[llvm] 79e3b04 - [AArch64] Add extra addp codegen tests. NFC
David Green via llvm-commits
llvm-commits at lists.llvm.org
Fri Jun 3 03:36:45 PDT 2022
Author: David Green
Date: 2022-06-03T11:36:40+01:00
New Revision: 79e3b043e59af643f2540b12a826c695f054e782
URL: https://github.com/llvm/llvm-project/commit/79e3b043e59af643f2540b12a826c695f054e782
DIFF: https://github.com/llvm/llvm-project/commit/79e3b043e59af643f2540b12a826c695f054e782.diff
LOG: [AArch64] Add extra addp codegen tests. NFC
Added:
Modified:
llvm/test/CodeGen/AArch64/arm64-addp.ll
Removed:
################################################################################
diff --git a/llvm/test/CodeGen/AArch64/arm64-addp.ll b/llvm/test/CodeGen/AArch64/arm64-addp.ll
index 982ce0a73a34..cd7794ff0ccc 100644
--- a/llvm/test/CodeGen/AArch64/arm64-addp.ll
+++ b/llvm/test/CodeGen/AArch64/arm64-addp.ll
@@ -1,9 +1,11 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=arm64-eabi -aarch64-neon-syntax=apple -mcpu=cyclone | FileCheck %s
define double @foo(<2 x double> %a) nounwind {
; CHECK-LABEL: foo:
-; CHECK: faddp.2d d0, v0
-; CHECK-NEXT: ret
+; CHECK: // %bb.0:
+; CHECK-NEXT: faddp.2d d0, v0
+; CHECK-NEXT: ret
%lane0.i = extractelement <2 x double> %a, i32 0
%lane1.i = extractelement <2 x double> %a, i32 1
%vpaddd.i = fadd double %lane0.i, %lane1.i
@@ -12,9 +14,10 @@ define double @foo(<2 x double> %a) nounwind {
define i64 @foo0(<2 x i64> %a) nounwind {
; CHECK-LABEL: foo0:
-; CHECK: addp.2d d0, v0
-; CHECK-NEXT: fmov x0, d0
-; CHECK-NEXT: ret
+; CHECK: // %bb.0:
+; CHECK-NEXT: addp.2d d0, v0
+; CHECK-NEXT: fmov x0, d0
+; CHECK-NEXT: ret
%lane0.i = extractelement <2 x i64> %a, i32 0
%lane1.i = extractelement <2 x i64> %a, i32 1
%vpaddd.i = add i64 %lane0.i, %lane1.i
@@ -23,10 +26,135 @@ define i64 @foo0(<2 x i64> %a) nounwind {
define float @foo1(<2 x float> %a) nounwind {
; CHECK-LABEL: foo1:
-; CHECK: faddp.2s
-; CHECK-NEXT: ret
+; CHECK: // %bb.0:
+; CHECK-NEXT: // kill: def $d0 killed $d0 def $q0
+; CHECK-NEXT: faddp.2s s0, v0
+; CHECK-NEXT: ret
%lane0.i = extractelement <2 x float> %a, i32 0
%lane1.i = extractelement <2 x float> %a, i32 1
%vpaddd.i = fadd float %lane0.i, %lane1.i
ret float %vpaddd.i
}
+
+
+define <2 x i64> @addp_v2i64(<2 x i64> %a) {
+; CHECK-LABEL: addp_v2i64:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ext.16b v1, v0, v0, #8
+; CHECK-NEXT: add.2d v0, v1, v0
+; CHECK-NEXT: ret
+entry:
+ %s = shufflevector <2 x i64> %a, <2 x i64> poison, <2 x i32> <i32 1, i32 0>
+ %b = add <2 x i64> %s, %a
+ ret <2 x i64> %b
+}
+
+define <4 x i64> @addp_v4i64(<4 x i64> %a) {
+; CHECK-LABEL: addp_v4i64:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ext.16b v2, v1, v1, #8
+; CHECK-NEXT: ext.16b v3, v0, v0, #8
+; CHECK-NEXT: add.2d v0, v3, v0
+; CHECK-NEXT: add.2d v1, v2, v1
+; CHECK-NEXT: ret
+entry:
+ %s = shufflevector <4 x i64> %a, <4 x i64> poison, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
+ %b = add <4 x i64> %s, %a
+ ret <4 x i64> %b
+}
+
+define <4 x i32> @addp_v4i32(<4 x i32> %a) {
+; CHECK-LABEL: addp_v4i32:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: rev64.4s v1, v0
+; CHECK-NEXT: add.4s v0, v1, v0
+; CHECK-NEXT: ret
+entry:
+ %s = shufflevector <4 x i32> %a, <4 x i32> poison, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
+ %b = add <4 x i32> %s, %a
+ ret <4 x i32> %b
+}
+
+define <8 x i32> @addp_v8i32(<8 x i32> %a) {
+; CHECK-LABEL: addp_v8i32:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: rev64.4s v2, v1
+; CHECK-NEXT: rev64.4s v3, v0
+; CHECK-NEXT: add.4s v0, v3, v0
+; CHECK-NEXT: add.4s v1, v2, v1
+; CHECK-NEXT: ret
+entry:
+ %s = shufflevector <8 x i32> %a, <8 x i32> poison, <8 x i32> <i32 1, i32 0, i32 3, i32 2, i32 5, i32 4, i32 7, i32 6>
+ %b = add <8 x i32> %s, %a
+ ret <8 x i32> %b
+}
+
+define <16 x i32> @addp_v16i32(<16 x i32> %a) {
+; CHECK-LABEL: addp_v16i32:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: rev64.4s v4, v3
+; CHECK-NEXT: rev64.4s v5, v2
+; CHECK-NEXT: rev64.4s v6, v1
+; CHECK-NEXT: rev64.4s v7, v0
+; CHECK-NEXT: add.4s v0, v7, v0
+; CHECK-NEXT: add.4s v1, v6, v1
+; CHECK-NEXT: add.4s v2, v5, v2
+; CHECK-NEXT: add.4s v3, v4, v3
+; CHECK-NEXT: ret
+entry:
+ %s = shufflevector <16 x i32> %a, <16 x i32> poison, <16 x i32> <i32 1, i32 0, i32 3, i32 2, i32 5, i32 4, i32 7, i32 6, i32 9, i32 8, i32 11, i32 10, i32 13, i32 12, i32 15, i32 14>
+ %b = add <16 x i32> %s, %a
+ ret <16 x i32> %b
+}
+
+define <8 x i16> @addp_v8i16(<8 x i16> %a) {
+; CHECK-LABEL: addp_v8i16:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: rev32.8h v1, v0
+; CHECK-NEXT: add.8h v0, v1, v0
+; CHECK-NEXT: ret
+entry:
+ %s = shufflevector <8 x i16> %a, <8 x i16> poison, <8 x i32> <i32 1, i32 0, i32 3, i32 2, i32 5, i32 4, i32 7, i32 6>
+ %b = add <8 x i16> %s, %a
+ ret <8 x i16> %b
+}
+
+define <16 x i16> @addp_v16i16(<16 x i16> %a) {
+; CHECK-LABEL: addp_v16i16:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: rev32.8h v2, v1
+; CHECK-NEXT: rev32.8h v3, v0
+; CHECK-NEXT: add.8h v0, v3, v0
+; CHECK-NEXT: add.8h v1, v2, v1
+; CHECK-NEXT: ret
+entry:
+ %s = shufflevector <16 x i16> %a, <16 x i16> poison, <16 x i32> <i32 1, i32 0, i32 3, i32 2, i32 5, i32 4, i32 7, i32 6, i32 9, i32 8, i32 11, i32 10, i32 13, i32 12, i32 15, i32 14>
+ %b = add <16 x i16> %s, %a
+ ret <16 x i16> %b
+}
+
+define <16 x i8> @addp_v16i8(<16 x i8> %a) {
+; CHECK-LABEL: addp_v16i8:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: rev16.16b v1, v0
+; CHECK-NEXT: add.16b v0, v1, v0
+; CHECK-NEXT: ret
+entry:
+ %s = shufflevector <16 x i8> %a, <16 x i8> poison, <16 x i32> <i32 1, i32 0, i32 3, i32 2, i32 5, i32 4, i32 7, i32 6, i32 9, i32 8, i32 11, i32 10, i32 13, i32 12, i32 15, i32 14>
+ %b = add <16 x i8> %s, %a
+ ret <16 x i8> %b
+}
+
+define <32 x i8> @addp_v32i8(<32 x i8> %a) {
+; CHECK-LABEL: addp_v32i8:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: rev16.16b v2, v1
+; CHECK-NEXT: rev16.16b v3, v0
+; CHECK-NEXT: add.16b v0, v3, v0
+; CHECK-NEXT: add.16b v1, v2, v1
+; CHECK-NEXT: ret
+entry:
+ %s = shufflevector <32 x i8> %a, <32 x i8> poison, <32 x i32> <i32 1, i32 0, i32 3, i32 2, i32 5, i32 4, i32 7, i32 6, i32 9, i32 8, i32 11, i32 10, i32 13, i32 12, i32 15, i32 14, i32 17, i32 16, i32 19, i32 18, i32 21, i32 20, i32 23, i32 22, i32 25, i32 24, i32 27, i32 26, i32 29, i32 28, i32 31, i32 30>
+ %b = add <32 x i8> %s, %a
+ ret <32 x i8> %b
+}
More information about the llvm-commits
mailing list