred_arit_z<0b01, opc, asm, ZPR16>;
def _S : sve2_int_un_pred_arit_z<0b10, opc, asm, ZPR32>;
def _D : sve2_int_un_pred_arit_z<0b11, opc, asm, ZPR64>;
+
+ def : SVE_3_Op_UndefZero_Pat<nxv16i8, op, nxv16i8, nxv16i1, nxv16i8, !cast<Instruction>(NAME # _B)>;
+ def : SVE_3_Op_UndefZero_Pat<nxv8i16, op, nxv8i16, nxv8i1, nxv8i16, !cast<Instruction>(NAME # _H)>;
+ def : SVE_3_Op_UndefZero_Pat<nxv4i32, op, nxv4i32, nxv4i1, nxv4i32, !cast<Instruction>(NAME # _S)>;
+ def : SVE_3_Op_UndefZero_Pat<nxv2i64, op, nxv2i64, nxv2i1, nxv2i64, !cast<Instruction>(NAME # _D)>;
}
//===----------------------------------------------------------------------===//
diff --git a/llvm/test/CodeGen/AArch64/zeroing-forms-urecpe-ursqrte-sqabs-sqneg.ll b/llvm/test/CodeGen/AArch64/zeroing-forms-urecpe-ursqrte-sqabs-sqneg.ll
new file mode 100644
index 00000000000000..3b072f71e30c56
--- /dev/null
+++ b/llvm/test/CodeGen/AArch64/zeroing-forms-urecpe-ursqrte-sqabs-sqneg.ll
@@ -0,0 +1,478 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
+; RUN: llc -mattr=+sve2 < %s | FileCheck %s
+; RUN: llc -mattr=+sve2p2 < %s | FileCheck %s -check-prefix CHECK-2p2
+
+; RUN: llc -mattr=+sme --force-streaming < %s | FileCheck %s
+; RUN: llc -mattr=+sme2p2 --force-streaming < %s | FileCheck %s -check-prefix CHECK-2p2
+
+target triple = "aarch64-linux"
+
+define <vscale x 4 x i32> @test_svrecpe_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrecpe_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: urecpe z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrecpe_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: urecpe z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.urecpe.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svrecpe_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrecpe_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: urecpe z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrecpe_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: urecpe z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.urecpe.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svrecpe_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrecpe_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: urecpe z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrecpe_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: urecpe z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.urecpe.nxv4i32(<vscale x 4 x i32> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svrsqrte_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrsqrte_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ursqrte z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrsqrte_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ursqrte z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.ursqrte.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svrsqrte_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrsqrte_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: ursqrte z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrsqrte_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ursqrte z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.ursqrte.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svrsqrte_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrsqrte_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: ursqrte z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrsqrte_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ursqrte z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.ursqrte.nxv4i32(<vscale x 4 x i32> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 16 x i8> @test_svqabs_s8_x_1(<vscale x 16 x i1> %pg, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svqabs_s8_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: sqabs z0.b, p0/m, z0.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svqabs_s8_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sqabs z0.b, p0/z, z0.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sqabs.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svqabs_s8_x_2(<vscale x 16 x i1> %pg, double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svqabs_s8_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: sqabs z0.b, p0/m, z1.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svqabs_s8_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sqabs z0.b, p0/z, z1.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sqabs.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svqabs_s8_z(<vscale x 16 x i1> %pg, double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svqabs_s8_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.b, #0 // =0x0
+; CHECK-NEXT: sqabs z0.b, p0/m, z1.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svqabs_s8_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sqabs z0.b, p0/z, z1.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sqabs.nxv16i8(<vscale x 16 x i8> zeroinitializer, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 8 x i16> @test_svqabs_s16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svqabs_s16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: sqabs z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svqabs_s16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sqabs z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sqabs.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svqabs_s16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svqabs_s16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: sqabs z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svqabs_s16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sqabs z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sqabs.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svqabs_s16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svqabs_s16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: sqabs z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svqabs_s16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sqabs z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sqabs.nxv8i16(<vscale x 8 x i16> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svqabs_s32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svqabs_s32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: sqabs z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svqabs_s32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sqabs z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sqabs.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svqabs_s32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svqabs_s32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: sqabs z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svqabs_s32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sqabs z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sqabs.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svqabs_s32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svqabs_s32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: sqabs z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svqabs_s32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sqabs z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sqabs.nxv4i32(<vscale x 4 x i32> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svqabs_s64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svqabs_s64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: sqabs z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svqabs_s64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sqabs z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sqabs.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svqabs_s64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svqabs_s64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: sqabs z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svqabs_s64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sqabs z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sqabs.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svqabs_s64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svqabs_s64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: sqabs z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svqabs_s64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sqabs z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sqabs.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 16 x i8> @test_svqneg_s8_x_1(<vscale x 16 x i1> %pg, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svqneg_s8_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: sqneg z0.b, p0/m, z0.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svqneg_s8_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sqneg z0.b, p0/z, z0.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sqneg.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svqneg_s8_x_2(<vscale x 16 x i1> %pg, double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svqneg_s8_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: sqneg z0.b, p0/m, z1.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svqneg_s8_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sqneg z0.b, p0/z, z1.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sqneg.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svqneg_s8_z(<vscale x 16 x i1> %pg, double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svqneg_s8_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.b, #0 // =0x0
+; CHECK-NEXT: sqneg z0.b, p0/m, z1.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svqneg_s8_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sqneg z0.b, p0/z, z1.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sqneg.nxv16i8(<vscale x 16 x i8> zeroinitializer, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 8 x i16> @test_svqneg_s16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svqneg_s16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: sqneg z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svqneg_s16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sqneg z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sqneg.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svqneg_s16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svqneg_s16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: sqneg z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svqneg_s16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sqneg z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sqneg.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svqneg_s16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svqneg_s16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: sqneg z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svqneg_s16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sqneg z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sqneg.nxv8i16(<vscale x 8 x i16> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svqneg_s32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svqneg_s32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: sqneg z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svqneg_s32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sqneg z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sqneg.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svqneg_s32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svqneg_s32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: sqneg z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svqneg_s32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sqneg z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sqneg.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svqneg_s32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svqneg_s32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: sqneg z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svqneg_s32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sqneg z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sqneg.nxv4i32(<vscale x 4 x i32> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svqneg_s64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svqneg_s64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: sqneg z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svqneg_s64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sqneg z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sqneg.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svqneg_s64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svqneg_s64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: sqneg z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svqneg_s64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sqneg z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sqneg.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svqneg_s64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svqneg_s64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: sqneg z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svqneg_s64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sqneg z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sqneg.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
>From 6dd286169f0dde146d0bd98f441ec132ac530a9c Mon Sep 17 00:00:00 2001
From: Momchil Velikov <momchil.velikov at arm.com>
Date: Fri, 20 Dec 2024 17:59:03 +0000
Subject: [PATCH 12/16] [fixup] Rebase, update tests, add tests for the new
all-true patterns, change undef to poison
---
llvm/lib/Target/AArch64/SVEInstrFormats.td | 10 +-
...eroing-forms-urecpe-ursqrte-sqabs-sqneg.ll | 420 +++++++++++++++++-
2 files changed, 405 insertions(+), 25 deletions(-)
diff --git a/llvm/lib/Target/AArch64/SVEInstrFormats.td b/llvm/lib/Target/AArch64/SVEInstrFormats.td
index b38c06976d66d5..c8f656c11b932f 100644
--- a/llvm/lib/Target/AArch64/SVEInstrFormats.td
+++ b/llvm/lib/Target/AArch64/SVEInstrFormats.td
@@ -4220,7 +4220,7 @@ multiclass sve2_int_un_pred_arit<bits<2> opc, string asm, SDPatternOperator op>
multiclass sve2_int_un_pred_arit_z_S<bits<2> opc, string asm, SDPatternOperator op> {
def _S : sve2_int_un_pred_arit_z<0b10, opc, asm, ZPR32>;
- def : SVE_3_Op_UndefZero_Pat<nxv4i32, op, nxv4i32, nxv4i1, nxv4i32, !cast<Instruction>(NAME # _S)>;
+ defm : SVE_3_Op_UndefZero_Pat<nxv4i32, op, nxv4i32, nxv4i1, nxv4i32, !cast<Instruction>(NAME # _S)>;
}
multiclass sve2_int_un_pred_arit_z<bits<2> opc, string asm, SDPatternOperator op> {
@@ -4229,10 +4229,10 @@ multiclass sve2_int_un_pred_arit_z<bits<2> opc, string asm, SDPatternOperator op
def _S : sve2_int_un_pred_arit_z<0b10, opc, asm, ZPR32>;
def _D : sve2_int_un_pred_arit_z<0b11, opc, asm, ZPR64>;
- def : SVE_3_Op_UndefZero_Pat<nxv16i8, op, nxv16i8, nxv16i1, nxv16i8, !cast<Instruction>(NAME # _B)>;
- def : SVE_3_Op_UndefZero_Pat<nxv8i16, op, nxv8i16, nxv8i1, nxv8i16, !cast<Instruction>(NAME # _H)>;
- def : SVE_3_Op_UndefZero_Pat<nxv4i32, op, nxv4i32, nxv4i1, nxv4i32, !cast<Instruction>(NAME # _S)>;
- def : SVE_3_Op_UndefZero_Pat<nxv2i64, op, nxv2i64, nxv2i1, nxv2i64, !cast<Instruction>(NAME # _D)>;
+ defm : SVE_3_Op_UndefZero_Pat<nxv16i8, op, nxv16i8, nxv16i1, nxv16i8, !cast<Instruction>(NAME # _B)>;
+ defm : SVE_3_Op_UndefZero_Pat<nxv8i16, op, nxv8i16, nxv8i1, nxv8i16, !cast<Instruction>(NAME # _H)>;
+ defm : SVE_3_Op_UndefZero_Pat<nxv4i32, op, nxv4i32, nxv4i1, nxv4i32, !cast<Instruction>(NAME # _S)>;
+ defm : SVE_3_Op_UndefZero_Pat<nxv2i64, op, nxv2i64, nxv2i1, nxv2i64, !cast<Instruction>(NAME # _D)>;
}
//===----------------------------------------------------------------------===//
diff --git a/llvm/test/CodeGen/AArch64/zeroing-forms-urecpe-ursqrte-sqabs-sqneg.ll b/llvm/test/CodeGen/AArch64/zeroing-forms-urecpe-ursqrte-sqabs-sqneg.ll
index 3b072f71e30c56..787ac4458079c6 100644
--- a/llvm/test/CodeGen/AArch64/zeroing-forms-urecpe-ursqrte-sqabs-sqneg.ll
+++ b/llvm/test/CodeGen/AArch64/zeroing-forms-urecpe-ursqrte-sqabs-sqneg.ll
@@ -18,7 +18,7 @@ define <vscale x 4 x i32> @test_svrecpe_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x
; CHECK-2p2-NEXT: urecpe z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.urecpe.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.urecpe.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -34,7 +34,7 @@ define <vscale x 4 x i32> @test_svrecpe_x_2(<vscale x 4 x i1> %pg, double %z0, <
; CHECK-2p2-NEXT: urecpe z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.urecpe.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.urecpe.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -65,7 +65,7 @@ define <vscale x 4 x i32> @test_svrsqrte_x_1(<vscale x 4 x i1> %pg, <vscale x 4
; CHECK-2p2-NEXT: ursqrte z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.ursqrte.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.ursqrte.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -81,7 +81,7 @@ define <vscale x 4 x i32> @test_svrsqrte_x_2(<vscale x 4 x i1> %pg, double %z0,
; CHECK-2p2-NEXT: ursqrte z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.ursqrte.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.ursqrte.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -112,7 +112,7 @@ define <vscale x 16 x i8> @test_svqabs_s8_x_1(<vscale x 16 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: sqabs z0.b, p0/z, z0.b
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sqabs.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sqabs.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
ret <vscale x 16 x i8> %0
}
@@ -128,7 +128,7 @@ define <vscale x 16 x i8> @test_svqabs_s8_x_2(<vscale x 16 x i1> %pg, double %z0
; CHECK-2p2-NEXT: sqabs z0.b, p0/z, z1.b
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sqabs.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sqabs.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
ret <vscale x 16 x i8> %0
}
@@ -159,7 +159,7 @@ define <vscale x 8 x i16> @test_svqabs_s16_x_1(<vscale x 8 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: sqabs z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sqabs.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sqabs.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x i16> %0
}
@@ -175,7 +175,7 @@ define <vscale x 8 x i16> @test_svqabs_s16_x_2(<vscale x 8 x i1> %pg, double %z0
; CHECK-2p2-NEXT: sqabs z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sqabs.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sqabs.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x i16> %0
}
@@ -206,7 +206,7 @@ define <vscale x 4 x i32> @test_svqabs_s32_x_1(<vscale x 4 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: sqabs z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sqabs.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sqabs.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -222,7 +222,7 @@ define <vscale x 4 x i32> @test_svqabs_s32_x_2(<vscale x 4 x i1> %pg, double %z0
; CHECK-2p2-NEXT: sqabs z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sqabs.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sqabs.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -253,7 +253,7 @@ define <vscale x 2 x i64> @test_svqabs_s64_x_1(<vscale x 2 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: sqabs z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sqabs.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sqabs.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -269,7 +269,7 @@ define <vscale x 2 x i64> @test_svqabs_s64_x_2(<vscale x 2 x i1> %pg, double %z0
; CHECK-2p2-NEXT: sqabs z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sqabs.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sqabs.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -300,7 +300,7 @@ define <vscale x 16 x i8> @test_svqneg_s8_x_1(<vscale x 16 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: sqneg z0.b, p0/z, z0.b
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sqneg.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sqneg.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
ret <vscale x 16 x i8> %0
}
@@ -316,7 +316,7 @@ define <vscale x 16 x i8> @test_svqneg_s8_x_2(<vscale x 16 x i1> %pg, double %z0
; CHECK-2p2-NEXT: sqneg z0.b, p0/z, z1.b
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sqneg.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sqneg.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
ret <vscale x 16 x i8> %0
}
@@ -347,7 +347,7 @@ define <vscale x 8 x i16> @test_svqneg_s16_x_1(<vscale x 8 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: sqneg z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sqneg.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sqneg.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x i16> %0
}
@@ -363,7 +363,7 @@ define <vscale x 8 x i16> @test_svqneg_s16_x_2(<vscale x 8 x i1> %pg, double %z0
; CHECK-2p2-NEXT: sqneg z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sqneg.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sqneg.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x i16> %0
}
@@ -394,7 +394,7 @@ define <vscale x 4 x i32> @test_svqneg_s32_x_1(<vscale x 4 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: sqneg z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sqneg.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sqneg.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -410,7 +410,7 @@ define <vscale x 4 x i32> @test_svqneg_s32_x_2(<vscale x 4 x i1> %pg, double %z0
; CHECK-2p2-NEXT: sqneg z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sqneg.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sqneg.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -441,7 +441,7 @@ define <vscale x 2 x i64> @test_svqneg_s64_x_1(<vscale x 2 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: sqneg z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sqneg.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sqneg.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -457,7 +457,7 @@ define <vscale x 2 x i64> @test_svqneg_s64_x_2(<vscale x 2 x i1> %pg, double %z0
; CHECK-2p2-NEXT: sqneg z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sqneg.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sqneg.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -476,3 +476,383 @@ entry:
%0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sqneg.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
+
+define <vscale x 4 x i32> @test_svurecpe_nxv4i32_ptrue_u(double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svurecpe_nxv4i32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: urecpe z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svurecpe_nxv4i32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: urecpe z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.urecpe.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svurecpe_nxv4i32_ptrue(double %z0, <vscale x 4 x i32> %x, <vscale x 4 x i32> %y) {
+; CHECK-LABEL: test_svurecpe_nxv4i32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: urecpe z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svurecpe_nxv4i32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: urecpe z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.urecpe.nxv4i32(<vscale x 4 x i32> %x, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %y)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svursqrte_nxv4i32_ptrue_u(double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svursqrte_nxv4i32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: ursqrte z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svursqrte_nxv4i32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: ursqrte z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.ursqrte.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svursqrte_nxv4i32_ptrue(double %z0, <vscale x 4 x i32> %x, <vscale x 4 x i32> %y) {
+; CHECK-LABEL: test_svursqrte_nxv4i32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: ursqrte z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svursqrte_nxv4i32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: ursqrte z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.ursqrte.nxv4i32(<vscale x 4 x i32> %x, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %y)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 16 x i8> @test_svsqabs_nxv16i8_ptrue_u(double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svsqabs_nxv16i8_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.b
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: sqabs z0.b, p0/m, z1.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqabs_nxv16i8_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.b
+; CHECK-2p2-NEXT: sqabs z0.b, p0/z, z1.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sqabs.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svsqabs_nxv16i8_ptrue(double %z0, <vscale x 16 x i8> %x, <vscale x 16 x i8> %y) {
+; CHECK-LABEL: test_svsqabs_nxv16i8_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.b
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: sqabs z0.b, p0/m, z2.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqabs_nxv16i8_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.b
+; CHECK-2p2-NEXT: sqabs z0.b, p0/z, z2.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sqabs.nxv16i8(<vscale x 16 x i8> %x, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %y)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 8 x i16> @test_svsqabs_nxv8i16_ptrue_u(double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svsqabs_nxv8i16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: sqabs z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqabs_nxv8i16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: sqabs z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sqabs.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svsqabs_nxv8i16_ptrue(double %z0, <vscale x 8 x i16> %x, <vscale x 8 x i16> %y) {
+; CHECK-LABEL: test_svsqabs_nxv8i16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: sqabs z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqabs_nxv8i16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: sqabs z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sqabs.nxv8i16(<vscale x 8 x i16> %x, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %y)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svsqabs_nxv4i32_ptrue_u(double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svsqabs_nxv4i32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: sqabs z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqabs_nxv4i32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: sqabs z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sqabs.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svsqabs_nxv4i32_ptrue(double %z0, <vscale x 4 x i32> %x, <vscale x 4 x i32> %y) {
+; CHECK-LABEL: test_svsqabs_nxv4i32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: sqabs z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqabs_nxv4i32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: sqabs z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sqabs.nxv4i32(<vscale x 4 x i32> %x, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %y)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svsqabs_nxv2i64_ptrue_u(double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svsqabs_nxv2i64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: sqabs z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqabs_nxv2i64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: sqabs z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sqabs.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svsqabs_nxv2i64_ptrue(double %z0, <vscale x 2 x i64> %x, <vscale x 2 x i64> %y) {
+; CHECK-LABEL: test_svsqabs_nxv2i64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: sqabs z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqabs_nxv2i64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: sqabs z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sqabs.nxv2i64(<vscale x 2 x i64> %x, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %y)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 16 x i8> @test_svsqneg_nxv16i8_ptrue_u(double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svsqneg_nxv16i8_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.b
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: sqneg z0.b, p0/m, z1.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqneg_nxv16i8_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.b
+; CHECK-2p2-NEXT: sqneg z0.b, p0/z, z1.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sqneg.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svsqneg_nxv16i8_ptrue(double %z0, <vscale x 16 x i8> %x, <vscale x 16 x i8> %y) {
+; CHECK-LABEL: test_svsqneg_nxv16i8_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.b
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: sqneg z0.b, p0/m, z2.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqneg_nxv16i8_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.b
+; CHECK-2p2-NEXT: sqneg z0.b, p0/z, z2.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sqneg.nxv16i8(<vscale x 16 x i8> %x, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %y)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 8 x i16> @test_svsqneg_nxv8i16_ptrue_u(double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svsqneg_nxv8i16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: sqneg z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqneg_nxv8i16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: sqneg z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sqneg.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svsqneg_nxv8i16_ptrue(double %z0, <vscale x 8 x i16> %x, <vscale x 8 x i16> %y) {
+; CHECK-LABEL: test_svsqneg_nxv8i16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: sqneg z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqneg_nxv8i16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: sqneg z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sqneg.nxv8i16(<vscale x 8 x i16> %x, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %y)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svsqneg_nxv4i32_ptrue_u(double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svsqneg_nxv4i32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: sqneg z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqneg_nxv4i32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: sqneg z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sqneg.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svsqneg_nxv4i32_ptrue(double %z0, <vscale x 4 x i32> %x, <vscale x 4 x i32> %y) {
+; CHECK-LABEL: test_svsqneg_nxv4i32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: sqneg z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqneg_nxv4i32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: sqneg z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sqneg.nxv4i32(<vscale x 4 x i32> %x, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %y)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svsqneg_nxv2i64_ptrue_u(double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svsqneg_nxv2i64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: sqneg z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqneg_nxv2i64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: sqneg z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sqneg.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svsqneg_nxv2i64_ptrue(double %z0, <vscale x 2 x i64> %x, <vscale x 2 x i64> %y) {
+; CHECK-LABEL: test_svsqneg_nxv2i64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: sqneg z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsqneg_nxv2i64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: sqneg z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sqneg.nxv2i64(<vscale x 2 x i64> %x, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %y)
+ ret <vscale x 2 x i64> %0
+}
>From 5d74f3e61ae28d99a836f0d474710819b534e0a7 Mon Sep 17 00:00:00 2001
From: Momchil Velikov <momchil.velikov at arm.com>
Date: Tue, 19 Nov 2024 14:23:44 +0000
Subject: [PATCH 13/16] [AArch64] Generate zeroing forms of certain SVE2.2
instructions (10/11)
---
.../lib/Target/AArch64/AArch64SVEInstrInfo.td | 10 +-
llvm/lib/Target/AArch64/SVEInstrFormats.td | 38 +-
.../test/CodeGen/AArch64/zeroing-forms-rev.ll | 1480 +++++++++++++++++
3 files changed, 1520 insertions(+), 8 deletions(-)
create mode 100644 llvm/test/CodeGen/AArch64/zeroing-forms-rev.ll
diff --git a/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td b/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
index 81180eb69b20f1..fc84e8056f496e 100644
--- a/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
+++ b/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
@@ -4339,11 +4339,11 @@ let Predicates = [HasSVE2p2orSME2p2] in {
defm LASTP_XPP : sve_int_pcount_pred_tmp<0b010, "lastp">;
// SVE reverse within elements, zeroing predicate
- defm RBIT_ZPzZ : sve_int_perm_rev_rbit_z<"rbit">;
- defm REVB_ZPzZ : sve_int_perm_rev_revb_z<"revb">;
- defm REVH_ZPzZ : sve_int_perm_rev_revh_z<"revh">;
- def REVW_ZPzZ : sve_int_perm_rev_z<0b11, 0b0110, "revw", ZPR64>;
- def REVD_ZPzZ : sve_int_perm_rev_z<0b00, 0b1110, "revd", ZPR128>;
+ defm RBIT_ZPzZ : sve_int_perm_rev_rbit_z<"rbit", AArch64rbit_mt>;
+ defm REVB_ZPzZ : sve_int_perm_rev_revb_z<"revb", AArch64revb_mt>;
+ defm REVH_ZPzZ : sve_int_perm_rev_revh_z<"revh", AArch64revh_mt>;
+ defm REVW_ZPzZ : sve_int_perm_rev_revw_z<"revw", AArch64revw_mt>;
+ defm REVD_ZPzZ : sve_int_perm_rev_revd_z<"revd", AArch64revd_mt>;
} // End HasSME2p2orSVE2p2
//===----------------------------------------------------------------------===//
diff --git a/llvm/lib/Target/AArch64/SVEInstrFormats.td b/llvm/lib/Target/AArch64/SVEInstrFormats.td
index c8f656c11b932f..bf4125c487a048 100644
--- a/llvm/lib/Target/AArch64/SVEInstrFormats.td
+++ b/llvm/lib/Target/AArch64/SVEInstrFormats.td
@@ -7674,22 +7674,54 @@ class sve_int_perm_rev_z<bits<2> sz, bits<4> opc, string asm,
let hasSideEffects = 0;
}
-multiclass sve_int_perm_rev_rbit_z<string asm> {
+multiclass sve_int_perm_rev_rbit_z<string asm, SDPatternOperator op> {
def _B : sve_int_perm_rev_z<0b00, 0b0111, asm, ZPR8>;
def _H : sve_int_perm_rev_z<0b01, 0b0111, asm, ZPR16>;
def _S : sve_int_perm_rev_z<0b10, 0b0111, asm, ZPR32>;
def _D : sve_int_perm_rev_z<0b11, 0b0111, asm, ZPR64>;
+
+ def : SVE_1_Op_PassthruUndefZero_Pat<nxv16i8, op, nxv16i1, nxv16i8, !cast<Instruction>(NAME # _B)>;
+ def : SVE_1_Op_PassthruUndefZero_Pat<nxv8i16, op, nxv8i1, nxv8i16, !cast<Instruction>(NAME # _H)>;
+ def : SVE_1_Op_PassthruUndefZero_Pat<nxv4i32, op, nxv4i1, nxv4i32, !cast<Instruction>(NAME # _S)>;
+ def : SVE_1_Op_PassthruUndefZero_Pat<nxv2i64, op, nxv2i1, nxv2i64, !cast<Instruction>(NAME # _D)>;
}
-multiclass sve_int_perm_rev_revb_z<string asm> {
+multiclass sve_int_perm_rev_revb_z<string asm, SDPatternOperator op> {
def _H : sve_int_perm_rev_z<0b01, 0b0100, asm, ZPR16>;
def _S : sve_int_perm_rev_z<0b10, 0b0100, asm, ZPR32>;
def _D : sve_int_perm_rev_z<0b11, 0b0100, asm, ZPR64>;
+
+ def : SVE_1_Op_PassthruUndefZero_Pat<nxv8i16, op, nxv8i1, nxv8i16, !cast<Instruction>(NAME # _H)>;
+ def : SVE_1_Op_PassthruUndefZero_Pat<nxv4i32, op, nxv4i1, nxv4i32, !cast<Instruction>(NAME # _S)>;
+ def : SVE_1_Op_PassthruUndefZero_Pat<nxv2i64, op, nxv2i1, nxv2i64, !cast<Instruction>(NAME # _D)>;
}
-multiclass sve_int_perm_rev_revh_z<string asm> {
+multiclass sve_int_perm_rev_revh_z<string asm, SDPatternOperator op> {
def _S : sve_int_perm_rev_z<0b10, 0b0101, asm, ZPR32>;
def _D : sve_int_perm_rev_z<0b11, 0b0101, asm, ZPR64>;
+
+ def : SVE_1_Op_PassthruUndefZero_Pat<nxv4i32, op, nxv4i1, nxv4i32, !cast<Instruction>(NAME # _S)>;
+ def : SVE_1_Op_PassthruUndefZero_Pat<nxv2i64, op, nxv2i1, nxv2i64, !cast<Instruction>(NAME # _D)>;
+}
+
+multiclass sve_int_perm_rev_revw_z<string asm, SDPatternOperator op> {
+ def _D : sve_int_perm_rev_z<0b11, 0b0110, asm, ZPR64>;
+
+ def : SVE_1_Op_PassthruUndefZero_Pat<nxv2i64, op, nxv2i1, nxv2i64, !cast<Instruction>(NAME # _D)>;
+}
+
+multiclass sve_int_perm_rev_revd_z<string asm, SDPatternOperator op> {
+ def NAME : sve_int_perm_rev_z<0b00, 0b1110, asm, ZPR128>;
+
+ def : SVE_1_Op_PassthruUndefZero_Pat<nxv16i8, op, nxv16i1, nxv16i8, !cast<Instruction>(NAME)>;
+ def : SVE_1_Op_PassthruUndefZero_Pat<nxv8i16, op, nxv8i1, nxv8i16, !cast<Instruction>(NAME)>;
+ def : SVE_1_Op_PassthruUndefZero_Pat<nxv4i32, op, nxv4i1, nxv4i32, !cast<Instruction>(NAME)>;
+ def : SVE_1_Op_PassthruUndefZero_Pat<nxv2i64, op, nxv2i1, nxv2i64, !cast<Instruction>(NAME)>;
+
+ def : SVE_1_Op_PassthruUndefZero_Pat<nxv8bf16, op, nxv8i1, nxv8bf16, !cast<Instruction>(NAME)>;
+ def : SVE_1_Op_PassthruUndefZero_Pat<nxv8f16, op, nxv8i1, nxv8f16, !cast<Instruction>(NAME)>;
+ def : SVE_1_Op_PassthruUndefZero_Pat<nxv4f32, op, nxv4i1, nxv4f32, !cast<Instruction>(NAME)>;
+ def : SVE_1_Op_PassthruUndefZero_Pat<nxv2f64, op, nxv2i1, nxv2f64, !cast<Instruction>(NAME)>;
}
class sve_int_perm_cpy_r<bits<2> sz8_64, string asm, ZPRRegOp zprty,
diff --git a/llvm/test/CodeGen/AArch64/zeroing-forms-rev.ll b/llvm/test/CodeGen/AArch64/zeroing-forms-rev.ll
new file mode 100644
index 00000000000000..b7c2b1fcd99849
--- /dev/null
+++ b/llvm/test/CodeGen/AArch64/zeroing-forms-rev.ll
@@ -0,0 +1,1480 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
+; RUN: llc -mattr=+sve2p1 < %s | FileCheck %s
+; RUN: llc -mattr=+sve2p2 < %s | FileCheck %s -check-prefix CHECK-2p2
+
+; RUN: llc -mattr=+sme -force-streaming < %s | FileCheck %s
+; RUN: llc -mattr=+sme2p2 -force-streaming < %s | FileCheck %s -check-prefix CHECK-2p2
+
+target triple = "aarch64-linux"
+
+define <vscale x 16 x i8> @test_svrbit_u8_x_1(<vscale x 16 x i1> %pg, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svrbit_u8_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: rbit z0.b, p0/m, z0.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_u8_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: rbit z0.b, p0/z, z0.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.rbit.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svrbit_u8_x_2(<vscale x 16 x i1> %pg, double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svrbit_u8_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: rbit z0.b, p0/m, z1.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_u8_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: rbit z0.b, p0/z, z1.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.rbit.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svrbit_u8_z(<vscale x 16 x i1> %pg, double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svrbit_u8_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.b, #0 // =0x0
+; CHECK-NEXT: rbit z0.b, p0/m, z1.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_u8_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: rbit z0.b, p0/z, z1.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.rbit.nxv16i8(<vscale x 16 x i8> zeroinitializer, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 8 x i16> @test_svrbit_u16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svrbit_u16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: rbit z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_u16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: rbit z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.rbit.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svrbit_u16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svrbit_u16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: rbit z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_u16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: rbit z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.rbit.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svrbit_u16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svrbit_u16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: rbit z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_u16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: rbit z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.rbit.nxv8i16(<vscale x 8 x i16> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svrbit_u32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrbit_u32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: rbit z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_u32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: rbit z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.rbit.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svrbit_u32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrbit_u32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: rbit z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_u32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: rbit z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.rbit.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svrbit_u32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrbit_u32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: rbit z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_u32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: rbit z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.rbit.nxv4i32(<vscale x 4 x i32> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svrbit_u64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrbit_u64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: rbit z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_u64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: rbit z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.rbit.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svrbit_u64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrbit_u64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: rbit z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_u64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: rbit z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.rbit.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svrbit_u64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrbit_u64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: rbit z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_u64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: rbit z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.rbit.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 16 x i8> @test_svrbit_s8_x_1(<vscale x 16 x i1> %pg, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svrbit_s8_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: rbit z0.b, p0/m, z0.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_s8_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: rbit z0.b, p0/z, z0.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.rbit.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svrbit_s8_x_2(<vscale x 16 x i1> %pg, double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svrbit_s8_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: rbit z0.b, p0/m, z1.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_s8_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: rbit z0.b, p0/z, z1.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.rbit.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svrbit_s8_z(<vscale x 16 x i1> %pg, double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svrbit_s8_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.b, #0 // =0x0
+; CHECK-NEXT: rbit z0.b, p0/m, z1.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_s8_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: rbit z0.b, p0/z, z1.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.rbit.nxv16i8(<vscale x 16 x i8> zeroinitializer, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 8 x i16> @test_svrbit_s16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svrbit_s16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: rbit z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_s16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: rbit z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.rbit.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svrbit_s16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svrbit_s16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: rbit z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_s16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: rbit z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.rbit.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svrbit_s16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svrbit_s16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: rbit z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_s16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: rbit z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.rbit.nxv8i16(<vscale x 8 x i16> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svrbit_s32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrbit_s32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: rbit z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_s32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: rbit z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.rbit.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svrbit_s32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrbit_s32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: rbit z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_s32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: rbit z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.rbit.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svrbit_s32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrbit_s32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: rbit z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_s32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: rbit z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.rbit.nxv4i32(<vscale x 4 x i32> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svrbit_s64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrbit_s64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: rbit z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_s64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: rbit z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.rbit.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svrbit_s64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrbit_s64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: rbit z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_s64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: rbit z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.rbit.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svrbit_s64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrbit_s64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: rbit z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_s64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: rbit z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.rbit.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 8 x i16> @test_svrevb_u16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svrevb_u16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revb z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevb_u16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revb z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.revb.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svrevb_u16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svrevb_u16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revb z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevb_u16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revb z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.revb.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svrevb_u16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svrevb_u16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: revb z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevb_u16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revb z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.revb.nxv8i16(<vscale x 8 x i16> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svrevb_u32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrevb_u32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revb z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevb_u32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revb z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revb.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svrevb_u32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrevb_u32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revb z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevb_u32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revb z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revb.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svrevb_u32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrevb_u32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: revb z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevb_u32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revb z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revb.nxv4i32(<vscale x 4 x i32> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svrevb_u64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrevb_u64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revb z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevb_u64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revb z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revb.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svrevb_u64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrevb_u64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revb z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevb_u64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revb z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revb.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svrevb_u64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrevb_u64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: revb z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevb_u64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revb z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revb.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 8 x i16> @test_svrevb_s16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svrevb_s16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revb z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevb_s16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revb z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.revb.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svrevb_s16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svrevb_s16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revb z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevb_s16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revb z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.revb.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svrevb_s16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svrevb_s16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: revb z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevb_s16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revb z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.revb.nxv8i16(<vscale x 8 x i16> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svrevb_s32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrevb_s32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revb z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevb_s32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revb z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revb.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svrevb_s32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrevb_s32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revb z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevb_s32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revb z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revb.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svrevb_s32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrevb_s32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: revb z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevb_s32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revb z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revb.nxv4i32(<vscale x 4 x i32> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svrevb_s64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrevb_s64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revb z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevb_s64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revb z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revb.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svrevb_s64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrevb_s64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revb z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevb_s64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revb z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revb.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svrevb_s64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrevb_s64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: revb z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevb_s64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revb z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revb.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 4 x i32> @test_svrevh_u32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrevh_u32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revh z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevh_u32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revh z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revh.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svrevh_u32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrevh_u32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revh z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevh_u32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revh z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revh.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svrevh_u32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrevh_u32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: revh z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevh_u32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revh z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revh.nxv4i32(<vscale x 4 x i32> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svrevh_u64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrevh_u64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revh z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevh_u64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revh z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revh.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svrevh_u64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrevh_u64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revh z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevh_u64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revh z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revh.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svrevh_u64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrevh_u64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: revh z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevh_u64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revh z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revh.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 4 x i32> @test_svrevh_s32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrevh_s32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revh z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevh_s32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revh z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revh.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svrevh_s32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrevh_s32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revh z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevh_s32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revh z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revh.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svrevh_s32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrevh_s32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: revh z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevh_s32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revh z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revh.nxv4i32(<vscale x 4 x i32> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svrevh_s64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrevh_s64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revh z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevh_s64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revh z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revh.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svrevh_s64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrevh_s64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revh z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevh_s64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revh z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revh.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svrevh_s64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrevh_s64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: revh z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevh_s64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revh z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revh.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svrevw_u64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrevw_u64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revw z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevw_u64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revw z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revw.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svrevw_u64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrevw_u64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revw z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevw_u64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revw z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revw.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svrevw_u64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrevw_u64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: revw z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevw_u64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revw z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revw.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svrevw_s64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrevw_s64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revw z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevw_s64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revw z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revw.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svrevw_s64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrevw_s64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revw z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevw_s64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revw z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revw.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svrevw_s64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrevw_s64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: revw z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevw_s64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revw z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revw.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 16 x i8> @test_svrevd_u8_x_1(<vscale x 16 x i1> %pg, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svrevd_u8_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revd z0.q, p0/m, z0.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_u8_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z0.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.revd.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svrevd_u8_x_2(<vscale x 16 x i1> %pg, double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svrevd_u8_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_u8_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.revd.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svrevd_u8_z(<vscale x 16 x i1> %pg, double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svrevd_u8_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.b, #0 // =0x0
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_u8_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.revd.nxv16i8(<vscale x 16 x i8> zeroinitializer, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 8 x i16> @test_svrevd_u16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svrevd_u16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revd z0.q, p0/m, z0.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_u16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z0.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.revd.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svrevd_u16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svrevd_u16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_u16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.revd.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svrevd_u16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svrevd_u16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_u16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.revd.nxv8i16(<vscale x 8 x i16> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svrevd_u32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrevd_u32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revd z0.q, p0/m, z0.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_u32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z0.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revd.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svrevd_u32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrevd_u32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_u32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revd.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svrevd_u32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrevd_u32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_u32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revd.nxv4i32(<vscale x 4 x i32> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svrevd_u64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrevd_u64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revd z0.q, p0/m, z0.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_u64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z0.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revd.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svrevd_u64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrevd_u64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_u64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revd.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svrevd_u64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrevd_u64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_u64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revd.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 16 x i8> @test_svrevd_s8_x_1(<vscale x 16 x i1> %pg, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svrevd_s8_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revd z0.q, p0/m, z0.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_s8_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z0.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.revd.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svrevd_s8_x_2(<vscale x 16 x i1> %pg, double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svrevd_s8_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_s8_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.revd.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svrevd_s8_z(<vscale x 16 x i1> %pg, double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svrevd_s8_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.b, #0 // =0x0
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_s8_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.revd.nxv16i8(<vscale x 16 x i8> zeroinitializer, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 8 x i16> @test_svrevd_s16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svrevd_s16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revd z0.q, p0/m, z0.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_s16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z0.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.revd.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svrevd_s16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svrevd_s16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_s16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.revd.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svrevd_s16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svrevd_s16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_s16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.revd.nxv8i16(<vscale x 8 x i16> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svrevd_s32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrevd_s32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revd z0.q, p0/m, z0.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_s32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z0.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revd.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svrevd_s32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrevd_s32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_s32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revd.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svrevd_s32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrevd_s32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_s32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revd.nxv4i32(<vscale x 4 x i32> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svrevd_s64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrevd_s64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revd z0.q, p0/m, z0.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_s64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z0.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revd.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svrevd_s64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrevd_s64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_s64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revd.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svrevd_s64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrevd_s64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_s64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revd.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 8 x half> @test_svrevd_f16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svrevd_f16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revd z0.q, p0/m, z0.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_f16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z0.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.revd.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svrevd_f16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svrevd_f16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_f16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.revd.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svrevd_f16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svrevd_f16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_f16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.revd.nxv8f16(<vscale x 8 x half> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x bfloat> @test_svrevd_bf16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x bfloat> %x) {
+; CHECK-LABEL: test_svrevd_bf16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revd z0.q, p0/m, z0.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_bf16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z0.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x bfloat> @llvm.aarch64.sve.revd.nxv8bf16(<vscale x 8 x bfloat> undef, <vscale x 8 x i1> %pg, <vscale x 8 x bfloat> %x)
+ ret <vscale x 8 x bfloat> %0
+}
+
+define <vscale x 8 x bfloat> @test_svrevd_bf16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x bfloat> %x) {
+; CHECK-LABEL: test_svrevd_bf16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_bf16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x bfloat> @llvm.aarch64.sve.revd.nxv8bf16(<vscale x 8 x bfloat> undef, <vscale x 8 x i1> %pg, <vscale x 8 x bfloat> %x)
+ ret <vscale x 8 x bfloat> %0
+}
+
+define <vscale x 8 x bfloat> @test_svrevd_bf16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x bfloat> %x) {
+; CHECK-LABEL: test_svrevd_bf16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_bf16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x bfloat> @llvm.aarch64.sve.revd.nxv8bf16(<vscale x 8 x bfloat> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x bfloat> %x)
+ ret <vscale x 8 x bfloat> %0
+}
+
+define <vscale x 4 x float> @test_svrevd_f32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svrevd_f32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revd z0.q, p0/m, z0.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_f32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z0.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.revd.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svrevd_f32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svrevd_f32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_f32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.revd.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svrevd_f32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svrevd_f32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_f32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.revd.nxv4f32(<vscale x 4 x float> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 2 x double> @test_svrevd_f64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svrevd_f64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revd z0.q, p0/m, z0.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_f64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z0.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.revd.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svrevd_f64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svrevd_f64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_f64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.revd.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svrevd_f64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svrevd_f64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_f64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.revd.nxv2f64(<vscale x 2 x double> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
>From a876e350fc6d2b9e94fde068dd5e8af94ae2ad2c Mon Sep 17 00:00:00 2001
From: Momchil Velikov <momchil.velikov at arm.com>
Date: Fri, 20 Dec 2024 18:01:48 +0000
Subject: [PATCH 14/16] [fixup] Rebase, update tests, add tests for the new
all-true patterns, change undef to poison
---
llvm/lib/Target/AArch64/SVEInstrFormats.td | 36 +-
.../test/CodeGen/AArch64/zeroing-forms-rev.ll | 1404 +++++++++--------
2 files changed, 731 insertions(+), 709 deletions(-)
diff --git a/llvm/lib/Target/AArch64/SVEInstrFormats.td b/llvm/lib/Target/AArch64/SVEInstrFormats.td
index bf4125c487a048..b2a245586c6e32 100644
--- a/llvm/lib/Target/AArch64/SVEInstrFormats.td
+++ b/llvm/lib/Target/AArch64/SVEInstrFormats.td
@@ -7680,10 +7680,10 @@ multiclass sve_int_perm_rev_rbit_z<string asm, SDPatternOperator op> {
def _S : sve_int_perm_rev_z<0b10, 0b0111, asm, ZPR32>;
def _D : sve_int_perm_rev_z<0b11, 0b0111, asm, ZPR64>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv16i8, op, nxv16i1, nxv16i8, !cast<Instruction>(NAME # _B)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv8i16, op, nxv8i1, nxv8i16, !cast<Instruction>(NAME # _H)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv4i32, op, nxv4i1, nxv4i32, !cast<Instruction>(NAME # _S)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv2i64, op, nxv2i1, nxv2i64, !cast<Instruction>(NAME # _D)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv16i8, op, nxv16i1, nxv16i8, !cast<Instruction>(NAME # _B)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv8i16, op, nxv8i1, nxv8i16, !cast<Instruction>(NAME # _H)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv4i32, op, nxv4i1, nxv4i32, !cast<Instruction>(NAME # _S)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv2i64, op, nxv2i1, nxv2i64, !cast<Instruction>(NAME # _D)>;
}
multiclass sve_int_perm_rev_revb_z<string asm, SDPatternOperator op> {
@@ -7691,37 +7691,37 @@ multiclass sve_int_perm_rev_revb_z<string asm, SDPatternOperator op> {
def _S : sve_int_perm_rev_z<0b10, 0b0100, asm, ZPR32>;
def _D : sve_int_perm_rev_z<0b11, 0b0100, asm, ZPR64>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv8i16, op, nxv8i1, nxv8i16, !cast<Instruction>(NAME # _H)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv4i32, op, nxv4i1, nxv4i32, !cast<Instruction>(NAME # _S)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv2i64, op, nxv2i1, nxv2i64, !cast<Instruction>(NAME # _D)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv8i16, op, nxv8i1, nxv8i16, !cast<Instruction>(NAME # _H)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv4i32, op, nxv4i1, nxv4i32, !cast<Instruction>(NAME # _S)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv2i64, op, nxv2i1, nxv2i64, !cast<Instruction>(NAME # _D)>;
}
multiclass sve_int_perm_rev_revh_z<string asm, SDPatternOperator op> {
def _S : sve_int_perm_rev_z<0b10, 0b0101, asm, ZPR32>;
def _D : sve_int_perm_rev_z<0b11, 0b0101, asm, ZPR64>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv4i32, op, nxv4i1, nxv4i32, !cast<Instruction>(NAME # _S)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv2i64, op, nxv2i1, nxv2i64, !cast<Instruction>(NAME # _D)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv4i32, op, nxv4i1, nxv4i32, !cast<Instruction>(NAME # _S)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv2i64, op, nxv2i1, nxv2i64, !cast<Instruction>(NAME # _D)>;
}
multiclass sve_int_perm_rev_revw_z<string asm, SDPatternOperator op> {
def _D : sve_int_perm_rev_z<0b11, 0b0110, asm, ZPR64>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv2i64, op, nxv2i1, nxv2i64, !cast<Instruction>(NAME # _D)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv2i64, op, nxv2i1, nxv2i64, !cast<Instruction>(NAME # _D)>;
}
multiclass sve_int_perm_rev_revd_z<string asm, SDPatternOperator op> {
def NAME : sve_int_perm_rev_z<0b00, 0b1110, asm, ZPR128>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv16i8, op, nxv16i1, nxv16i8, !cast<Instruction>(NAME)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv8i16, op, nxv8i1, nxv8i16, !cast<Instruction>(NAME)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv4i32, op, nxv4i1, nxv4i32, !cast<Instruction>(NAME)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv2i64, op, nxv2i1, nxv2i64, !cast<Instruction>(NAME)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv16i8, op, nxv16i1, nxv16i8, !cast<Instruction>(NAME)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv8i16, op, nxv8i1, nxv8i16, !cast<Instruction>(NAME)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv4i32, op, nxv4i1, nxv4i32, !cast<Instruction>(NAME)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv2i64, op, nxv2i1, nxv2i64, !cast<Instruction>(NAME)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv8bf16, op, nxv8i1, nxv8bf16, !cast<Instruction>(NAME)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv8f16, op, nxv8i1, nxv8f16, !cast<Instruction>(NAME)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv4f32, op, nxv4i1, nxv4f32, !cast<Instruction>(NAME)>;
- def : SVE_1_Op_PassthruUndefZero_Pat<nxv2f64, op, nxv2i1, nxv2f64, !cast<Instruction>(NAME)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv8bf16, op, nxv8i1, nxv8bf16, !cast<Instruction>(NAME)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv8f16, op, nxv8i1, nxv8f16, !cast<Instruction>(NAME)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv4f32, op, nxv4i1, nxv4f32, !cast<Instruction>(NAME)>;
+ defm : SVE_1_Op_PassthruUndefZero_Pat<nxv2f64, op, nxv2i1, nxv2f64, !cast<Instruction>(NAME)>;
}
class sve_int_perm_cpy_r<bits<2> sz8_64, string asm, ZPRRegOp zprty,
diff --git a/llvm/test/CodeGen/AArch64/zeroing-forms-rev.ll b/llvm/test/CodeGen/AArch64/zeroing-forms-rev.ll
index b7c2b1fcd99849..d7a51c8cf8062d 100644
--- a/llvm/test/CodeGen/AArch64/zeroing-forms-rev.ll
+++ b/llvm/test/CodeGen/AArch64/zeroing-forms-rev.ll
@@ -7,190 +7,6 @@
target triple = "aarch64-linux"
-define <vscale x 16 x i8> @test_svrbit_u8_x_1(<vscale x 16 x i1> %pg, <vscale x 16 x i8> %x) {
-; CHECK-LABEL: test_svrbit_u8_x_1:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: rbit z0.b, p0/m, z0.b
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrbit_u8_x_1:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: rbit z0.b, p0/z, z0.b
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.rbit.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
- ret <vscale x 16 x i8> %0
-}
-
-define <vscale x 16 x i8> @test_svrbit_u8_x_2(<vscale x 16 x i1> %pg, double %z0, <vscale x 16 x i8> %x) {
-; CHECK-LABEL: test_svrbit_u8_x_2:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: rbit z0.b, p0/m, z1.b
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrbit_u8_x_2:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: rbit z0.b, p0/z, z1.b
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.rbit.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
- ret <vscale x 16 x i8> %0
-}
-
-define <vscale x 16 x i8> @test_svrbit_u8_z(<vscale x 16 x i1> %pg, double %z0, <vscale x 16 x i8> %x) {
-; CHECK-LABEL: test_svrbit_u8_z:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: mov z0.b, #0 // =0x0
-; CHECK-NEXT: rbit z0.b, p0/m, z1.b
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrbit_u8_z:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: rbit z0.b, p0/z, z1.b
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.rbit.nxv16i8(<vscale x 16 x i8> zeroinitializer, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
- ret <vscale x 16 x i8> %0
-}
-
-define <vscale x 8 x i16> @test_svrbit_u16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %x) {
-; CHECK-LABEL: test_svrbit_u16_x_1:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: rbit z0.h, p0/m, z0.h
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrbit_u16_x_1:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: rbit z0.h, p0/z, z0.h
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.rbit.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
- ret <vscale x 8 x i16> %0
-}
-
-define <vscale x 8 x i16> @test_svrbit_u16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
-; CHECK-LABEL: test_svrbit_u16_x_2:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: rbit z0.h, p0/m, z1.h
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrbit_u16_x_2:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: rbit z0.h, p0/z, z1.h
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.rbit.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
- ret <vscale x 8 x i16> %0
-}
-
-define <vscale x 8 x i16> @test_svrbit_u16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
-; CHECK-LABEL: test_svrbit_u16_z:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: mov z0.h, #0 // =0x0
-; CHECK-NEXT: rbit z0.h, p0/m, z1.h
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrbit_u16_z:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: rbit z0.h, p0/z, z1.h
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.rbit.nxv8i16(<vscale x 8 x i16> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
- ret <vscale x 8 x i16> %0
-}
-
-define <vscale x 4 x i32> @test_svrbit_u32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %x) {
-; CHECK-LABEL: test_svrbit_u32_x_1:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: rbit z0.s, p0/m, z0.s
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrbit_u32_x_1:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: rbit z0.s, p0/z, z0.s
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.rbit.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
- ret <vscale x 4 x i32> %0
-}
-
-define <vscale x 4 x i32> @test_svrbit_u32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
-; CHECK-LABEL: test_svrbit_u32_x_2:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: rbit z0.s, p0/m, z1.s
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrbit_u32_x_2:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: rbit z0.s, p0/z, z1.s
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.rbit.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
- ret <vscale x 4 x i32> %0
-}
-
-define <vscale x 4 x i32> @test_svrbit_u32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
-; CHECK-LABEL: test_svrbit_u32_z:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: mov z0.s, #0 // =0x0
-; CHECK-NEXT: rbit z0.s, p0/m, z1.s
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrbit_u32_z:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: rbit z0.s, p0/z, z1.s
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.rbit.nxv4i32(<vscale x 4 x i32> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
- ret <vscale x 4 x i32> %0
-}
-
-define <vscale x 2 x i64> @test_svrbit_u64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
-; CHECK-LABEL: test_svrbit_u64_x_1:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: rbit z0.d, p0/m, z0.d
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrbit_u64_x_1:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: rbit z0.d, p0/z, z0.d
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.rbit.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
- ret <vscale x 2 x i64> %0
-}
-
-define <vscale x 2 x i64> @test_svrbit_u64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
-; CHECK-LABEL: test_svrbit_u64_x_2:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: rbit z0.d, p0/m, z1.d
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrbit_u64_x_2:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: rbit z0.d, p0/z, z1.d
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.rbit.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
- ret <vscale x 2 x i64> %0
-}
-
-define <vscale x 2 x i64> @test_svrbit_u64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
-; CHECK-LABEL: test_svrbit_u64_z:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: mov z0.d, #0 // =0x0
-; CHECK-NEXT: rbit z0.d, p0/m, z1.d
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrbit_u64_z:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: rbit z0.d, p0/z, z1.d
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.rbit.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
- ret <vscale x 2 x i64> %0
-}
-
define <vscale x 16 x i8> @test_svrbit_s8_x_1(<vscale x 16 x i1> %pg, <vscale x 16 x i8> %x) {
; CHECK-LABEL: test_svrbit_s8_x_1:
; CHECK: // %bb.0: // %entry
@@ -202,7 +18,7 @@ define <vscale x 16 x i8> @test_svrbit_s8_x_1(<vscale x 16 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: rbit z0.b, p0/z, z0.b
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.rbit.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.rbit.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
ret <vscale x 16 x i8> %0
}
@@ -217,7 +33,7 @@ define <vscale x 16 x i8> @test_svrbit_s8_x_2(<vscale x 16 x i1> %pg, double %z0
; CHECK-2p2-NEXT: rbit z0.b, p0/z, z1.b
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.rbit.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.rbit.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
ret <vscale x 16 x i8> %0
}
@@ -248,7 +64,7 @@ define <vscale x 8 x i16> @test_svrbit_s16_x_1(<vscale x 8 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: rbit z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.rbit.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.rbit.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x i16> %0
}
@@ -263,7 +79,7 @@ define <vscale x 8 x i16> @test_svrbit_s16_x_2(<vscale x 8 x i1> %pg, double %z0
; CHECK-2p2-NEXT: rbit z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.rbit.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.rbit.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x i16> %0
}
@@ -294,7 +110,7 @@ define <vscale x 4 x i32> @test_svrbit_s32_x_1(<vscale x 4 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: rbit z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.rbit.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.rbit.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -309,7 +125,7 @@ define <vscale x 4 x i32> @test_svrbit_s32_x_2(<vscale x 4 x i1> %pg, double %z0
; CHECK-2p2-NEXT: rbit z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.rbit.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.rbit.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -340,7 +156,7 @@ define <vscale x 2 x i64> @test_svrbit_s64_x_1(<vscale x 2 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: rbit z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.rbit.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.rbit.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -355,7 +171,7 @@ define <vscale x 2 x i64> @test_svrbit_s64_x_2(<vscale x 2 x i1> %pg, double %z0
; CHECK-2p2-NEXT: rbit z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.rbit.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.rbit.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -375,144 +191,6 @@ entry:
ret <vscale x 2 x i64> %0
}
-define <vscale x 8 x i16> @test_svrevb_u16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %x) {
-; CHECK-LABEL: test_svrevb_u16_x_1:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: revb z0.h, p0/m, z0.h
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrevb_u16_x_1:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: revb z0.h, p0/z, z0.h
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.revb.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
- ret <vscale x 8 x i16> %0
-}
-
-define <vscale x 8 x i16> @test_svrevb_u16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
-; CHECK-LABEL: test_svrevb_u16_x_2:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: revb z0.h, p0/m, z1.h
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrevb_u16_x_2:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: revb z0.h, p0/z, z1.h
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.revb.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
- ret <vscale x 8 x i16> %0
-}
-
-define <vscale x 8 x i16> @test_svrevb_u16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
-; CHECK-LABEL: test_svrevb_u16_z:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: mov z0.h, #0 // =0x0
-; CHECK-NEXT: revb z0.h, p0/m, z1.h
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrevb_u16_z:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: revb z0.h, p0/z, z1.h
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.revb.nxv8i16(<vscale x 8 x i16> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
- ret <vscale x 8 x i16> %0
-}
-
-define <vscale x 4 x i32> @test_svrevb_u32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %x) {
-; CHECK-LABEL: test_svrevb_u32_x_1:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: revb z0.s, p0/m, z0.s
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrevb_u32_x_1:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: revb z0.s, p0/z, z0.s
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revb.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
- ret <vscale x 4 x i32> %0
-}
-
-define <vscale x 4 x i32> @test_svrevb_u32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
-; CHECK-LABEL: test_svrevb_u32_x_2:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: revb z0.s, p0/m, z1.s
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrevb_u32_x_2:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: revb z0.s, p0/z, z1.s
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revb.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
- ret <vscale x 4 x i32> %0
-}
-
-define <vscale x 4 x i32> @test_svrevb_u32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
-; CHECK-LABEL: test_svrevb_u32_z:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: mov z0.s, #0 // =0x0
-; CHECK-NEXT: revb z0.s, p0/m, z1.s
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrevb_u32_z:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: revb z0.s, p0/z, z1.s
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revb.nxv4i32(<vscale x 4 x i32> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
- ret <vscale x 4 x i32> %0
-}
-
-define <vscale x 2 x i64> @test_svrevb_u64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
-; CHECK-LABEL: test_svrevb_u64_x_1:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: revb z0.d, p0/m, z0.d
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrevb_u64_x_1:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: revb z0.d, p0/z, z0.d
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revb.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
- ret <vscale x 2 x i64> %0
-}
-
-define <vscale x 2 x i64> @test_svrevb_u64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
-; CHECK-LABEL: test_svrevb_u64_x_2:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: revb z0.d, p0/m, z1.d
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrevb_u64_x_2:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: revb z0.d, p0/z, z1.d
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revb.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
- ret <vscale x 2 x i64> %0
-}
-
-define <vscale x 2 x i64> @test_svrevb_u64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
-; CHECK-LABEL: test_svrevb_u64_z:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: mov z0.d, #0 // =0x0
-; CHECK-NEXT: revb z0.d, p0/m, z1.d
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrevb_u64_z:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: revb z0.d, p0/z, z1.d
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revb.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
- ret <vscale x 2 x i64> %0
-}
-
define <vscale x 8 x i16> @test_svrevb_s16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %x) {
; CHECK-LABEL: test_svrevb_s16_x_1:
; CHECK: // %bb.0: // %entry
@@ -524,7 +202,7 @@ define <vscale x 8 x i16> @test_svrevb_s16_x_1(<vscale x 8 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: revb z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.revb.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.revb.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x i16> %0
}
@@ -539,7 +217,7 @@ define <vscale x 8 x i16> @test_svrevb_s16_x_2(<vscale x 8 x i1> %pg, double %z0
; CHECK-2p2-NEXT: revb z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.revb.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.revb.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x i16> %0
}
@@ -570,7 +248,7 @@ define <vscale x 4 x i32> @test_svrevb_s32_x_1(<vscale x 4 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: revb z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revb.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revb.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -585,7 +263,7 @@ define <vscale x 4 x i32> @test_svrevb_s32_x_2(<vscale x 4 x i1> %pg, double %z0
; CHECK-2p2-NEXT: revb z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revb.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revb.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -616,7 +294,7 @@ define <vscale x 2 x i64> @test_svrevb_s64_x_1(<vscale x 2 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: revb z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revb.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revb.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -631,7 +309,7 @@ define <vscale x 2 x i64> @test_svrevb_s64_x_2(<vscale x 2 x i1> %pg, double %z0
; CHECK-2p2-NEXT: revb z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revb.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revb.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -651,44 +329,44 @@ entry:
ret <vscale x 2 x i64> %0
}
-define <vscale x 4 x i32> @test_svrevh_u32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %x) {
-; CHECK-LABEL: test_svrevh_u32_x_1:
+define <vscale x 4 x i32> @test_svrevh_s32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrevh_s32_x_1:
; CHECK: // %bb.0: // %entry
; CHECK-NEXT: revh z0.s, p0/m, z0.s
; CHECK-NEXT: ret
;
-; CHECK-2p2-LABEL: test_svrevh_u32_x_1:
+; CHECK-2p2-LABEL: test_svrevh_s32_x_1:
; CHECK-2p2: // %bb.0: // %entry
; CHECK-2p2-NEXT: revh z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revh.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revh.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
-define <vscale x 4 x i32> @test_svrevh_u32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
-; CHECK-LABEL: test_svrevh_u32_x_2:
+define <vscale x 4 x i32> @test_svrevh_s32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrevh_s32_x_2:
; CHECK: // %bb.0: // %entry
; CHECK-NEXT: revh z0.s, p0/m, z1.s
; CHECK-NEXT: ret
;
-; CHECK-2p2-LABEL: test_svrevh_u32_x_2:
+; CHECK-2p2-LABEL: test_svrevh_s32_x_2:
; CHECK-2p2: // %bb.0: // %entry
; CHECK-2p2-NEXT: revh z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revh.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revh.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
-define <vscale x 4 x i32> @test_svrevh_u32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
-; CHECK-LABEL: test_svrevh_u32_z:
+define <vscale x 4 x i32> @test_svrevh_s32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrevh_s32_z:
; CHECK: // %bb.0: // %entry
; CHECK-NEXT: mov z0.s, #0 // =0x0
; CHECK-NEXT: revh z0.s, p0/m, z1.s
; CHECK-NEXT: ret
;
-; CHECK-2p2-LABEL: test_svrevh_u32_z:
+; CHECK-2p2-LABEL: test_svrevh_s32_z:
; CHECK-2p2: // %bb.0: // %entry
; CHECK-2p2-NEXT: revh z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
@@ -697,100 +375,8 @@ entry:
ret <vscale x 4 x i32> %0
}
-define <vscale x 2 x i64> @test_svrevh_u64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
-; CHECK-LABEL: test_svrevh_u64_x_1:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: revh z0.d, p0/m, z0.d
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrevh_u64_x_1:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: revh z0.d, p0/z, z0.d
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revh.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
- ret <vscale x 2 x i64> %0
-}
-
-define <vscale x 2 x i64> @test_svrevh_u64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
-; CHECK-LABEL: test_svrevh_u64_x_2:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: revh z0.d, p0/m, z1.d
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrevh_u64_x_2:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: revh z0.d, p0/z, z1.d
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revh.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
- ret <vscale x 2 x i64> %0
-}
-
-define <vscale x 2 x i64> @test_svrevh_u64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
-; CHECK-LABEL: test_svrevh_u64_z:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: mov z0.d, #0 // =0x0
-; CHECK-NEXT: revh z0.d, p0/m, z1.d
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrevh_u64_z:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: revh z0.d, p0/z, z1.d
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revh.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
- ret <vscale x 2 x i64> %0
-}
-
-define <vscale x 4 x i32> @test_svrevh_s32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %x) {
-; CHECK-LABEL: test_svrevh_s32_x_1:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: revh z0.s, p0/m, z0.s
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrevh_s32_x_1:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: revh z0.s, p0/z, z0.s
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revh.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
- ret <vscale x 4 x i32> %0
-}
-
-define <vscale x 4 x i32> @test_svrevh_s32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
-; CHECK-LABEL: test_svrevh_s32_x_2:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: revh z0.s, p0/m, z1.s
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrevh_s32_x_2:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: revh z0.s, p0/z, z1.s
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revh.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
- ret <vscale x 4 x i32> %0
-}
-
-define <vscale x 4 x i32> @test_svrevh_s32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
-; CHECK-LABEL: test_svrevh_s32_z:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: mov z0.s, #0 // =0x0
-; CHECK-NEXT: revh z0.s, p0/m, z1.s
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrevh_s32_z:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: revh z0.s, p0/z, z1.s
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revh.nxv4i32(<vscale x 4 x i32> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
- ret <vscale x 4 x i32> %0
-}
-
-define <vscale x 2 x i64> @test_svrevh_s64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
-; CHECK-LABEL: test_svrevh_s64_x_1:
+define <vscale x 2 x i64> @test_svrevh_s64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrevh_s64_x_1:
; CHECK: // %bb.0: // %entry
; CHECK-NEXT: revh z0.d, p0/m, z0.d
; CHECK-NEXT: ret
@@ -800,7 +386,7 @@ define <vscale x 2 x i64> @test_svrevh_s64_x_1(<vscale x 2 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: revh z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revh.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revh.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -815,7 +401,7 @@ define <vscale x 2 x i64> @test_svrevh_s64_x_2(<vscale x 2 x i1> %pg, double %z0
; CHECK-2p2-NEXT: revh z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revh.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revh.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -835,52 +421,6 @@ entry:
ret <vscale x 2 x i64> %0
}
-define <vscale x 2 x i64> @test_svrevw_u64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
-; CHECK-LABEL: test_svrevw_u64_x_1:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: revw z0.d, p0/m, z0.d
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrevw_u64_x_1:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: revw z0.d, p0/z, z0.d
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revw.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
- ret <vscale x 2 x i64> %0
-}
-
-define <vscale x 2 x i64> @test_svrevw_u64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
-; CHECK-LABEL: test_svrevw_u64_x_2:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: revw z0.d, p0/m, z1.d
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrevw_u64_x_2:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: revw z0.d, p0/z, z1.d
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revw.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
- ret <vscale x 2 x i64> %0
-}
-
-define <vscale x 2 x i64> @test_svrevw_u64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
-; CHECK-LABEL: test_svrevw_u64_z:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: mov z0.d, #0 // =0x0
-; CHECK-NEXT: revw z0.d, p0/m, z1.d
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrevw_u64_z:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: revw z0.d, p0/z, z1.d
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revw.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
- ret <vscale x 2 x i64> %0
-}
-
define <vscale x 2 x i64> @test_svrevw_s64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
; CHECK-LABEL: test_svrevw_s64_x_1:
; CHECK: // %bb.0: // %entry
@@ -892,7 +432,7 @@ define <vscale x 2 x i64> @test_svrevw_s64_x_1(<vscale x 2 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: revw z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revw.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revw.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -907,7 +447,7 @@ define <vscale x 2 x i64> @test_svrevw_s64_x_2(<vscale x 2 x i1> %pg, double %z0
; CHECK-2p2-NEXT: revw z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revw.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revw.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -927,190 +467,6 @@ entry:
ret <vscale x 2 x i64> %0
}
-define <vscale x 16 x i8> @test_svrevd_u8_x_1(<vscale x 16 x i1> %pg, <vscale x 16 x i8> %x) {
-; CHECK-LABEL: test_svrevd_u8_x_1:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: revd z0.q, p0/m, z0.q
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrevd_u8_x_1:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: revd z0.q, p0/z, z0.q
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.revd.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
- ret <vscale x 16 x i8> %0
-}
-
-define <vscale x 16 x i8> @test_svrevd_u8_x_2(<vscale x 16 x i1> %pg, double %z0, <vscale x 16 x i8> %x) {
-; CHECK-LABEL: test_svrevd_u8_x_2:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: revd z0.q, p0/m, z1.q
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrevd_u8_x_2:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.revd.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
- ret <vscale x 16 x i8> %0
-}
-
-define <vscale x 16 x i8> @test_svrevd_u8_z(<vscale x 16 x i1> %pg, double %z0, <vscale x 16 x i8> %x) {
-; CHECK-LABEL: test_svrevd_u8_z:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: mov z0.b, #0 // =0x0
-; CHECK-NEXT: revd z0.q, p0/m, z1.q
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrevd_u8_z:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.revd.nxv16i8(<vscale x 16 x i8> zeroinitializer, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
- ret <vscale x 16 x i8> %0
-}
-
-define <vscale x 8 x i16> @test_svrevd_u16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %x) {
-; CHECK-LABEL: test_svrevd_u16_x_1:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: revd z0.q, p0/m, z0.q
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrevd_u16_x_1:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: revd z0.q, p0/z, z0.q
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.revd.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
- ret <vscale x 8 x i16> %0
-}
-
-define <vscale x 8 x i16> @test_svrevd_u16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
-; CHECK-LABEL: test_svrevd_u16_x_2:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: revd z0.q, p0/m, z1.q
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrevd_u16_x_2:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.revd.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
- ret <vscale x 8 x i16> %0
-}
-
-define <vscale x 8 x i16> @test_svrevd_u16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
-; CHECK-LABEL: test_svrevd_u16_z:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: mov z0.h, #0 // =0x0
-; CHECK-NEXT: revd z0.q, p0/m, z1.q
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrevd_u16_z:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.revd.nxv8i16(<vscale x 8 x i16> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
- ret <vscale x 8 x i16> %0
-}
-
-define <vscale x 4 x i32> @test_svrevd_u32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %x) {
-; CHECK-LABEL: test_svrevd_u32_x_1:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: revd z0.q, p0/m, z0.q
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrevd_u32_x_1:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: revd z0.q, p0/z, z0.q
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revd.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
- ret <vscale x 4 x i32> %0
-}
-
-define <vscale x 4 x i32> @test_svrevd_u32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
-; CHECK-LABEL: test_svrevd_u32_x_2:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: revd z0.q, p0/m, z1.q
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrevd_u32_x_2:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revd.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
- ret <vscale x 4 x i32> %0
-}
-
-define <vscale x 4 x i32> @test_svrevd_u32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
-; CHECK-LABEL: test_svrevd_u32_z:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: mov z0.s, #0 // =0x0
-; CHECK-NEXT: revd z0.q, p0/m, z1.q
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrevd_u32_z:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revd.nxv4i32(<vscale x 4 x i32> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
- ret <vscale x 4 x i32> %0
-}
-
-define <vscale x 2 x i64> @test_svrevd_u64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
-; CHECK-LABEL: test_svrevd_u64_x_1:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: revd z0.q, p0/m, z0.q
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrevd_u64_x_1:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: revd z0.q, p0/z, z0.q
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revd.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
- ret <vscale x 2 x i64> %0
-}
-
-define <vscale x 2 x i64> @test_svrevd_u64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
-; CHECK-LABEL: test_svrevd_u64_x_2:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: revd z0.q, p0/m, z1.q
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrevd_u64_x_2:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revd.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
- ret <vscale x 2 x i64> %0
-}
-
-define <vscale x 2 x i64> @test_svrevd_u64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
-; CHECK-LABEL: test_svrevd_u64_z:
-; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: mov z0.d, #0 // =0x0
-; CHECK-NEXT: revd z0.q, p0/m, z1.q
-; CHECK-NEXT: ret
-;
-; CHECK-2p2-LABEL: test_svrevd_u64_z:
-; CHECK-2p2: // %bb.0: // %entry
-; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
-; CHECK-2p2-NEXT: ret
-entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revd.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
- ret <vscale x 2 x i64> %0
-}
-
define <vscale x 16 x i8> @test_svrevd_s8_x_1(<vscale x 16 x i1> %pg, <vscale x 16 x i8> %x) {
; CHECK-LABEL: test_svrevd_s8_x_1:
; CHECK: // %bb.0: // %entry
@@ -1122,7 +478,7 @@ define <vscale x 16 x i8> @test_svrevd_s8_x_1(<vscale x 16 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: revd z0.q, p0/z, z0.q
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.revd.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.revd.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
ret <vscale x 16 x i8> %0
}
@@ -1137,7 +493,7 @@ define <vscale x 16 x i8> @test_svrevd_s8_x_2(<vscale x 16 x i1> %pg, double %z0
; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.revd.nxv16i8(<vscale x 16 x i8> undef, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.revd.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
ret <vscale x 16 x i8> %0
}
@@ -1168,7 +524,7 @@ define <vscale x 8 x i16> @test_svrevd_s16_x_1(<vscale x 8 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: revd z0.q, p0/z, z0.q
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.revd.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.revd.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x i16> %0
}
@@ -1183,7 +539,7 @@ define <vscale x 8 x i16> @test_svrevd_s16_x_2(<vscale x 8 x i1> %pg, double %z0
; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.revd.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.revd.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x i16> %0
}
@@ -1214,7 +570,7 @@ define <vscale x 4 x i32> @test_svrevd_s32_x_1(<vscale x 4 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: revd z0.q, p0/z, z0.q
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revd.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revd.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -1229,7 +585,7 @@ define <vscale x 4 x i32> @test_svrevd_s32_x_2(<vscale x 4 x i1> %pg, double %z0
; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revd.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revd.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -1260,7 +616,7 @@ define <vscale x 2 x i64> @test_svrevd_s64_x_1(<vscale x 2 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: revd z0.q, p0/z, z0.q
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revd.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revd.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -1275,7 +631,7 @@ define <vscale x 2 x i64> @test_svrevd_s64_x_2(<vscale x 2 x i1> %pg, double %z0
; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revd.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revd.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -1306,7 +662,7 @@ define <vscale x 8 x half> @test_svrevd_f16_x_1(<vscale x 8 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: revd z0.q, p0/z, z0.q
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.revd.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.revd.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x half> %0
}
@@ -1321,7 +677,7 @@ define <vscale x 8 x half> @test_svrevd_f16_x_2(<vscale x 8 x i1> %pg, double %z
; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.revd.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.revd.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
ret <vscale x 8 x half> %0
}
@@ -1352,7 +708,7 @@ define <vscale x 8 x bfloat> @test_svrevd_bf16_x_1(<vscale x 8 x i1> %pg, <vscal
; CHECK-2p2-NEXT: revd z0.q, p0/z, z0.q
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x bfloat> @llvm.aarch64.sve.revd.nxv8bf16(<vscale x 8 x bfloat> undef, <vscale x 8 x i1> %pg, <vscale x 8 x bfloat> %x)
+ %0 = tail call <vscale x 8 x bfloat> @llvm.aarch64.sve.revd.nxv8bf16(<vscale x 8 x bfloat> poison, <vscale x 8 x i1> %pg, <vscale x 8 x bfloat> %x)
ret <vscale x 8 x bfloat> %0
}
@@ -1367,7 +723,7 @@ define <vscale x 8 x bfloat> @test_svrevd_bf16_x_2(<vscale x 8 x i1> %pg, double
; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x bfloat> @llvm.aarch64.sve.revd.nxv8bf16(<vscale x 8 x bfloat> undef, <vscale x 8 x i1> %pg, <vscale x 8 x bfloat> %x)
+ %0 = tail call <vscale x 8 x bfloat> @llvm.aarch64.sve.revd.nxv8bf16(<vscale x 8 x bfloat> poison, <vscale x 8 x i1> %pg, <vscale x 8 x bfloat> %x)
ret <vscale x 8 x bfloat> %0
}
@@ -1398,7 +754,7 @@ define <vscale x 4 x float> @test_svrevd_f32_x_1(<vscale x 4 x i1> %pg, <vscale
; CHECK-2p2-NEXT: revd z0.q, p0/z, z0.q
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.revd.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.revd.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x float> %0
}
@@ -1413,7 +769,7 @@ define <vscale x 4 x float> @test_svrevd_f32_x_2(<vscale x 4 x i1> %pg, double %
; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.revd.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.revd.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
ret <vscale x 4 x float> %0
}
@@ -1444,7 +800,7 @@ define <vscale x 2 x double> @test_svrevd_f64_x_1(<vscale x 2 x i1> %pg, <vscale
; CHECK-2p2-NEXT: revd z0.q, p0/z, z0.q
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.revd.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.revd.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x double> %0
}
@@ -1459,7 +815,7 @@ define <vscale x 2 x double> @test_svrevd_f64_x_2(<vscale x 2 x i1> %pg, double
; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.revd.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.revd.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x double> %0
}
@@ -1478,3 +834,669 @@ entry:
%0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.revd.nxv2f64(<vscale x 2 x double> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
ret <vscale x 2 x double> %0
}
+
+define <vscale x 16 x i8> @test_svrbit_nxv16i8_ptrue_u(double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svrbit_nxv16i8_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.b
+; CHECK-NEXT: rbit z0.b, p0/m, z1.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_nxv16i8_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.b
+; CHECK-2p2-NEXT: rbit z0.b, p0/z, z1.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.rbit.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svrbit_nxv16i8_ptrue(double %z0, <vscale x 16 x i8> %x, <vscale x 16 x i8> %y) {
+; CHECK-LABEL: test_svrbit_nxv16i8_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.b
+; CHECK-NEXT: rbit z0.b, p0/m, z2.b
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_nxv16i8_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.b
+; CHECK-2p2-NEXT: rbit z0.b, p0/z, z2.b
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.rbit.nxv16i8(<vscale x 16 x i8> %x, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %y)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 8 x i16> @test_svrbit_nxv8i16_ptrue_u(double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svrbit_nxv8i16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: rbit z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_nxv8i16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: rbit z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.rbit.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svrbit_nxv8i16_ptrue(double %z0, <vscale x 8 x i16> %x, <vscale x 8 x i16> %y) {
+; CHECK-LABEL: test_svrbit_nxv8i16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: rbit z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_nxv8i16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: rbit z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.rbit.nxv8i16(<vscale x 8 x i16> %x, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %y)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svrbit_nxv4i32_ptrue_u(double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrbit_nxv4i32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: rbit z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_nxv4i32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: rbit z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.rbit.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svrbit_nxv4i32_ptrue(double %z0, <vscale x 4 x i32> %x, <vscale x 4 x i32> %y) {
+; CHECK-LABEL: test_svrbit_nxv4i32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: rbit z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_nxv4i32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: rbit z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.rbit.nxv4i32(<vscale x 4 x i32> %x, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %y)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svrbit_nxv2i64_ptrue_u(double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrbit_nxv2i64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: rbit z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_nxv2i64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: rbit z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.rbit.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svrbit_nxv2i64_ptrue(double %z0, <vscale x 2 x i64> %x, <vscale x 2 x i64> %y) {
+; CHECK-LABEL: test_svrbit_nxv2i64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: rbit z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrbit_nxv2i64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: rbit z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.rbit.nxv2i64(<vscale x 2 x i64> %x, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %y)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 8 x i16> @test_svrevb_nxv8i16_ptrue_u(double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svrevb_nxv8i16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: revb z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevb_nxv8i16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: revb z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.revb.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svrevb_nxv8i16_ptrue(double %z0, <vscale x 8 x i16> %x, <vscale x 8 x i16> %y) {
+; CHECK-LABEL: test_svrevb_nxv8i16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: revb z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevb_nxv8i16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: revb z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.revb.nxv8i16(<vscale x 8 x i16> %x, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %y)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svrevb_nxv4i32_ptrue_u(double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrevb_nxv4i32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: revb z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevb_nxv4i32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: revb z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revb.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svrevb_nxv4i32_ptrue(double %z0, <vscale x 4 x i32> %x, <vscale x 4 x i32> %y) {
+; CHECK-LABEL: test_svrevb_nxv4i32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: revb z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevb_nxv4i32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: revb z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revb.nxv4i32(<vscale x 4 x i32> %x, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %y)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svrevb_nxv2i64_ptrue_u(double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrevb_nxv2i64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: revb z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevb_nxv2i64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: revb z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revb.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svrevb_nxv2i64_ptrue(double %z0, <vscale x 2 x i64> %x, <vscale x 2 x i64> %y) {
+; CHECK-LABEL: test_svrevb_nxv2i64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: revb z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevb_nxv2i64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: revb z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revb.nxv2i64(<vscale x 2 x i64> %x, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %y)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 4 x i32> @test_svrevh_nxv4i32_ptrue_u(double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrevh_nxv4i32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: revh z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevh_nxv4i32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: revh z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revh.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svrevh_nxv4i32_ptrue(double %z0, <vscale x 4 x i32> %x, <vscale x 4 x i32> %y) {
+; CHECK-LABEL: test_svrevh_nxv4i32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: revh z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevh_nxv4i32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: revh z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revh.nxv4i32(<vscale x 4 x i32> %x, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %y)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svrevh_nxv2i64_ptrue_u(double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrevh_nxv2i64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: revh z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevh_nxv2i64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: revh z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revh.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svrevh_nxv2i64_ptrue(double %z0, <vscale x 2 x i64> %x, <vscale x 2 x i64> %y) {
+; CHECK-LABEL: test_svrevh_nxv2i64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: revh z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevh_nxv2i64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: revh z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revh.nxv2i64(<vscale x 2 x i64> %x, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %y)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svrevw_nxv2i64_ptrue_u(double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrevw_nxv2i64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: revw z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevw_nxv2i64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: revw z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revw.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svrevw_nxv2i64_ptrue(double %z0, <vscale x 2 x i64> %x, <vscale x 2 x i64> %y) {
+; CHECK-LABEL: test_svrevw_nxv2i64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: revw z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevw_nxv2i64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: revw z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revw.nxv2i64(<vscale x 2 x i64> %x, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %y)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 16 x i8> @test_svrevd_nxv16i8_ptrue_u(double %z0, <vscale x 16 x i8> %x) {
+; CHECK-LABEL: test_svrevd_nxv16i8_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.b
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_nxv16i8_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.b
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.revd.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %x)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 16 x i8> @test_svrevd_nxv16i8_ptrue(double %z0, <vscale x 16 x i8> %x, <vscale x 16 x i8> %y) {
+; CHECK-LABEL: test_svrevd_nxv16i8_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.b
+; CHECK-NEXT: revd z0.q, p0/m, z2.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_nxv16i8_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.b
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z2.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
+ %0 = tail call <vscale x 16 x i8> @llvm.aarch64.sve.revd.nxv16i8(<vscale x 16 x i8> %x, <vscale x 16 x i1> %pg, <vscale x 16 x i8> %y)
+ ret <vscale x 16 x i8> %0
+}
+
+define <vscale x 8 x i16> @test_svrevd_nxv8i16_ptrue_u(double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svrevd_nxv8i16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_nxv8i16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.revd.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svrevd_nxv8i16_ptrue(double %z0, <vscale x 8 x i16> %x, <vscale x 8 x i16> %y) {
+; CHECK-LABEL: test_svrevd_nxv8i16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: revd z0.q, p0/m, z2.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_nxv8i16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z2.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.revd.nxv8i16(<vscale x 8 x i16> %x, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %y)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svrevd_nxv4i32_ptrue_u(double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svrevd_nxv4i32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_nxv4i32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revd.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svrevd_nxv4i32_ptrue(double %z0, <vscale x 4 x i32> %x, <vscale x 4 x i32> %y) {
+; CHECK-LABEL: test_svrevd_nxv4i32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: revd z0.q, p0/m, z2.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_nxv4i32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z2.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.revd.nxv4i32(<vscale x 4 x i32> %x, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %y)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svrevd_nxv2i64_ptrue_u(double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svrevd_nxv2i64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_nxv2i64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revd.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svrevd_nxv2i64_ptrue(double %z0, <vscale x 2 x i64> %x, <vscale x 2 x i64> %y) {
+; CHECK-LABEL: test_svrevd_nxv2i64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: revd z0.q, p0/m, z2.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_nxv2i64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z2.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.revd.nxv2i64(<vscale x 2 x i64> %x, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %y)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 8 x half> @test_svrevd_nxv8f16_ptrue_u(double %z0, <vscale x 8 x half> %x) {
+; CHECK-LABEL: test_svrevd_nxv8f16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_nxv8f16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.revd.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> %pg, <vscale x 8 x half> %x)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x half> @test_svrevd_nxv8f16_ptrue(double %z0, <vscale x 8 x half> %x, <vscale x 8 x half> %y) {
+; CHECK-LABEL: test_svrevd_nxv8f16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: revd z0.q, p0/m, z2.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_nxv8f16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z2.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x half> @llvm.aarch64.sve.revd.nxv8f16(<vscale x 8 x half> %x, <vscale x 8 x i1> %pg, <vscale x 8 x half> %y)
+ ret <vscale x 8 x half> %0
+}
+
+define <vscale x 8 x bfloat> @test_svrevd_nxv8bf16_ptrue_u(double %z0, <vscale x 8 x bfloat> %x) {
+; CHECK-LABEL: test_svrevd_nxv8bf16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_nxv8bf16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x bfloat> @llvm.aarch64.sve.revd.nxv8bf16(<vscale x 8 x bfloat> poison, <vscale x 8 x i1> %pg, <vscale x 8 x bfloat> %x)
+ ret <vscale x 8 x bfloat> %0
+}
+
+define <vscale x 8 x bfloat> @test_svrevd_nxv8bf16_ptrue(double %z0, <vscale x 8 x bfloat> %x, <vscale x 8 x bfloat> %y) {
+; CHECK-LABEL: test_svrevd_nxv8bf16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: revd z0.q, p0/m, z2.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_nxv8bf16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z2.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x bfloat> @llvm.aarch64.sve.revd.nxv8bf16(<vscale x 8 x bfloat> %x, <vscale x 8 x i1> %pg, <vscale x 8 x bfloat> %y)
+ ret <vscale x 8 x bfloat> %0
+}
+
+define <vscale x 4 x float> @test_svrevd_nxv4f32_ptrue_u(double %z0, <vscale x 4 x float> %x) {
+; CHECK-LABEL: test_svrevd_nxv4f32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_nxv4f32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.revd.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> %pg, <vscale x 4 x float> %x)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 4 x float> @test_svrevd_nxv4f32_ptrue(double %z0, <vscale x 4 x float> %x, <vscale x 4 x float> %y) {
+; CHECK-LABEL: test_svrevd_nxv4f32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: revd z0.q, p0/m, z2.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_nxv4f32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z2.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x float> @llvm.aarch64.sve.revd.nxv4f32(<vscale x 4 x float> %x, <vscale x 4 x i1> %pg, <vscale x 4 x float> %y)
+ ret <vscale x 4 x float> %0
+}
+
+define <vscale x 2 x double> @test_svrevd_nxv2f64_ptrue_u(double %z0, <vscale x 2 x double> %x) {
+; CHECK-LABEL: test_svrevd_nxv2f64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: revd z0.q, p0/m, z1.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_nxv2f64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z1.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.revd.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> %pg, <vscale x 2 x double> %x)
+ ret <vscale x 2 x double> %0
+}
+
+define <vscale x 2 x double> @test_svrevd_nxv2f64_ptrue(double %z0, <vscale x 2 x double> %x, <vscale x 2 x double> %y) {
+; CHECK-LABEL: test_svrevd_nxv2f64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, z1.d
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: revd z0.q, p0/m, z2.q
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svrevd_nxv2f64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: revd z0.q, p0/z, z2.q
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x double> @llvm.aarch64.sve.revd.nxv2f64(<vscale x 2 x double> %x, <vscale x 2 x i1> %pg, <vscale x 2 x double> %y)
+ ret <vscale x 2 x double> %0
+}
>From 589d57325a3ec732a8948926dd32177cdd3bdf54 Mon Sep 17 00:00:00 2001
From: Momchil Velikov <momchil.velikov at arm.com>
Date: Tue, 19 Nov 2024 14:49:38 +0000
Subject: [PATCH 15/16] [AArch64] Generate zeroing forms of certain SVE2.2
instructions (11/11)
---
.../lib/Target/AArch64/AArch64SVEInstrInfo.td | 12 +-
llvm/lib/Target/AArch64/SVEInstrFormats.td | 22 +-
.../test/CodeGen/AArch64/zeroing-forms-ext.ll | 572 ++++++++++++++++++
3 files changed, 598 insertions(+), 8 deletions(-)
create mode 100644 llvm/test/CodeGen/AArch64/zeroing-forms-ext.ll
diff --git a/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td b/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
index fc84e8056f496e..0233c8d1124d5c 100644
--- a/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
+++ b/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
@@ -4325,14 +4325,14 @@ let Predicates = [HasSVE2p2orSME2p2] in {
defm FNEG_ZPzZ : sve_int_un_pred_arit_bitwise_fp_z<0b101, "fneg", AArch64fneg_mt>;
// SVE2p2 integer unary arithmetic, zeroing predicate
- defm SXTB_ZPzZ : sve_int_un_pred_arit_h_z<0b000, "sxtb">;
- defm UXTB_ZPzZ : sve_int_un_pred_arit_h_z<0b001, "uxtb">;
- defm SXTH_ZPzZ : sve_int_un_pred_arit_w_z<0b010, "sxth">;
- defm UXTH_ZPzZ : sve_int_un_pred_arit_w_z<0b011, "uxth">;
+ defm SXTB_ZPzZ : sve_int_un_pred_arit_h_z<0b000, "sxtb", AArch64sxt_mt>;
+ defm UXTB_ZPzZ : sve_int_un_pred_arit_h_z<0b001, "uxtb", AArch64uxt_mt>;
+ defm SXTH_ZPzZ : sve_int_un_pred_arit_w_z<0b010, "sxth", AArch64sxt_mt>;
+ defm UXTH_ZPzZ : sve_int_un_pred_arit_w_z<0b011, "uxth", AArch64uxt_mt>;
defm ABS_ZPzZ : sve_int_un_pred_arit_z< 0b110, "abs", AArch64abs_mt>;
defm NEG_ZPzZ : sve_int_un_pred_arit_z< 0b111, "neg", AArch64neg_mt>;
- def SXTW_ZPzZ_D : sve_int_un_pred_arit_z<0b11, 0b1000, "sxtw", ZPR64>;
- def UXTW_ZPzZ_D : sve_int_un_pred_arit_z<0b11, 0b1010, "uxtw", ZPR64>;
+ defm SXTW_ZPzZ : sve_int_un_pred_arit_d_z<0b100, "sxtw", AArch64sxt_mt>;
+ defm UXTW_ZPzZ : sve_int_un_pred_arit_d_z<0b101, "uxtw", AArch64uxt_mt>;
// SVE predicate count
defm FIRSTP_XPP : sve_int_pcount_pred_tmp<0b001, "firstp">;
diff --git a/llvm/lib/Target/AArch64/SVEInstrFormats.td b/llvm/lib/Target/AArch64/SVEInstrFormats.td
index b2a245586c6e32..41db627e76ce69 100644
--- a/llvm/lib/Target/AArch64/SVEInstrFormats.td
+++ b/llvm/lib/Target/AArch64/SVEInstrFormats.td
@@ -662,6 +662,11 @@ multiclass SVE_InReg_Extend_PassthruUndef<ValueType vt, SDPatternOperator op, Va
(inst $PassThru, $Pg, $Src)>;
}
+class SVE_InReg_Extend_PassthruUndefZero<ValueType vt, SDPatternOperator op, ValueType pt,
+ ValueType inreg_vt, Instruction inst>
+ : Pat<(vt (op pt:$Pg, vt:$Src, inreg_vt, (vt (SVEDup0Undef)))),
+ (inst $Pg, $Src)>;
+
class SVE_Shift_DupImm_Pred_Pat<ValueType vt, SDPatternOperator op,
ValueType pt, ValueType it,
ComplexPattern cast, Instruction inst>
@@ -4947,10 +4952,14 @@ multiclass sve_int_un_pred_arit_h<bits<3> opc, string asm,
defm : SVE_InReg_Extend_PassthruUndef<nxv2i64, op, nxv2i1, nxv2i8, !cast<Pseudo>(NAME # _D_UNDEF)>;
}
-multiclass sve_int_un_pred_arit_h_z<bits<3> opc, string asm> {
+multiclass sve_int_un_pred_arit_h_z<bits<3> opc, string asm, SDPatternOperator op> {
def _H : sve_int_un_pred_arit_z<0b01, { opc, 0b0 }, asm, ZPR16>;
def _S : sve_int_un_pred_arit_z<0b10, { opc, 0b0 }, asm, ZPR32>;
def _D : sve_int_un_pred_arit_z<0b11, { opc, 0b0 }, asm, ZPR64>;
+
+ def : SVE_InReg_Extend_PassthruUndefZero<nxv8i16, op, nxv8i1, nxv8i8, !cast<Instruction>(NAME # _H)>;
+ def : SVE_InReg_Extend_PassthruUndefZero<nxv4i32, op, nxv4i1, nxv4i8, !cast<Instruction>(NAME # _S)>;
+ def : SVE_InReg_Extend_PassthruUndefZero<nxv2i64, op, nxv2i1, nxv2i8, !cast<Instruction>(NAME # _D)>;
}
multiclass sve_int_un_pred_arit_w<bits<3> opc, string asm,
@@ -4970,9 +4979,12 @@ multiclass sve_int_un_pred_arit_w<bits<3> opc, string asm,
defm : SVE_InReg_Extend_PassthruUndef<nxv2i64, op, nxv2i1, nxv2i16, !cast<Pseudo>(NAME # _D_UNDEF)>;
}
-multiclass sve_int_un_pred_arit_w_z<bits<3> opc, string asm> {
+multiclass sve_int_un_pred_arit_w_z<bits<3> opc, string asm, SDPatternOperator op> {
def _S : sve_int_un_pred_arit_z<0b10, { opc, 0b0 }, asm, ZPR32>;
def _D : sve_int_un_pred_arit_z<0b11, { opc, 0b0 }, asm, ZPR64>;
+
+ def : SVE_InReg_Extend_PassthruUndefZero<nxv4i32, op, nxv4i1, nxv4i16, !cast<Instruction>(NAME # _S)>;
+ def : SVE_InReg_Extend_PassthruUndefZero<nxv2i64, op, nxv2i1, nxv2i16, !cast<Instruction>(NAME # _D)>;
}
multiclass sve_int_un_pred_arit_d<bits<3> opc, string asm,
@@ -4987,6 +4999,12 @@ multiclass sve_int_un_pred_arit_d<bits<3> opc, string asm,
defm : SVE_InReg_Extend_PassthruUndef<nxv2i64, op, nxv2i1, nxv2i32, !cast<Pseudo>(NAME # _D_UNDEF)>;
}
+multiclass sve_int_un_pred_arit_d_z<bits<3> opc, string asm, SDPatternOperator op> {
+ def _D : sve_int_un_pred_arit_z<0b11, {opc, 0b0}, asm, ZPR64>;
+
+ def : SVE_InReg_Extend_PassthruUndefZero<nxv2i64, op, nxv2i1, nxv2i32, !cast<Instruction>(NAME # _D)>;
+}
+
multiclass sve_int_un_pred_arit_bitwise<bits<3> opc, string asm,
SDPatternOperator op> {
def _B : sve_int_un_pred_arit<0b00, { opc, 0b1 }, asm, ZPR8>,
diff --git a/llvm/test/CodeGen/AArch64/zeroing-forms-ext.ll b/llvm/test/CodeGen/AArch64/zeroing-forms-ext.ll
new file mode 100644
index 00000000000000..9ecb4171091d60
--- /dev/null
+++ b/llvm/test/CodeGen/AArch64/zeroing-forms-ext.ll
@@ -0,0 +1,572 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
+; RUN: llc -mattr=+sve < %s | FileCheck %s
+; RUN: llc -mattr=+sve2p2 < %s | FileCheck %s -check-prefix CHECK-2p2
+
+; RUN: llc -mattr=+sme -force-streaming < %s | FileCheck %s
+; RUN: llc -mattr=+sme2p2 -force-streaming < %s | FileCheck %s -check-prefix CHECK-2p2
+
+target triple = "aarch64-linux"
+
+define <vscale x 8 x i16> @test_svextb_s16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svextb_s16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: sxtb z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svextb_s16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sxtb z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sxtb.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svextb_s16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svextb_s16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: sxtb z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svextb_s16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sxtb z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sxtb.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svextb_s16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svextb_s16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: sxtb z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svextb_s16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sxtb z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sxtb.nxv8i16(<vscale x 8 x i16> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svextb_s32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svextb_s32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: sxtb z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svextb_s32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sxtb z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sxtb.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svextb_s32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svextb_s32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: sxtb z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svextb_s32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sxtb z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sxtb.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svextb_s32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svextb_s32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: sxtb z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svextb_s32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sxtb z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sxtb.nxv4i32(<vscale x 4 x i32> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svextb_s64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svextb_s64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: sxtb z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svextb_s64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sxtb z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sxtb.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svextb_s64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svextb_s64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: sxtb z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svextb_s64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sxtb z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sxtb.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svextb_s64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svextb_s64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: sxtb z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svextb_s64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sxtb z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sxtb.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 8 x i16> @test_svextb_u16_x_1(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svextb_u16_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: uxtb z0.h, p0/m, z0.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svextb_u16_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: uxtb z0.h, p0/z, z0.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.uxtb.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svextb_u16_x_2(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svextb_u16_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: uxtb z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svextb_u16_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: uxtb z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.uxtb.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svextb_u16_z(<vscale x 8 x i1> %pg, double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svextb_u16_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.h, #0 // =0x0
+; CHECK-NEXT: uxtb z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svextb_u16_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: uxtb z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.uxtb.nxv8i16(<vscale x 8 x i16> zeroinitializer, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svextb_u32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svextb_u32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: uxtb z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svextb_u32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: uxtb z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.uxtb.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svextb_u32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svextb_u32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: uxtb z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svextb_u32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: uxtb z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.uxtb.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svextb_u32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svextb_u32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: uxtb z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svextb_u32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: uxtb z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.uxtb.nxv4i32(<vscale x 4 x i32> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svextb_u64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svextb_u64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: uxtb z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svextb_u64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: uxtb z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.uxtb.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svextb_u64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svextb_u64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: uxtb z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svextb_u64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: uxtb z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.uxtb.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svextb_u64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svextb_u64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: uxtb z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svextb_u64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: uxtb z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.uxtb.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 4 x i32> @test_svexth_s32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svexth_s32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: sxth z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svexth_s32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sxth z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sxth.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svexth_s32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svexth_s32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: sxth z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svexth_s32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sxth z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sxth.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svexth_s32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svexth_s32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: sxth z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svexth_s32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sxth z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sxth.nxv4i32(<vscale x 4 x i32> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svexth_s64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svexth_s64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: sxth z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svexth_s64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sxth z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sxth.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svexth_s64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svexth_s64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: sxth z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svexth_s64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sxth z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sxth.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svexth_s64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svexth_s64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: sxth z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svexth_s64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sxth z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sxth.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 4 x i32> @test_svexth_u32_x_1(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svexth_u32_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: uxth z0.s, p0/m, z0.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svexth_u32_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: uxth z0.s, p0/z, z0.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.uxth.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svexth_u32_x_2(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svexth_u32_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: uxth z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svexth_u32_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: uxth z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.uxth.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svexth_u32_z(<vscale x 4 x i1> %pg, double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svexth_u32_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.s, #0 // =0x0
+; CHECK-NEXT: uxth z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svexth_u32_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: uxth z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.uxth.nxv4i32(<vscale x 4 x i32> zeroinitializer, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svexth_u64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svexth_u64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: uxth z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svexth_u64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: uxth z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.uxth.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svexth_u64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svexth_u64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: uxth z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svexth_u64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: uxth z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.uxth.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svexth_u64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svexth_u64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: uxth z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svexth_u64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: uxth z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.uxth.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svextw_s64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svextw_s64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: sxtw z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svextw_s64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sxtw z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sxtw.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svextw_s64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svextw_s64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: sxtw z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svextw_s64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sxtw z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sxtw.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svextw_s64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svextw_s64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: sxtw z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svextw_s64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: sxtw z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sxtw.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svextw_u64_x_1(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svextw_u64_x_1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: uxtw z0.d, p0/m, z0.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svextw_u64_x_1:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: uxtw z0.d, p0/z, z0.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.uxtw.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svextw_u64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svextw_u64_x_2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: uxtw z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svextw_u64_x_2:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: uxtw z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.uxtw.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svextw_u64_z(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svextw_u64_z:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov z0.d, #0 // =0x0
+; CHECK-NEXT: uxtw z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svextw_u64_z:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: uxtw z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.uxtw.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
>From 82d5f95280f5701ec84ee62736892aa68fedb442 Mon Sep 17 00:00:00 2001
From: Momchil Velikov <momchil.velikov at arm.com>
Date: Fri, 20 Dec 2024 18:03:56 +0000
Subject: [PATCH 16/16] [fixup] Rebase, update tests, add SVEAllActive
patterns, add tests for the new all-true patterns, change undef to poison
---
llvm/lib/Target/AArch64/SVEInstrFormats.td | 26 +-
.../test/CodeGen/AArch64/zeroing-forms-ext.ll | 504 +++++++++++++++++-
2 files changed, 496 insertions(+), 34 deletions(-)
diff --git a/llvm/lib/Target/AArch64/SVEInstrFormats.td b/llvm/lib/Target/AArch64/SVEInstrFormats.td
index 41db627e76ce69..67baf2826ecac5 100644
--- a/llvm/lib/Target/AArch64/SVEInstrFormats.td
+++ b/llvm/lib/Target/AArch64/SVEInstrFormats.td
@@ -662,10 +662,16 @@ multiclass SVE_InReg_Extend_PassthruUndef<ValueType vt, SDPatternOperator op, Va
(inst $PassThru, $Pg, $Src)>;
}
-class SVE_InReg_Extend_PassthruUndefZero<ValueType vt, SDPatternOperator op, ValueType pt,
- ValueType inreg_vt, Instruction inst>
- : Pat<(vt (op pt:$Pg, vt:$Src, inreg_vt, (vt (SVEDup0Undef)))),
- (inst $Pg, $Src)>;
+multiclass SVE_InReg_Extend_PassthruUndefZero<ValueType vt, SDPatternOperator op, ValueType pt,
+ ValueType inreg_vt, Instruction inst> {
+ let AddedComplexity = 1 in {
+ def : Pat<(vt (op pt:$Pg, vt:$Src, inreg_vt, (vt (SVEDup0Undef)))),
+ (inst $Pg, $Src)>;
+
+ def : Pat<(vt (op (pt (SVEAllActive:$Pg)), vt:$Src, inreg_vt, (vt (SVEAny)))),
+ (inst $Pg, $Src)>;
+ }
+}
class SVE_Shift_DupImm_Pred_Pat<ValueType vt, SDPatternOperator op,
ValueType pt, ValueType it,
@@ -4957,9 +4963,9 @@ multiclass sve_int_un_pred_arit_h_z<bits<3> opc, string asm, SDPatternOperator o
def _S : sve_int_un_pred_arit_z<0b10, { opc, 0b0 }, asm, ZPR32>;
def _D : sve_int_un_pred_arit_z<0b11, { opc, 0b0 }, asm, ZPR64>;
- def : SVE_InReg_Extend_PassthruUndefZero<nxv8i16, op, nxv8i1, nxv8i8, !cast<Instruction>(NAME # _H)>;
- def : SVE_InReg_Extend_PassthruUndefZero<nxv4i32, op, nxv4i1, nxv4i8, !cast<Instruction>(NAME # _S)>;
- def : SVE_InReg_Extend_PassthruUndefZero<nxv2i64, op, nxv2i1, nxv2i8, !cast<Instruction>(NAME # _D)>;
+ defm : SVE_InReg_Extend_PassthruUndefZero<nxv8i16, op, nxv8i1, nxv8i8, !cast<Instruction>(NAME # _H)>;
+ defm : SVE_InReg_Extend_PassthruUndefZero<nxv4i32, op, nxv4i1, nxv4i8, !cast<Instruction>(NAME # _S)>;
+ defm : SVE_InReg_Extend_PassthruUndefZero<nxv2i64, op, nxv2i1, nxv2i8, !cast<Instruction>(NAME # _D)>;
}
multiclass sve_int_un_pred_arit_w<bits<3> opc, string asm,
@@ -4983,8 +4989,8 @@ multiclass sve_int_un_pred_arit_w_z<bits<3> opc, string asm, SDPatternOperator o
def _S : sve_int_un_pred_arit_z<0b10, { opc, 0b0 }, asm, ZPR32>;
def _D : sve_int_un_pred_arit_z<0b11, { opc, 0b0 }, asm, ZPR64>;
- def : SVE_InReg_Extend_PassthruUndefZero<nxv4i32, op, nxv4i1, nxv4i16, !cast<Instruction>(NAME # _S)>;
- def : SVE_InReg_Extend_PassthruUndefZero<nxv2i64, op, nxv2i1, nxv2i16, !cast<Instruction>(NAME # _D)>;
+ defm : SVE_InReg_Extend_PassthruUndefZero<nxv4i32, op, nxv4i1, nxv4i16, !cast<Instruction>(NAME # _S)>;
+ defm : SVE_InReg_Extend_PassthruUndefZero<nxv2i64, op, nxv2i1, nxv2i16, !cast<Instruction>(NAME # _D)>;
}
multiclass sve_int_un_pred_arit_d<bits<3> opc, string asm,
@@ -5002,7 +5008,7 @@ multiclass sve_int_un_pred_arit_d<bits<3> opc, string asm,
multiclass sve_int_un_pred_arit_d_z<bits<3> opc, string asm, SDPatternOperator op> {
def _D : sve_int_un_pred_arit_z<0b11, {opc, 0b0}, asm, ZPR64>;
- def : SVE_InReg_Extend_PassthruUndefZero<nxv2i64, op, nxv2i1, nxv2i32, !cast<Instruction>(NAME # _D)>;
+ defm : SVE_InReg_Extend_PassthruUndefZero<nxv2i64, op, nxv2i1, nxv2i32, !cast<Instruction>(NAME # _D)>;
}
multiclass sve_int_un_pred_arit_bitwise<bits<3> opc, string asm,
diff --git a/llvm/test/CodeGen/AArch64/zeroing-forms-ext.ll b/llvm/test/CodeGen/AArch64/zeroing-forms-ext.ll
index 9ecb4171091d60..b29805c2b8f05a 100644
--- a/llvm/test/CodeGen/AArch64/zeroing-forms-ext.ll
+++ b/llvm/test/CodeGen/AArch64/zeroing-forms-ext.ll
@@ -18,7 +18,7 @@ define <vscale x 8 x i16> @test_svextb_s16_x_1(<vscale x 8 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: sxtb z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sxtb.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sxtb.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x i16> %0
}
@@ -34,7 +34,7 @@ define <vscale x 8 x i16> @test_svextb_s16_x_2(<vscale x 8 x i1> %pg, double %z0
; CHECK-2p2-NEXT: sxtb z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sxtb.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sxtb.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x i16> %0
}
@@ -65,7 +65,7 @@ define <vscale x 4 x i32> @test_svextb_s32_x_1(<vscale x 4 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: sxtb z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sxtb.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sxtb.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -81,7 +81,7 @@ define <vscale x 4 x i32> @test_svextb_s32_x_2(<vscale x 4 x i1> %pg, double %z0
; CHECK-2p2-NEXT: sxtb z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sxtb.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sxtb.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -112,7 +112,7 @@ define <vscale x 2 x i64> @test_svextb_s64_x_1(<vscale x 2 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: sxtb z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sxtb.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sxtb.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -128,7 +128,7 @@ define <vscale x 2 x i64> @test_svextb_s64_x_2(<vscale x 2 x i1> %pg, double %z0
; CHECK-2p2-NEXT: sxtb z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sxtb.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sxtb.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -159,7 +159,7 @@ define <vscale x 8 x i16> @test_svextb_u16_x_1(<vscale x 8 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: uxtb z0.h, p0/z, z0.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.uxtb.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.uxtb.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x i16> %0
}
@@ -175,7 +175,7 @@ define <vscale x 8 x i16> @test_svextb_u16_x_2(<vscale x 8 x i1> %pg, double %z0
; CHECK-2p2-NEXT: uxtb z0.h, p0/z, z1.h
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.uxtb.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.uxtb.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
ret <vscale x 8 x i16> %0
}
@@ -206,7 +206,7 @@ define <vscale x 4 x i32> @test_svextb_u32_x_1(<vscale x 4 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: uxtb z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.uxtb.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.uxtb.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -222,7 +222,7 @@ define <vscale x 4 x i32> @test_svextb_u32_x_2(<vscale x 4 x i1> %pg, double %z0
; CHECK-2p2-NEXT: uxtb z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.uxtb.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.uxtb.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -253,7 +253,7 @@ define <vscale x 2 x i64> @test_svextb_u64_x_1(<vscale x 2 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: uxtb z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.uxtb.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.uxtb.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -269,7 +269,7 @@ define <vscale x 2 x i64> @test_svextb_u64_x_2(<vscale x 2 x i1> %pg, double %z0
; CHECK-2p2-NEXT: uxtb z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.uxtb.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.uxtb.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -300,7 +300,7 @@ define <vscale x 4 x i32> @test_svexth_s32_x_1(<vscale x 4 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: sxth z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sxth.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sxth.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -316,7 +316,7 @@ define <vscale x 4 x i32> @test_svexth_s32_x_2(<vscale x 4 x i1> %pg, double %z0
; CHECK-2p2-NEXT: sxth z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sxth.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sxth.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -347,7 +347,7 @@ define <vscale x 2 x i64> @test_svexth_s64_x_1(<vscale x 2 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: sxth z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sxth.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sxth.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -363,7 +363,7 @@ define <vscale x 2 x i64> @test_svexth_s64_x_2(<vscale x 2 x i1> %pg, double %z0
; CHECK-2p2-NEXT: sxth z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sxth.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sxth.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -394,7 +394,7 @@ define <vscale x 4 x i32> @test_svexth_u32_x_1(<vscale x 4 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: uxth z0.s, p0/z, z0.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.uxth.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.uxth.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -410,7 +410,7 @@ define <vscale x 4 x i32> @test_svexth_u32_x_2(<vscale x 4 x i1> %pg, double %z0
; CHECK-2p2-NEXT: uxth z0.s, p0/z, z1.s
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.uxth.nxv4i32(<vscale x 4 x i32> undef, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.uxth.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
ret <vscale x 4 x i32> %0
}
@@ -441,7 +441,7 @@ define <vscale x 2 x i64> @test_svexth_u64_x_1(<vscale x 2 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: uxth z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.uxth.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.uxth.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -457,7 +457,7 @@ define <vscale x 2 x i64> @test_svexth_u64_x_2(<vscale x 2 x i1> %pg, double %z0
; CHECK-2p2-NEXT: uxth z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.uxth.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.uxth.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -488,7 +488,7 @@ define <vscale x 2 x i64> @test_svextw_s64_x_1(<vscale x 2 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: sxtw z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sxtw.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sxtw.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -504,7 +504,7 @@ define <vscale x 2 x i64> @test_svextw_s64_x_2(<vscale x 2 x i1> %pg, double %z0
; CHECK-2p2-NEXT: sxtw z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sxtw.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sxtw.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -535,7 +535,7 @@ define <vscale x 2 x i64> @test_svextw_u64_x_1(<vscale x 2 x i1> %pg, <vscale x
; CHECK-2p2-NEXT: uxtw z0.d, p0/z, z0.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.uxtw.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.uxtw.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -551,7 +551,7 @@ define <vscale x 2 x i64> @test_svextw_u64_x_2(<vscale x 2 x i1> %pg, double %z0
; CHECK-2p2-NEXT: uxtw z0.d, p0/z, z1.d
; CHECK-2p2-NEXT: ret
entry:
- %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.uxtw.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.uxtw.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
@@ -570,3 +570,459 @@ entry:
%0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.uxtw.nxv2i64(<vscale x 2 x i64> zeroinitializer, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
ret <vscale x 2 x i64> %0
}
+
+define <vscale x 8 x i16> @test_svsxtb_nxv8i16_ptrue_u(double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svsxtb_nxv8i16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: sxtb z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsxtb_nxv8i16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: sxtb z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sxtb.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svsxtb_nxv8i16_ptrue(double %z0, <vscale x 8 x i16> %x, <vscale x 8 x i16> %y) {
+; CHECK-LABEL: test_svsxtb_nxv8i16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: sxtb z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsxtb_nxv8i16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: sxtb z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sxtb.nxv8i16(<vscale x 8 x i16> %x, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %y)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svsxtb_nxv4i32_ptrue_u(double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svsxtb_nxv4i32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: sxtb z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsxtb_nxv4i32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: sxtb z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sxtb.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svsxtb_nxv4i32_ptrue(double %z0, <vscale x 4 x i32> %x, <vscale x 4 x i32> %y) {
+; CHECK-LABEL: test_svsxtb_nxv4i32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: sxtb z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsxtb_nxv4i32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: sxtb z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sxtb.nxv4i32(<vscale x 4 x i32> %x, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %y)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svsxtb_nxv2i64_ptrue_u(double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svsxtb_nxv2i64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: sxtb z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsxtb_nxv2i64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: sxtb z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sxtb.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svsxtb_nxv2i64_ptrue(double %z0, <vscale x 2 x i64> %x, <vscale x 2 x i64> %y) {
+; CHECK-LABEL: test_svsxtb_nxv2i64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: sxtb z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsxtb_nxv2i64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: sxtb z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sxtb.nxv2i64(<vscale x 2 x i64> %x, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %y)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 8 x i16> @test_svuxtb_nxv8i16_ptrue_u(double %z0, <vscale x 8 x i16> %x) {
+; CHECK-LABEL: test_svuxtb_nxv8i16_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: uxtb z0.h, p0/m, z1.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svuxtb_nxv8i16_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: uxtb z0.h, p0/z, z1.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.uxtb.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %x)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 8 x i16> @test_svuxtb_nxv8i16_ptrue(double %z0, <vscale x 8 x i16> %x, <vscale x 8 x i16> %y) {
+; CHECK-LABEL: test_svuxtb_nxv8i16_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.h
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: uxtb z0.h, p0/m, z2.h
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svuxtb_nxv8i16_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.h
+; CHECK-2p2-NEXT: uxtb z0.h, p0/z, z2.h
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
+ %0 = tail call <vscale x 8 x i16> @llvm.aarch64.sve.uxtb.nxv8i16(<vscale x 8 x i16> %x, <vscale x 8 x i1> %pg, <vscale x 8 x i16> %y)
+ ret <vscale x 8 x i16> %0
+}
+
+define <vscale x 4 x i32> @test_svuxtb_nxv4i32_ptrue_u(double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svuxtb_nxv4i32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: uxtb z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svuxtb_nxv4i32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: uxtb z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.uxtb.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svuxtb_nxv4i32_ptrue(double %z0, <vscale x 4 x i32> %x, <vscale x 4 x i32> %y) {
+; CHECK-LABEL: test_svuxtb_nxv4i32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: uxtb z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svuxtb_nxv4i32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: uxtb z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.uxtb.nxv4i32(<vscale x 4 x i32> %x, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %y)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svuxtb_nxv2i64_ptrue_u(double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svuxtb_nxv2i64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: uxtb z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svuxtb_nxv2i64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: uxtb z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.uxtb.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svuxtb_nxv2i64_ptrue(double %z0, <vscale x 2 x i64> %x, <vscale x 2 x i64> %y) {
+; CHECK-LABEL: test_svuxtb_nxv2i64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: uxtb z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svuxtb_nxv2i64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: uxtb z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.uxtb.nxv2i64(<vscale x 2 x i64> %x, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %y)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 4 x i32> @test_svsxth_nxv4i32_ptrue_u(double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svsxth_nxv4i32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: sxth z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsxth_nxv4i32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: sxth z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sxth.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svsxth_nxv4i32_ptrue(double %z0, <vscale x 4 x i32> %x, <vscale x 4 x i32> %y) {
+; CHECK-LABEL: test_svsxth_nxv4i32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: sxth z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsxth_nxv4i32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: sxth z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sxth.nxv4i32(<vscale x 4 x i32> %x, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %y)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svsxth_nxv2i64_ptrue_u(double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svsxth_nxv2i64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: sxth z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsxth_nxv2i64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: sxth z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sxth.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svsxth_nxv2i64_ptrue(double %z0, <vscale x 2 x i64> %x, <vscale x 2 x i64> %y) {
+; CHECK-LABEL: test_svsxth_nxv2i64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: sxth z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsxth_nxv2i64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: sxth z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sxth.nxv2i64(<vscale x 2 x i64> %x, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %y)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 4 x i32> @test_svuxth_nxv4i32_ptrue_u(double %z0, <vscale x 4 x i32> %x) {
+; CHECK-LABEL: test_svuxth_nxv4i32_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: uxth z0.s, p0/m, z1.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svuxth_nxv4i32_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: uxth z0.s, p0/z, z1.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.uxth.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %x)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 4 x i32> @test_svuxth_nxv4i32_ptrue(double %z0, <vscale x 4 x i32> %x, <vscale x 4 x i32> %y) {
+; CHECK-LABEL: test_svuxth_nxv4i32_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.s
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: uxth z0.s, p0/m, z2.s
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svuxth_nxv4i32_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.s
+; CHECK-2p2-NEXT: uxth z0.s, p0/z, z2.s
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
+ %0 = tail call <vscale x 4 x i32> @llvm.aarch64.sve.uxth.nxv4i32(<vscale x 4 x i32> %x, <vscale x 4 x i1> %pg, <vscale x 4 x i32> %y)
+ ret <vscale x 4 x i32> %0
+}
+
+define <vscale x 2 x i64> @test_svuxth_nxv2i64_ptrue_u(double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svuxth_nxv2i64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: uxth z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svuxth_nxv2i64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: uxth z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.uxth.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svuxth_nxv2i64_ptrue(double %z0, <vscale x 2 x i64> %x, <vscale x 2 x i64> %y) {
+; CHECK-LABEL: test_svuxth_nxv2i64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: uxth z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svuxth_nxv2i64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: uxth z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.uxth.nxv2i64(<vscale x 2 x i64> %x, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %y)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svsxtw_nxv2i64_ptrue_u(double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svsxtw_nxv2i64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: sxtw z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsxtw_nxv2i64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: sxtw z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sxtw.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svsxtw_nxv2i64_ptrue(double %z0, <vscale x 2 x i64> %x, <vscale x 2 x i64> %y) {
+; CHECK-LABEL: test_svsxtw_nxv2i64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: sxtw z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svsxtw_nxv2i64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: sxtw z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sxtw.nxv2i64(<vscale x 2 x i64> %x, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %y)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svuxtw_nxv2i64_ptrue_u(double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABEL: test_svuxtw_nxv2i64_ptrue_u:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z1
+; CHECK-NEXT: uxtw z0.d, p0/m, z1.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svuxtw_nxv2i64_ptrue_u:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: uxtw z0.d, p0/z, z1.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.uxtw.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %x)
+ ret <vscale x 2 x i64> %0
+}
+
+define <vscale x 2 x i64> @test_svuxtw_nxv2i64_ptrue(double %z0, <vscale x 2 x i64> %x, <vscale x 2 x i64> %y) {
+; CHECK-LABEL: test_svuxtw_nxv2i64_ptrue:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ptrue p0.d
+; CHECK-NEXT: movprfx z0, z2
+; CHECK-NEXT: uxtw z0.d, p0/m, z2.d
+; CHECK-NEXT: ret
+;
+; CHECK-2p2-LABEL: test_svuxtw_nxv2i64_ptrue:
+; CHECK-2p2: // %bb.0: // %entry
+; CHECK-2p2-NEXT: ptrue p0.d
+; CHECK-2p2-NEXT: uxtw z0.d, p0/z, z2.d
+; CHECK-2p2-NEXT: ret
+entry:
+ %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+ %0 = tail call <vscale x 2 x i64> @llvm.aarch64.sve.uxtw.nxv2i64(<vscale x 2 x i64> %x, <vscale x 2 x i1> %pg, <vscale x 2 x i64> %y)
+ ret <vscale x 2 x i64> %0
+}
More information about the llvm-commits
mailing list