[llvm] b31a531 - [AArch64][SVE2] Implement while comparison intrinsics
Cullen Rhodes via llvm-commits
llvm-commits at lists.llvm.org
Fri Dec 6 03:39:28 PST 2019
Author: Cullen Rhodes
Date: 2019-12-06T11:29:34Z
New Revision: b31a531f9bbf7a3ae193df6009f5ee542f760017
URL: https://github.com/llvm/llvm-project/commit/b31a531f9bbf7a3ae193df6009f5ee542f760017
DIFF: https://github.com/llvm/llvm-project/commit/b31a531f9bbf7a3ae193df6009f5ee542f760017.diff
LOG: [AArch64][SVE2] Implement while comparison intrinsics
Summary:
Adds the following intrinsics:
* whilege, whilegt, whilehi, whilehs
Reviewers: sdesmalen, rovka, dancgr, efriedma, rengolin, huntergr
Reviewed By: sdesmalen
Subscribers: tschuett, kristof.beyls, hiraditya, rkruppe, psnobl, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D70909
Added:
llvm/test/CodeGen/AArch64/sve2-intrinsics-while.ll
Modified:
llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
Removed:
################################################################################
diff --git a/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td b/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
index e374ce4edd7e..571dc775e879 100644
--- a/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
+++ b/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
@@ -1521,16 +1521,15 @@ let Predicates = [HasSVE2] in {
defm TBX_ZZZ : sve2_int_perm_tbx<"tbx">;
// SVE2 integer compare scalar count and limit
- defm WHILEGE_PWW : sve_int_while4_rr<0b000, "whilege", null_frag>;
- defm WHILEGT_PWW : sve_int_while4_rr<0b001, "whilegt", null_frag>;
- defm WHILEHS_PWW : sve_int_while4_rr<0b100, "whilehs", null_frag>;
- defm WHILEHI_PWW : sve_int_while4_rr<0b101, "whilehi", null_frag>;
-
- defm WHILEGE_PXX : sve_int_while8_rr<0b000, "whilege", null_frag>;
- defm WHILEGT_PXX : sve_int_while8_rr<0b001, "whilegt", null_frag>;
- defm WHILEHS_PXX : sve_int_while8_rr<0b100, "whilehs", null_frag>;
- defm WHILEHI_PXX : sve_int_while8_rr<0b101, "whilehi", null_frag>;
-
+ defm WHILEGE_PWW : sve_int_while4_rr<0b000, "whilege", int_aarch64_sve_whilege>;
+ defm WHILEGT_PWW : sve_int_while4_rr<0b001, "whilegt", int_aarch64_sve_whilegt>;
+ defm WHILEHS_PWW : sve_int_while4_rr<0b100, "whilehs", int_aarch64_sve_whilehs>;
+ defm WHILEHI_PWW : sve_int_while4_rr<0b101, "whilehi", int_aarch64_sve_whilehi>;
+
+ defm WHILEGE_PXX : sve_int_while8_rr<0b000, "whilege", int_aarch64_sve_whilege>;
+ defm WHILEGT_PXX : sve_int_while8_rr<0b001, "whilegt", int_aarch64_sve_whilegt>;
+ defm WHILEHS_PXX : sve_int_while8_rr<0b100, "whilehs", int_aarch64_sve_whilehs>;
+ defm WHILEHI_PXX : sve_int_while8_rr<0b101, "whilehi", int_aarch64_sve_whilehi>;
// SVE2 pointer conflict compare
defm WHILEWR_PXX : sve2_int_while_rr<0b0, "whilewr">;
diff --git a/llvm/test/CodeGen/AArch64/sve2-intrinsics-while.ll b/llvm/test/CodeGen/AArch64/sve2-intrinsics-while.ll
new file mode 100644
index 000000000000..e53c981b64d8
--- /dev/null
+++ b/llvm/test/CodeGen/AArch64/sve2-intrinsics-while.ll
@@ -0,0 +1,309 @@
+; RUN: llc -mtriple=aarch64-linux-gnu -mattr=+sve2 < %s | FileCheck %s
+
+;
+; WHILEGE
+;
+
+define <vscale x 16 x i1> @whilege_b_ww(i32 %a, i32 %b) {
+; CHECK-LABEL: whilege_b_ww:
+; CHECK: whilege p0.b, w0, w1
+; CHECK-NEXT: ret
+ %out = call <vscale x 16 x i1> @llvm.aarch64.sve.whilege.nxv16i1.i32(i32 %a, i32 %b)
+ ret <vscale x 16 x i1> %out
+}
+
+define <vscale x 16 x i1> @whilege_b_xx(i64 %a, i64 %b) {
+; CHECK-LABEL: whilege_b_xx:
+; CHECK: whilege p0.b, x0, x1
+; CHECK-NEXT: ret
+ %out = call <vscale x 16 x i1> @llvm.aarch64.sve.whilege.nxv16i1.i64(i64 %a, i64 %b)
+ ret <vscale x 16 x i1> %out
+}
+
+define <vscale x 8 x i1> @whilege_h_ww(i32 %a, i32 %b) {
+; CHECK-LABEL: whilege_h_ww:
+; CHECK: whilege p0.h, w0, w1
+; CHECK-NEXT: ret
+ %out = call <vscale x 8 x i1> @llvm.aarch64.sve.whilege.nxv8i1.i32(i32 %a, i32 %b)
+ ret <vscale x 8 x i1> %out
+}
+
+define <vscale x 8 x i1> @whilege_h_xx(i64 %a, i64 %b) {
+; CHECK-LABEL: whilege_h_xx:
+; CHECK: whilege p0.h, x0, x1
+; CHECK-NEXT: ret
+ %out = call <vscale x 8 x i1> @llvm.aarch64.sve.whilege.nxv8i1.i64(i64 %a, i64 %b)
+ ret <vscale x 8 x i1> %out
+}
+
+define <vscale x 4 x i1> @whilege_s_ww(i32 %a, i32 %b) {
+; CHECK-LABEL: whilege_s_ww:
+; CHECK: whilege p0.s, w0, w1
+; CHECK-NEXT: ret
+ %out = call <vscale x 4 x i1> @llvm.aarch64.sve.whilege.nxv4i1.i32(i32 %a, i32 %b)
+ ret <vscale x 4 x i1> %out
+}
+
+define <vscale x 4 x i1> @whilege_s_xx(i64 %a, i64 %b) {
+; CHECK-LABEL: whilege_s_xx:
+; CHECK: whilege p0.s, x0, x1
+; CHECK-NEXT: ret
+ %out = call <vscale x 4 x i1> @llvm.aarch64.sve.whilege.nxv4i1.i64(i64 %a, i64 %b)
+ ret <vscale x 4 x i1> %out
+}
+
+define <vscale x 2 x i1> @whilege_d_ww(i32 %a, i32 %b) {
+; CHECK-LABEL: whilege_d_ww:
+; CHECK: whilege p0.d, w0, w1
+; CHECK-NEXT: ret
+ %out = call <vscale x 2 x i1> @llvm.aarch64.sve.whilege.nxv2i1.i32(i32 %a, i32 %b)
+ ret <vscale x 2 x i1> %out
+}
+
+define <vscale x 2 x i1> @whilege_d_xx(i64 %a, i64 %b) {
+; CHECK-LABEL: whilege_d_xx:
+; CHECK: whilege p0.d, x0, x1
+; CHECK-NEXT: ret
+ %out = call <vscale x 2 x i1> @llvm.aarch64.sve.whilege.nxv2i1.i64(i64 %a, i64 %b)
+ ret <vscale x 2 x i1> %out
+}
+
+;
+; WHILEHS
+;
+
+define <vscale x 16 x i1> @whilehs_b_ww(i32 %a, i32 %b) {
+; CHECK-LABEL: whilehs_b_ww:
+; CHECK: whilehs p0.b, w0, w1
+; CHECK-NEXT: ret
+ %out = call <vscale x 16 x i1> @llvm.aarch64.sve.whilehs.nxv16i1.i32(i32 %a, i32 %b)
+ ret <vscale x 16 x i1> %out
+}
+
+define <vscale x 16 x i1> @whilehs_b_xx(i64 %a, i64 %b) {
+; CHECK-LABEL: whilehs_b_xx:
+; CHECK: whilehs p0.b, x0, x1
+; CHECK-NEXT: ret
+ %out = call <vscale x 16 x i1> @llvm.aarch64.sve.whilehs.nxv16i1.i64(i64 %a, i64 %b)
+ ret <vscale x 16 x i1> %out
+}
+
+define <vscale x 8 x i1> @whilehs_h_ww(i32 %a, i32 %b) {
+; CHECK-LABEL: whilehs_h_ww:
+; CHECK: whilehs p0.h, w0, w1
+; CHECK-NEXT: ret
+ %out = call <vscale x 8 x i1> @llvm.aarch64.sve.whilehs.nxv8i1.i32(i32 %a, i32 %b)
+ ret <vscale x 8 x i1> %out
+}
+
+define <vscale x 8 x i1> @whilehs_h_xx(i64 %a, i64 %b) {
+; CHECK-LABEL: whilehs_h_xx:
+; CHECK: whilehs p0.h, x0, x1
+; CHECK-NEXT: ret
+ %out = call <vscale x 8 x i1> @llvm.aarch64.sve.whilehs.nxv8i1.i64(i64 %a, i64 %b)
+ ret <vscale x 8 x i1> %out
+}
+
+define <vscale x 4 x i1> @whilehs_s_ww(i32 %a, i32 %b) {
+; CHECK-LABEL: whilehs_s_ww:
+; CHECK: whilehs p0.s, w0, w1
+; CHECK-NEXT: ret
+ %out = call <vscale x 4 x i1> @llvm.aarch64.sve.whilehs.nxv4i1.i32(i32 %a, i32 %b)
+ ret <vscale x 4 x i1> %out
+}
+
+define <vscale x 4 x i1> @whilehs_s_xx(i64 %a, i64 %b) {
+; CHECK-LABEL: whilehs_s_xx:
+; CHECK: whilehs p0.s, x0, x1
+; CHECK-NEXT: ret
+ %out = call <vscale x 4 x i1> @llvm.aarch64.sve.whilehs.nxv4i1.i64(i64 %a, i64 %b)
+ ret <vscale x 4 x i1> %out
+}
+
+define <vscale x 2 x i1> @whilehs_d_ww(i32 %a, i32 %b) {
+; CHECK-LABEL: whilehs_d_ww:
+; CHECK: whilehs p0.d, w0, w1
+; CHECK-NEXT: ret
+ %out = call <vscale x 2 x i1> @llvm.aarch64.sve.whilehs.nxv2i1.i32(i32 %a, i32 %b)
+ ret <vscale x 2 x i1> %out
+}
+
+define <vscale x 2 x i1> @whilehs_d_xx(i64 %a, i64 %b) {
+; CHECK-LABEL: whilehs_d_xx:
+; CHECK: whilehs p0.d, x0, x1
+; CHECK-NEXT: ret
+ %out = call <vscale x 2 x i1> @llvm.aarch64.sve.whilehs.nxv2i1.i64(i64 %a, i64 %b)
+ ret <vscale x 2 x i1> %out
+}
+
+;
+; WHILEGT
+;
+
+define <vscale x 16 x i1> @whilegt_b_ww(i32 %a, i32 %b) {
+; CHECK-LABEL: whilegt_b_ww:
+; CHECK: whilegt p0.b, w0, w1
+; CHECK-NEXT: ret
+ %out = call <vscale x 16 x i1> @llvm.aarch64.sve.whilegt.nxv16i1.i32(i32 %a, i32 %b)
+ ret <vscale x 16 x i1> %out
+}
+
+define <vscale x 16 x i1> @whilegt_b_xx(i64 %a, i64 %b) {
+; CHECK-LABEL: whilegt_b_xx:
+; CHECK: whilegt p0.b, x0, x1
+; CHECK-NEXT: ret
+ %out = call <vscale x 16 x i1> @llvm.aarch64.sve.whilegt.nxv16i1.i64(i64 %a, i64 %b)
+ ret <vscale x 16 x i1> %out
+}
+
+define <vscale x 8 x i1> @whilegt_h_ww(i32 %a, i32 %b) {
+; CHECK-LABEL: whilegt_h_ww:
+; CHECK: whilegt p0.h, w0, w1
+; CHECK-NEXT: ret
+ %out = call <vscale x 8 x i1> @llvm.aarch64.sve.whilegt.nxv8i1.i32(i32 %a, i32 %b)
+ ret <vscale x 8 x i1> %out
+}
+
+define <vscale x 8 x i1> @whilegt_h_xx(i64 %a, i64 %b) {
+; CHECK-LABEL: whilegt_h_xx:
+; CHECK: whilegt p0.h, x0, x1
+; CHECK-NEXT: ret
+ %out = call <vscale x 8 x i1> @llvm.aarch64.sve.whilegt.nxv8i1.i64(i64 %a, i64 %b)
+ ret <vscale x 8 x i1> %out
+}
+
+define <vscale x 4 x i1> @whilegt_s_ww(i32 %a, i32 %b) {
+; CHECK-LABEL: whilegt_s_ww:
+; CHECK: whilegt p0.s, w0, w1
+; CHECK-NEXT: ret
+ %out = call <vscale x 4 x i1> @llvm.aarch64.sve.whilegt.nxv4i1.i32(i32 %a, i32 %b)
+ ret <vscale x 4 x i1> %out
+}
+
+define <vscale x 4 x i1> @whilegt_s_xx(i64 %a, i64 %b) {
+; CHECK-LABEL: whilegt_s_xx:
+; CHECK: whilegt p0.s, x0, x1
+; CHECK-NEXT: ret
+ %out = call <vscale x 4 x i1> @llvm.aarch64.sve.whilegt.nxv4i1.i64(i64 %a, i64 %b)
+ ret <vscale x 4 x i1> %out
+}
+
+define <vscale x 2 x i1> @whilegt_d_ww(i32 %a, i32 %b) {
+; CHECK-LABEL: whilegt_d_ww:
+; CHECK: whilegt p0.d, w0, w1
+; CHECK-NEXT: ret
+ %out = call <vscale x 2 x i1> @llvm.aarch64.sve.whilegt.nxv2i1.i32(i32 %a, i32 %b)
+ ret <vscale x 2 x i1> %out
+}
+
+define <vscale x 2 x i1> @whilegt_d_xx(i64 %a, i64 %b) {
+; CHECK-LABEL: whilegt_d_xx:
+; CHECK: whilegt p0.d, x0, x1
+; CHECK-NEXT: ret
+ %out = call <vscale x 2 x i1> @llvm.aarch64.sve.whilegt.nxv2i1.i64(i64 %a, i64 %b)
+ ret <vscale x 2 x i1> %out
+}
+
+;
+; WHILEHI
+;
+
+define <vscale x 16 x i1> @whilehi_b_ww(i32 %a, i32 %b) {
+; CHECK-LABEL: whilehi_b_ww:
+; CHECK: whilehi p0.b, w0, w1
+; CHECK-NEXT: ret
+ %out = call <vscale x 16 x i1> @llvm.aarch64.sve.whilehi.nxv16i1.i32(i32 %a, i32 %b)
+ ret <vscale x 16 x i1> %out
+}
+
+define <vscale x 16 x i1> @whilehi_b_xx(i64 %a, i64 %b) {
+; CHECK-LABEL: whilehi_b_xx:
+; CHECK: whilehi p0.b, x0, x1
+; CHECK-NEXT: ret
+ %out = call <vscale x 16 x i1> @llvm.aarch64.sve.whilehi.nxv16i1.i64(i64 %a, i64 %b)
+ ret <vscale x 16 x i1> %out
+}
+
+define <vscale x 8 x i1> @whilehi_h_ww(i32 %a, i32 %b) {
+; CHECK-LABEL: whilehi_h_ww:
+; CHECK: whilehi p0.h, w0, w1
+; CHECK-NEXT: ret
+ %out = call <vscale x 8 x i1> @llvm.aarch64.sve.whilehi.nxv8i1.i32(i32 %a, i32 %b)
+ ret <vscale x 8 x i1> %out
+}
+
+define <vscale x 8 x i1> @whilehi_h_xx(i64 %a, i64 %b) {
+; CHECK-LABEL: whilehi_h_xx:
+; CHECK: whilehi p0.h, x0, x1
+; CHECK-NEXT: ret
+ %out = call <vscale x 8 x i1> @llvm.aarch64.sve.whilehi.nxv8i1.i64(i64 %a, i64 %b)
+ ret <vscale x 8 x i1> %out
+}
+
+define <vscale x 4 x i1> @whilehi_s_ww(i32 %a, i32 %b) {
+; CHECK-LABEL: whilehi_s_ww:
+; CHECK: whilehi p0.s, w0, w1
+; CHECK-NEXT: ret
+ %out = call <vscale x 4 x i1> @llvm.aarch64.sve.whilehi.nxv4i1.i32(i32 %a, i32 %b)
+ ret <vscale x 4 x i1> %out
+}
+
+define <vscale x 4 x i1> @whilehi_s_xx(i64 %a, i64 %b) {
+; CHECK-LABEL: whilehi_s_xx:
+; CHECK: whilehi p0.s, x0, x1
+; CHECK-NEXT: ret
+ %out = call <vscale x 4 x i1> @llvm.aarch64.sve.whilehi.nxv4i1.i64(i64 %a, i64 %b)
+ ret <vscale x 4 x i1> %out
+}
+
+define <vscale x 2 x i1> @whilehi_d_ww(i32 %a, i32 %b) {
+; CHECK-LABEL: whilehi_d_ww:
+; CHECK: whilehi p0.d, w0, w1
+; CHECK-NEXT: ret
+ %out = call <vscale x 2 x i1> @llvm.aarch64.sve.whilehi.nxv2i1.i32(i32 %a, i32 %b)
+ ret <vscale x 2 x i1> %out
+}
+
+define <vscale x 2 x i1> @whilehi_d_xx(i64 %a, i64 %b) {
+; CHECK-LABEL: whilehi_d_xx:
+; CHECK: whilehi p0.d, x0, x1
+; CHECK-NEXT: ret
+ %out = call <vscale x 2 x i1> @llvm.aarch64.sve.whilehi.nxv2i1.i64(i64 %a, i64 %b)
+ ret <vscale x 2 x i1> %out
+}
+
+declare <vscale x 16 x i1> @llvm.aarch64.sve.whilege.nxv16i1.i32(i32, i32)
+declare <vscale x 16 x i1> @llvm.aarch64.sve.whilege.nxv16i1.i64(i64, i64)
+declare <vscale x 8 x i1> @llvm.aarch64.sve.whilege.nxv8i1.i32(i32, i32)
+declare <vscale x 8 x i1> @llvm.aarch64.sve.whilege.nxv8i1.i64(i64, i64)
+declare <vscale x 4 x i1> @llvm.aarch64.sve.whilege.nxv4i1.i32(i32, i32)
+declare <vscale x 4 x i1> @llvm.aarch64.sve.whilege.nxv4i1.i64(i64, i64)
+declare <vscale x 2 x i1> @llvm.aarch64.sve.whilege.nxv2i1.i32(i32, i32)
+declare <vscale x 2 x i1> @llvm.aarch64.sve.whilege.nxv2i1.i64(i64, i64)
+
+declare <vscale x 16 x i1> @llvm.aarch64.sve.whilehs.nxv16i1.i32(i32, i32)
+declare <vscale x 16 x i1> @llvm.aarch64.sve.whilehs.nxv16i1.i64(i64, i64)
+declare <vscale x 8 x i1> @llvm.aarch64.sve.whilehs.nxv8i1.i32(i32, i32)
+declare <vscale x 8 x i1> @llvm.aarch64.sve.whilehs.nxv8i1.i64(i64, i64)
+declare <vscale x 4 x i1> @llvm.aarch64.sve.whilehs.nxv4i1.i32(i32, i32)
+declare <vscale x 4 x i1> @llvm.aarch64.sve.whilehs.nxv4i1.i64(i64, i64)
+declare <vscale x 2 x i1> @llvm.aarch64.sve.whilehs.nxv2i1.i32(i32, i32)
+declare <vscale x 2 x i1> @llvm.aarch64.sve.whilehs.nxv2i1.i64(i64, i64)
+
+declare <vscale x 16 x i1> @llvm.aarch64.sve.whilegt.nxv16i1.i32(i32, i32)
+declare <vscale x 16 x i1> @llvm.aarch64.sve.whilegt.nxv16i1.i64(i64, i64)
+declare <vscale x 8 x i1> @llvm.aarch64.sve.whilegt.nxv8i1.i32(i32, i32)
+declare <vscale x 8 x i1> @llvm.aarch64.sve.whilegt.nxv8i1.i64(i64, i64)
+declare <vscale x 4 x i1> @llvm.aarch64.sve.whilegt.nxv4i1.i32(i32, i32)
+declare <vscale x 4 x i1> @llvm.aarch64.sve.whilegt.nxv4i1.i64(i64, i64)
+declare <vscale x 2 x i1> @llvm.aarch64.sve.whilegt.nxv2i1.i32(i32, i32)
+declare <vscale x 2 x i1> @llvm.aarch64.sve.whilegt.nxv2i1.i64(i64, i64)
+
+declare <vscale x 16 x i1> @llvm.aarch64.sve.whilehi.nxv16i1.i32(i32, i32)
+declare <vscale x 16 x i1> @llvm.aarch64.sve.whilehi.nxv16i1.i64(i64, i64)
+declare <vscale x 8 x i1> @llvm.aarch64.sve.whilehi.nxv8i1.i32(i32, i32)
+declare <vscale x 8 x i1> @llvm.aarch64.sve.whilehi.nxv8i1.i64(i64, i64)
+declare <vscale x 4 x i1> @llvm.aarch64.sve.whilehi.nxv4i1.i32(i32, i32)
+declare <vscale x 4 x i1> @llvm.aarch64.sve.whilehi.nxv4i1.i64(i64, i64)
+declare <vscale x 2 x i1> @llvm.aarch64.sve.whilehi.nxv2i1.i32(i32, i32)
+declare <vscale x 2 x i1> @llvm.aarch64.sve.whilehi.nxv2i1.i64(i64, i64)
More information about the llvm-commits
mailing list