[llvm] [RISCV] Use mask undisturbed policy when silencing sNans for strict rounding ops. (PR #93356)
via llvm-commits
llvm-commits at lists.llvm.org
Fri May 24 16:56:58 PDT 2024
llvmbot wrote:
<!--LLVM PR SUMMARY COMMENT-->
@llvm/pr-subscribers-backend-risc-v
Author: Craig Topper (topperc)
<details>
<summary>Changes</summary>
The elements that aren't sNans need to get passed through this fadd instruction unchanged. With the agnostic mask policy they might be forced to all ones.
---
Patch is 143.67 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/93356.diff
13 Files Affected:
- (modified) llvm/lib/Target/RISCV/RISCVISelLowering.cpp (+1-1)
- (modified) llvm/test/CodeGen/RISCV/rvv/fceil-constrained-sdnode.ll (+30-15)
- (modified) llvm/test/CodeGen/RISCV/rvv/ffloor-constrained-sdnode.ll (+30-15)
- (modified) llvm/test/CodeGen/RISCV/rvv/fixed-vectors-fceil-constrained-sdnode.ll (+30-15)
- (modified) llvm/test/CodeGen/RISCV/rvv/fixed-vectors-ffloor-constrained-sdnode.ll (+30-15)
- (modified) llvm/test/CodeGen/RISCV/rvv/fixed-vectors-fnearbyint-constrained-sdnode.ll (+24-12)
- (modified) llvm/test/CodeGen/RISCV/rvv/fixed-vectors-fround-constrained-sdnode.ll (+30-15)
- (modified) llvm/test/CodeGen/RISCV/rvv/fixed-vectors-froundeven-constrained-sdnode.ll (+30-15)
- (modified) llvm/test/CodeGen/RISCV/rvv/fixed-vectors-ftrunc-constrained-sdnode.ll (+30-15)
- (modified) llvm/test/CodeGen/RISCV/rvv/fnearbyint-constrained-sdnode.ll (+30-15)
- (modified) llvm/test/CodeGen/RISCV/rvv/fround-constrained-sdnode.ll (+30-15)
- (modified) llvm/test/CodeGen/RISCV/rvv/froundeven-constrained-sdnode.ll (+30-15)
- (modified) llvm/test/CodeGen/RISCV/rvv/ftrunc-constrained-sdnode.ll (+30-15)
``````````diff
diff --git a/llvm/lib/Target/RISCV/RISCVISelLowering.cpp b/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
index f0e5a7d393b6c..c826892c1668e 100644
--- a/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
+++ b/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
@@ -3125,7 +3125,7 @@ lowerVectorStrictFTRUNC_FCEIL_FFLOOR_FROUND(SDValue Op, SelectionDAG &DAG,
Chain = Unorder.getValue(1);
Src = DAG.getNode(RISCVISD::STRICT_FADD_VL, DL,
DAG.getVTList(ContainerVT, MVT::Other),
- {Chain, Src, Src, DAG.getUNDEF(ContainerVT), Unorder, VL});
+ {Chain, Src, Src, Src, Unorder, VL});
Chain = Src.getValue(1);
// We do the conversion on the absolute value and fix the sign at the end.
diff --git a/llvm/test/CodeGen/RISCV/rvv/fceil-constrained-sdnode.ll b/llvm/test/CodeGen/RISCV/rvv/fceil-constrained-sdnode.ll
index 75747a6674b7b..d8781495abd75 100644
--- a/llvm/test/CodeGen/RISCV/rvv/fceil-constrained-sdnode.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/fceil-constrained-sdnode.ll
@@ -7,7 +7,7 @@
define <vscale x 1 x half> @ceil_nxv1f16(<vscale x 1 x half> %x) strictfp {
; CHECK-LABEL: ceil_nxv1f16:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT: vsetvli a0, zero, e16, mf4, ta, mu
; CHECK-NEXT: vmfne.vv v0, v8, v8
; CHECK-NEXT: lui a0, %hi(.LCPI0_0)
; CHECK-NEXT: flh fa5, %lo(.LCPI0_0)(a0)
@@ -15,6 +15,7 @@ define <vscale x 1 x half> @ceil_nxv1f16(<vscale x 1 x half> %x) strictfp {
; CHECK-NEXT: vfabs.v v9, v8
; CHECK-NEXT: vmflt.vf v0, v9, fa5
; CHECK-NEXT: fsrmi a0, 3
+; CHECK-NEXT: vsetvli zero, zero, e16, mf4, ta, ma
; CHECK-NEXT: vfcvt.x.f.v v9, v8, v0.t
; CHECK-NEXT: fsrm a0
; CHECK-NEXT: vfcvt.f.x.v v9, v9, v0.t
@@ -29,7 +30,7 @@ declare <vscale x 1 x half> @llvm.experimental.constrained.ceil.nxv1f16(<vscale
define <vscale x 2 x half> @ceil_nxv2f16(<vscale x 2 x half> %x) strictfp {
; CHECK-LABEL: ceil_nxv2f16:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT: vsetvli a0, zero, e16, mf2, ta, mu
; CHECK-NEXT: vmfne.vv v0, v8, v8
; CHECK-NEXT: lui a0, %hi(.LCPI1_0)
; CHECK-NEXT: flh fa5, %lo(.LCPI1_0)(a0)
@@ -37,6 +38,7 @@ define <vscale x 2 x half> @ceil_nxv2f16(<vscale x 2 x half> %x) strictfp {
; CHECK-NEXT: vfabs.v v9, v8
; CHECK-NEXT: vmflt.vf v0, v9, fa5
; CHECK-NEXT: fsrmi a0, 3
+; CHECK-NEXT: vsetvli zero, zero, e16, mf2, ta, ma
; CHECK-NEXT: vfcvt.x.f.v v9, v8, v0.t
; CHECK-NEXT: fsrm a0
; CHECK-NEXT: vfcvt.f.x.v v9, v9, v0.t
@@ -51,7 +53,7 @@ declare <vscale x 2 x half> @llvm.experimental.constrained.ceil.nxv2f16(<vscale
define <vscale x 4 x half> @ceil_nxv4f16(<vscale x 4 x half> %x) strictfp {
; CHECK-LABEL: ceil_nxv4f16:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT: vsetvli a0, zero, e16, m1, ta, mu
; CHECK-NEXT: vmfne.vv v0, v8, v8
; CHECK-NEXT: lui a0, %hi(.LCPI2_0)
; CHECK-NEXT: flh fa5, %lo(.LCPI2_0)(a0)
@@ -59,6 +61,7 @@ define <vscale x 4 x half> @ceil_nxv4f16(<vscale x 4 x half> %x) strictfp {
; CHECK-NEXT: vfabs.v v9, v8
; CHECK-NEXT: vmflt.vf v0, v9, fa5
; CHECK-NEXT: fsrmi a0, 3
+; CHECK-NEXT: vsetvli zero, zero, e16, m1, ta, ma
; CHECK-NEXT: vfcvt.x.f.v v9, v8, v0.t
; CHECK-NEXT: fsrm a0
; CHECK-NEXT: vfcvt.f.x.v v9, v9, v0.t
@@ -73,7 +76,7 @@ declare <vscale x 4 x half> @llvm.experimental.constrained.ceil.nxv4f16(<vscale
define <vscale x 8 x half> @ceil_nxv8f16(<vscale x 8 x half> %x) strictfp {
; CHECK-LABEL: ceil_nxv8f16:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT: vsetvli a0, zero, e16, m2, ta, mu
; CHECK-NEXT: vmfne.vv v0, v8, v8
; CHECK-NEXT: lui a0, %hi(.LCPI3_0)
; CHECK-NEXT: flh fa5, %lo(.LCPI3_0)(a0)
@@ -81,6 +84,7 @@ define <vscale x 8 x half> @ceil_nxv8f16(<vscale x 8 x half> %x) strictfp {
; CHECK-NEXT: vfabs.v v10, v8
; CHECK-NEXT: vmflt.vf v0, v10, fa5
; CHECK-NEXT: fsrmi a0, 3
+; CHECK-NEXT: vsetvli zero, zero, e16, m2, ta, ma
; CHECK-NEXT: vfcvt.x.f.v v10, v8, v0.t
; CHECK-NEXT: fsrm a0
; CHECK-NEXT: vfcvt.f.x.v v10, v10, v0.t
@@ -95,7 +99,7 @@ declare <vscale x 8 x half> @llvm.experimental.constrained.ceil.nxv8f16(<vscale
define <vscale x 16 x half> @ceil_nxv16f16(<vscale x 16 x half> %x) strictfp {
; CHECK-LABEL: ceil_nxv16f16:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e16, m4, ta, ma
+; CHECK-NEXT: vsetvli a0, zero, e16, m4, ta, mu
; CHECK-NEXT: vmfne.vv v0, v8, v8
; CHECK-NEXT: lui a0, %hi(.LCPI4_0)
; CHECK-NEXT: flh fa5, %lo(.LCPI4_0)(a0)
@@ -103,6 +107,7 @@ define <vscale x 16 x half> @ceil_nxv16f16(<vscale x 16 x half> %x) strictfp {
; CHECK-NEXT: vfabs.v v12, v8
; CHECK-NEXT: vmflt.vf v0, v12, fa5
; CHECK-NEXT: fsrmi a0, 3
+; CHECK-NEXT: vsetvli zero, zero, e16, m4, ta, ma
; CHECK-NEXT: vfcvt.x.f.v v12, v8, v0.t
; CHECK-NEXT: fsrm a0
; CHECK-NEXT: vfcvt.f.x.v v12, v12, v0.t
@@ -117,7 +122,7 @@ declare <vscale x 16 x half> @llvm.experimental.constrained.ceil.nxv16f16(<vscal
define <vscale x 32 x half> @ceil_nxv32f16(<vscale x 32 x half> %x) strictfp {
; CHECK-LABEL: ceil_nxv32f16:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e16, m8, ta, ma
+; CHECK-NEXT: vsetvli a0, zero, e16, m8, ta, mu
; CHECK-NEXT: vmfne.vv v0, v8, v8
; CHECK-NEXT: lui a0, %hi(.LCPI5_0)
; CHECK-NEXT: flh fa5, %lo(.LCPI5_0)(a0)
@@ -125,6 +130,7 @@ define <vscale x 32 x half> @ceil_nxv32f16(<vscale x 32 x half> %x) strictfp {
; CHECK-NEXT: vfabs.v v16, v8
; CHECK-NEXT: vmflt.vf v0, v16, fa5
; CHECK-NEXT: fsrmi a0, 3
+; CHECK-NEXT: vsetvli zero, zero, e16, m8, ta, ma
; CHECK-NEXT: vfcvt.x.f.v v16, v8, v0.t
; CHECK-NEXT: fsrm a0
; CHECK-NEXT: vfcvt.f.x.v v16, v16, v0.t
@@ -139,7 +145,7 @@ declare <vscale x 32 x half> @llvm.experimental.constrained.ceil.nxv32f16(<vscal
define <vscale x 1 x float> @ceil_nxv1f32(<vscale x 1 x float> %x) strictfp {
; CHECK-LABEL: ceil_nxv1f32:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT: vsetvli a0, zero, e32, mf2, ta, mu
; CHECK-NEXT: vmfne.vv v0, v8, v8
; CHECK-NEXT: vfadd.vv v8, v8, v8, v0.t
; CHECK-NEXT: vfabs.v v9, v8
@@ -147,6 +153,7 @@ define <vscale x 1 x float> @ceil_nxv1f32(<vscale x 1 x float> %x) strictfp {
; CHECK-NEXT: fmv.w.x fa5, a0
; CHECK-NEXT: vmflt.vf v0, v9, fa5
; CHECK-NEXT: fsrmi a0, 3
+; CHECK-NEXT: vsetvli zero, zero, e32, mf2, ta, ma
; CHECK-NEXT: vfcvt.x.f.v v9, v8, v0.t
; CHECK-NEXT: fsrm a0
; CHECK-NEXT: vfcvt.f.x.v v9, v9, v0.t
@@ -161,7 +168,7 @@ declare <vscale x 1 x float> @llvm.experimental.constrained.ceil.nxv1f32(<vscale
define <vscale x 2 x float> @ceil_nxv2f32(<vscale x 2 x float> %x) strictfp {
; CHECK-LABEL: ceil_nxv2f32:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT: vsetvli a0, zero, e32, m1, ta, mu
; CHECK-NEXT: vmfne.vv v0, v8, v8
; CHECK-NEXT: vfadd.vv v8, v8, v8, v0.t
; CHECK-NEXT: vfabs.v v9, v8
@@ -169,6 +176,7 @@ define <vscale x 2 x float> @ceil_nxv2f32(<vscale x 2 x float> %x) strictfp {
; CHECK-NEXT: fmv.w.x fa5, a0
; CHECK-NEXT: vmflt.vf v0, v9, fa5
; CHECK-NEXT: fsrmi a0, 3
+; CHECK-NEXT: vsetvli zero, zero, e32, m1, ta, ma
; CHECK-NEXT: vfcvt.x.f.v v9, v8, v0.t
; CHECK-NEXT: fsrm a0
; CHECK-NEXT: vfcvt.f.x.v v9, v9, v0.t
@@ -183,7 +191,7 @@ declare <vscale x 2 x float> @llvm.experimental.constrained.ceil.nxv2f32(<vscale
define <vscale x 4 x float> @ceil_nxv4f32(<vscale x 4 x float> %x) strictfp {
; CHECK-LABEL: ceil_nxv4f32:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e32, m2, ta, ma
+; CHECK-NEXT: vsetvli a0, zero, e32, m2, ta, mu
; CHECK-NEXT: vmfne.vv v0, v8, v8
; CHECK-NEXT: vfadd.vv v8, v8, v8, v0.t
; CHECK-NEXT: vfabs.v v10, v8
@@ -191,6 +199,7 @@ define <vscale x 4 x float> @ceil_nxv4f32(<vscale x 4 x float> %x) strictfp {
; CHECK-NEXT: fmv.w.x fa5, a0
; CHECK-NEXT: vmflt.vf v0, v10, fa5
; CHECK-NEXT: fsrmi a0, 3
+; CHECK-NEXT: vsetvli zero, zero, e32, m2, ta, ma
; CHECK-NEXT: vfcvt.x.f.v v10, v8, v0.t
; CHECK-NEXT: fsrm a0
; CHECK-NEXT: vfcvt.f.x.v v10, v10, v0.t
@@ -205,7 +214,7 @@ declare <vscale x 4 x float> @llvm.experimental.constrained.ceil.nxv4f32(<vscale
define <vscale x 8 x float> @ceil_nxv8f32(<vscale x 8 x float> %x) strictfp {
; CHECK-LABEL: ceil_nxv8f32:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT: vsetvli a0, zero, e32, m4, ta, mu
; CHECK-NEXT: vmfne.vv v0, v8, v8
; CHECK-NEXT: vfadd.vv v8, v8, v8, v0.t
; CHECK-NEXT: vfabs.v v12, v8
@@ -213,6 +222,7 @@ define <vscale x 8 x float> @ceil_nxv8f32(<vscale x 8 x float> %x) strictfp {
; CHECK-NEXT: fmv.w.x fa5, a0
; CHECK-NEXT: vmflt.vf v0, v12, fa5
; CHECK-NEXT: fsrmi a0, 3
+; CHECK-NEXT: vsetvli zero, zero, e32, m4, ta, ma
; CHECK-NEXT: vfcvt.x.f.v v12, v8, v0.t
; CHECK-NEXT: fsrm a0
; CHECK-NEXT: vfcvt.f.x.v v12, v12, v0.t
@@ -227,7 +237,7 @@ declare <vscale x 8 x float> @llvm.experimental.constrained.ceil.nxv8f32(<vscale
define <vscale x 16 x float> @ceil_nxv16f32(<vscale x 16 x float> %x) strictfp {
; CHECK-LABEL: ceil_nxv16f32:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e32, m8, ta, ma
+; CHECK-NEXT: vsetvli a0, zero, e32, m8, ta, mu
; CHECK-NEXT: vmfne.vv v0, v8, v8
; CHECK-NEXT: vfadd.vv v8, v8, v8, v0.t
; CHECK-NEXT: vfabs.v v16, v8
@@ -235,6 +245,7 @@ define <vscale x 16 x float> @ceil_nxv16f32(<vscale x 16 x float> %x) strictfp {
; CHECK-NEXT: fmv.w.x fa5, a0
; CHECK-NEXT: vmflt.vf v0, v16, fa5
; CHECK-NEXT: fsrmi a0, 3
+; CHECK-NEXT: vsetvli zero, zero, e32, m8, ta, ma
; CHECK-NEXT: vfcvt.x.f.v v16, v8, v0.t
; CHECK-NEXT: fsrm a0
; CHECK-NEXT: vfcvt.f.x.v v16, v16, v0.t
@@ -249,7 +260,7 @@ declare <vscale x 16 x float> @llvm.experimental.constrained.ceil.nxv16f32(<vsca
define <vscale x 1 x double> @ceil_nxv1f64(<vscale x 1 x double> %x) strictfp {
; CHECK-LABEL: ceil_nxv1f64:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e64, m1, ta, ma
+; CHECK-NEXT: vsetvli a0, zero, e64, m1, ta, mu
; CHECK-NEXT: vmfne.vv v0, v8, v8
; CHECK-NEXT: lui a0, %hi(.LCPI11_0)
; CHECK-NEXT: fld fa5, %lo(.LCPI11_0)(a0)
@@ -257,6 +268,7 @@ define <vscale x 1 x double> @ceil_nxv1f64(<vscale x 1 x double> %x) strictfp {
; CHECK-NEXT: vfabs.v v9, v8
; CHECK-NEXT: vmflt.vf v0, v9, fa5
; CHECK-NEXT: fsrmi a0, 3
+; CHECK-NEXT: vsetvli zero, zero, e64, m1, ta, ma
; CHECK-NEXT: vfcvt.x.f.v v9, v8, v0.t
; CHECK-NEXT: fsrm a0
; CHECK-NEXT: vfcvt.f.x.v v9, v9, v0.t
@@ -271,7 +283,7 @@ declare <vscale x 1 x double> @llvm.experimental.constrained.ceil.nxv1f64(<vscal
define <vscale x 2 x double> @ceil_nxv2f64(<vscale x 2 x double> %x) strictfp {
; CHECK-LABEL: ceil_nxv2f64:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e64, m2, ta, ma
+; CHECK-NEXT: vsetvli a0, zero, e64, m2, ta, mu
; CHECK-NEXT: vmfne.vv v0, v8, v8
; CHECK-NEXT: lui a0, %hi(.LCPI12_0)
; CHECK-NEXT: fld fa5, %lo(.LCPI12_0)(a0)
@@ -279,6 +291,7 @@ define <vscale x 2 x double> @ceil_nxv2f64(<vscale x 2 x double> %x) strictfp {
; CHECK-NEXT: vfabs.v v10, v8
; CHECK-NEXT: vmflt.vf v0, v10, fa5
; CHECK-NEXT: fsrmi a0, 3
+; CHECK-NEXT: vsetvli zero, zero, e64, m2, ta, ma
; CHECK-NEXT: vfcvt.x.f.v v10, v8, v0.t
; CHECK-NEXT: fsrm a0
; CHECK-NEXT: vfcvt.f.x.v v10, v10, v0.t
@@ -293,7 +306,7 @@ declare <vscale x 2 x double> @llvm.experimental.constrained.ceil.nxv2f64(<vscal
define <vscale x 4 x double> @ceil_nxv4f64(<vscale x 4 x double> %x) strictfp {
; CHECK-LABEL: ceil_nxv4f64:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e64, m4, ta, ma
+; CHECK-NEXT: vsetvli a0, zero, e64, m4, ta, mu
; CHECK-NEXT: vmfne.vv v0, v8, v8
; CHECK-NEXT: lui a0, %hi(.LCPI13_0)
; CHECK-NEXT: fld fa5, %lo(.LCPI13_0)(a0)
@@ -301,6 +314,7 @@ define <vscale x 4 x double> @ceil_nxv4f64(<vscale x 4 x double> %x) strictfp {
; CHECK-NEXT: vfabs.v v12, v8
; CHECK-NEXT: vmflt.vf v0, v12, fa5
; CHECK-NEXT: fsrmi a0, 3
+; CHECK-NEXT: vsetvli zero, zero, e64, m4, ta, ma
; CHECK-NEXT: vfcvt.x.f.v v12, v8, v0.t
; CHECK-NEXT: fsrm a0
; CHECK-NEXT: vfcvt.f.x.v v12, v12, v0.t
@@ -315,7 +329,7 @@ declare <vscale x 4 x double> @llvm.experimental.constrained.ceil.nxv4f64(<vscal
define <vscale x 8 x double> @ceil_nxv8f64(<vscale x 8 x double> %x) strictfp {
; CHECK-LABEL: ceil_nxv8f64:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e64, m8, ta, ma
+; CHECK-NEXT: vsetvli a0, zero, e64, m8, ta, mu
; CHECK-NEXT: vmfne.vv v0, v8, v8
; CHECK-NEXT: lui a0, %hi(.LCPI14_0)
; CHECK-NEXT: fld fa5, %lo(.LCPI14_0)(a0)
@@ -323,6 +337,7 @@ define <vscale x 8 x double> @ceil_nxv8f64(<vscale x 8 x double> %x) strictfp {
; CHECK-NEXT: vfabs.v v16, v8
; CHECK-NEXT: vmflt.vf v0, v16, fa5
; CHECK-NEXT: fsrmi a0, 3
+; CHECK-NEXT: vsetvli zero, zero, e64, m8, ta, ma
; CHECK-NEXT: vfcvt.x.f.v v16, v8, v0.t
; CHECK-NEXT: fsrm a0
; CHECK-NEXT: vfcvt.f.x.v v16, v16, v0.t
diff --git a/llvm/test/CodeGen/RISCV/rvv/ffloor-constrained-sdnode.ll b/llvm/test/CodeGen/RISCV/rvv/ffloor-constrained-sdnode.ll
index 31a9453204457..1df452d8641c5 100644
--- a/llvm/test/CodeGen/RISCV/rvv/ffloor-constrained-sdnode.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/ffloor-constrained-sdnode.ll
@@ -7,7 +7,7 @@
define <vscale x 1 x half> @floor_nxv1f16(<vscale x 1 x half> %x) strictfp {
; CHECK-LABEL: floor_nxv1f16:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT: vsetvli a0, zero, e16, mf4, ta, mu
; CHECK-NEXT: vmfne.vv v0, v8, v8
; CHECK-NEXT: lui a0, %hi(.LCPI0_0)
; CHECK-NEXT: flh fa5, %lo(.LCPI0_0)(a0)
@@ -15,6 +15,7 @@ define <vscale x 1 x half> @floor_nxv1f16(<vscale x 1 x half> %x) strictfp {
; CHECK-NEXT: vfabs.v v9, v8
; CHECK-NEXT: vmflt.vf v0, v9, fa5
; CHECK-NEXT: fsrmi a0, 2
+; CHECK-NEXT: vsetvli zero, zero, e16, mf4, ta, ma
; CHECK-NEXT: vfcvt.x.f.v v9, v8, v0.t
; CHECK-NEXT: fsrm a0
; CHECK-NEXT: vfcvt.f.x.v v9, v9, v0.t
@@ -29,7 +30,7 @@ declare <vscale x 1 x half> @llvm.experimental.constrained.floor.nxv1f16(<vscale
define <vscale x 2 x half> @floor_nxv2f16(<vscale x 2 x half> %x) strictfp {
; CHECK-LABEL: floor_nxv2f16:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT: vsetvli a0, zero, e16, mf2, ta, mu
; CHECK-NEXT: vmfne.vv v0, v8, v8
; CHECK-NEXT: lui a0, %hi(.LCPI1_0)
; CHECK-NEXT: flh fa5, %lo(.LCPI1_0)(a0)
@@ -37,6 +38,7 @@ define <vscale x 2 x half> @floor_nxv2f16(<vscale x 2 x half> %x) strictfp {
; CHECK-NEXT: vfabs.v v9, v8
; CHECK-NEXT: vmflt.vf v0, v9, fa5
; CHECK-NEXT: fsrmi a0, 2
+; CHECK-NEXT: vsetvli zero, zero, e16, mf2, ta, ma
; CHECK-NEXT: vfcvt.x.f.v v9, v8, v0.t
; CHECK-NEXT: fsrm a0
; CHECK-NEXT: vfcvt.f.x.v v9, v9, v0.t
@@ -51,7 +53,7 @@ declare <vscale x 2 x half> @llvm.experimental.constrained.floor.nxv2f16(<vscale
define <vscale x 4 x half> @floor_nxv4f16(<vscale x 4 x half> %x) strictfp {
; CHECK-LABEL: floor_nxv4f16:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT: vsetvli a0, zero, e16, m1, ta, mu
; CHECK-NEXT: vmfne.vv v0, v8, v8
; CHECK-NEXT: lui a0, %hi(.LCPI2_0)
; CHECK-NEXT: flh fa5, %lo(.LCPI2_0)(a0)
@@ -59,6 +61,7 @@ define <vscale x 4 x half> @floor_nxv4f16(<vscale x 4 x half> %x) strictfp {
; CHECK-NEXT: vfabs.v v9, v8
; CHECK-NEXT: vmflt.vf v0, v9, fa5
; CHECK-NEXT: fsrmi a0, 2
+; CHECK-NEXT: vsetvli zero, zero, e16, m1, ta, ma
; CHECK-NEXT: vfcvt.x.f.v v9, v8, v0.t
; CHECK-NEXT: fsrm a0
; CHECK-NEXT: vfcvt.f.x.v v9, v9, v0.t
@@ -73,7 +76,7 @@ declare <vscale x 4 x half> @llvm.experimental.constrained.floor.nxv4f16(<vscale
define <vscale x 8 x half> @floor_nxv8f16(<vscale x 8 x half> %x) strictfp {
; CHECK-LABEL: floor_nxv8f16:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT: vsetvli a0, zero, e16, m2, ta, mu
; CHECK-NEXT: vmfne.vv v0, v8, v8
; CHECK-NEXT: lui a0, %hi(.LCPI3_0)
; CHECK-NEXT: flh fa5, %lo(.LCPI3_0)(a0)
@@ -81,6 +84,7 @@ define <vscale x 8 x half> @floor_nxv8f16(<vscale x 8 x half> %x) strictfp {
; CHECK-NEXT: vfabs.v v10, v8
; CHECK-NEXT: vmflt.vf v0, v10, fa5
; CHECK-NEXT: fsrmi a0, 2
+; CHECK-NEXT: vsetvli zero, zero, e16, m2, ta, ma
; CHECK-NEXT: vfcvt.x.f.v v10, v8, v0.t
; CHECK-NEXT: fsrm a0
; CHECK-NEXT: vfcvt.f.x.v v10, v10, v0.t
@@ -95,7 +99,7 @@ declare <vscale x 8 x half> @llvm.experimental.constrained.floor.nxv8f16(<vscale
define <vscale x 16 x half> @floor_nxv16f16(<vscale x 16 x half> %x) strictfp {
; CHECK-LABEL: floor_nxv16f16:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e16, m4, ta, ma
+; CHECK-NEXT: vsetvli a0, zero, e16, m4, ta, mu
; CHECK-NEXT: vmfne.vv v0, v8, v8
; CHECK-NEXT: lui a0, %hi(.LCPI4_0)
; CHECK-NEXT: flh fa5, %lo(.LCPI4_0)(a0)
@@ -103,6 +107,7 @@ define <vscale x 16 x half> @floor_nxv16f16(<vscale x 16 x half> %x) strictfp {
; CHECK-NEXT: vfabs.v v12, v8
; CHECK-NEXT: vmflt.vf v0, v12, fa5
; CHECK-NEXT: fsrmi a0, 2
+; CHECK-NEXT: vsetvli zero, zero, e16, m4, ta, ma
; CHECK-NEXT: vfcvt.x.f.v v12, v8, v0.t
; CHECK-NEXT: fsrm a0
; CHECK-NEXT: vfcvt.f.x.v v12, v12, v0.t
@@ -117,7 +122,7 @@ declare <vscale x 16 x half> @llvm.experimental.constrained.floor.nxv16f16(<vsca
define <vscale x 32 x half> @floor_nxv32f16(<vscale x 32 x half> %x) strictfp {
; CHECK-LABEL: floor_nxv32f16:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e16, m8, ta, ma
+; CHECK-NEXT: vsetvli a0, zero, e16, m8, ta, mu
; CHECK-NEXT: vmfne.vv v0, v8, v8
; CHECK-NEXT: lui a0, %hi(.LCPI5_0)
; CHECK-NEXT: flh fa5, %lo(.LCPI5_0)(a0)
@@ -125,6 +130,7 @@ define <vscale x 32 x half> @floor_nxv32f16(<vscale x 32 x half> %x) strictfp {
; CHECK-NEXT: vfabs.v v16, v8
; CHECK-NEXT: vmflt.vf v0, v16, fa5
; CHECK-NEXT: fsrmi a0, 2
+; CHECK-NEXT: vsetvli zero, zero, e16, m8, ta, ma
; CHECK-NEXT: vfcvt.x.f.v v16, v8, v0.t
; CHECK-NEXT: fsrm a0
; CHECK-NEXT: vfcvt.f.x.v v16, v16, v0.t
@@ -139,7 +145,7 @@ declare <vscale x 32 x half> @llvm.experimental.constrained.floor.nxv32f16(<vsca
define <vscale x 1 x float> @floor_nxv1f32(<vscale x 1 x float> %x) strictfp {
; CHECK-LABEL: floor_nxv1f32:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT: vsetvli a0, zero, e32, mf2, ta, mu
; CHECK-NEXT: vmfne.vv v0, v8, v8
; CHECK-NEXT: vfadd.vv v8, v8, v8, v0.t
; CHECK-NEXT: vfabs.v v9, v8
@@ -147,6 +153,7 @@ define <vscale x 1 x float> @floor_nxv1f32(<vscale x 1 x float> %x) strictfp {
; CHECK-NEXT: fmv.w.x fa5, a0
; CHECK-NEXT: vmflt.vf v0, v9, fa5
; CHECK-NEXT: fsrmi a0, 2
+; CHECK-NEXT: vsetvli zero, zero, e32, mf2, ta, ma
; CHECK-NEXT: vfcvt.x.f.v v9, v8, v0.t
; CHECK-NEXT: fsrm a0
; CHECK-NEXT: vfcvt.f.x.v v9, v9, v0.t
@@ -161,7 +168,7 @@ declare <vscale x 1 x float> @llvm.experimental.constrained.floor.nxv1f32(<vscal
define <vscale x 2 x float> @floor_nxv2f32(<vscale x 2 x float> %x) strictfp {
; CHECK-LABEL: floor_nxv2f32:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT: vsetvli a0, zero, e32, m1, ta, mu
; CHECK-NEXT: vmfne.vv v0, v8, v8
; CHECK-NEXT: vfadd.vv v8, v8, v8, v0.t
; CHECK-NEXT: vfabs.v v9, v8
@@ -169,6 +176,7 @@ define <vscale x 2 x float> @floor_nxv2f32(<vscale x 2 x float> %x) strictfp {
; CHECK-NEXT: fmv.w.x fa5, a0
; CHECK-NEXT: vmflt.vf v0, v9, fa5
; CHECK-NEXT: fsrmi a0, 2
+; CHECK-NEXT: vsetvli zero, zero, e32, m1, ta, ma
; CHECK-NEXT: vfcvt.x.f.v v9, v8, v0.t
; CHECK-NEXT: fsrm a0
; CHECK-NEXT: vfcvt.f.x.v v9, v9, v0.t
@@ -183,7 +191,7 @@ declare <vscale x 2 x float> @llvm.experimental.constrained.floor.nxv2f32(<vscal
define <vscale x 4 x float> @floor_nxv4f32(<vscale x 4 x float> %x) strictfp {
; CHECK-LABEL: floor_nxv4f32:
; CHECK: # %bb.0:
-; CHECK-NEXT: ...
[truncated]
``````````
</details>
https://github.com/llvm/llvm-project/pull/93356
More information about the llvm-commits
mailing list