Skip to content

[AArch64] Generate zeroing forms of certain SVE2.2 instructions (9/11) #116835

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Merged
momchil-velikov merged 2 commits into from
Jan 27, 2025
Merged

[AArch64] Generate zeroing forms of certain SVE2.2 instructions (9/11) #116835

momchil-velikov merged 2 commits into from
Jan 27, 2025

Conversation

@momchil-velikov
Copy link
Collaborator

@momchil-velikov momchil-velikov commented Nov 19, 2024

SVE2.2 introduces instructions with predicated forms with zeroing of
the inactive lanes. This allows in some cases to save a movprfx or
a mov instruction when emitting code for _x or _z variants of
intrinsics.

This patch adds support for emitting the zeroing forms of certain
URECPE, URSQRTE, SQABS and SQNEG instructions.

@momchil-velikov momchil-velikov force-pushed the zeroing-09.1-urecpe-ursqrte-sabs-sqneg branch from 8554270 to 6dd2861 Compare December 23, 2024 14:32
@momchil-velikov momchil-velikov force-pushed the zeroing-09.1-urecpe-ursqrte-sabs-sqneg branch from 6dd2861 to c54760e Compare December 24, 2024 10:44
@llvmbot
Copy link
Member

llvmbot commented Dec 24, 2024

@llvm/pr-subscribers-backend-aarch64

Author: Momchil Velikov (momchil-velikov)

Changes

SVE2.2 introduces instructions with predicated forms with zeroing of
the inactive lanes. This allows in some cases to save a movprfx or
a mov instruction when emitting code for _x or _z variants of
intrinsics.

This patch adds support for emitting the zeroing forms of certain
URECPE, URSQRTE, SQABS and SQNEG instructions.

Patch is 35.97 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/116835.diff

3 Files Affected:

  • (modified) llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td (+4-4)
  • (modified) llvm/lib/Target/AArch64/SVEInstrFormats.td (+12-1)
  • (added) llvm/test/CodeGen/AArch64/zeroing-forms-urecpe-ursqrte-sqabs-sqneg.ll (+858)
diff --git a/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td b/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
index 7dd6d49bf20227..c805acd56fff5f 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">;
 
   // 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 0ef862fc1a27cf..7198e151e0eab3 100644
--- a/llvm/lib/Target/AArch64/SVEInstrFormats.td
+++ b/llvm/lib/Target/AArch64/SVEInstrFormats.td
@@ -4180,11 +4180,22 @@ 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>;
+
+  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> {
   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>;
+
+  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
new file mode 100644
index 00000000000000..787ac4458079c6
--- /dev/null
+++ b/llvm/test/CodeGen/AArch64/zeroing-forms-urecpe-ursqrte-sqabs-sqneg.ll
@@ -0,0 +1,858 @@
+; 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> 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_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> 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_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> 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_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> 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_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> 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_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> 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_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> 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_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> 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_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> 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_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> 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_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> 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_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> 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_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> 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_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> 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_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> 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_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> 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_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> 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_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> 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_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> 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_svqneg_s64_x_2(<vscale x 2 x i1> %pg, double %z0, <vscale x 2 x i64> %x) {
+; CHECK-LABE...
[truncated]

jthackray
jthackray approved these changes Jan 20, 2025
View reviewed changes
Copy link
Contributor

@jthackray jthackray left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

Lukacma
Lukacma approved these changes Jan 24, 2025
View reviewed changes
Copy link
Contributor

@Lukacma Lukacma left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

@momchil-velikov momchil-velikov merged commit f95f10c into llvm:main Jan 27, 2025
9 checks passed
@momchil-velikov momchil-velikov deleted the zeroing-09.1-urecpe-ursqrte-sabs-sqneg branch January 29, 2025 10:54
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@jthackray jthackray jthackray approved these changes

@Lukacma Lukacma Lukacma approved these changes

@rgwott rgwott Awaiting requested review from rgwott

@CarolineConcatto CarolineConcatto Awaiting requested review from CarolineConcatto

@tmatheson-arm tmatheson-arm Awaiting requested review from tmatheson-arm

Assignees

No one assigned

Labels

backend:AArch64

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

4 participants

@momchil-velikov @llvmbot @jthackray @Lukacma