[llvm] 57c083e - [X86] combine-pmadd.ll - add tests for concatenations of pmaddwd/pmaddubsw subvectors

[Simon Pilgrim via llvm-commits]

Author: Simon Pilgrim
Date: 2024-06-21T11:22:21+01:00
New Revision: 57c083ecfb63a15656d974d097793c709c6552c4

URL: https://github.com/llvm/llvm-project/commit/57c083ecfb63a15656d974d097793c709c6552c4
DIFF: https://github.com/llvm/llvm-project/commit/57c083ecfb63a15656d974d097793c709c6552c4.diff

LOG: [X86] combine-pmadd.ll - add tests for concatenations of pmaddwd/pmaddubsw subvectors

Added: 
    

Modified: 
    llvm/test/CodeGen/X86/combine-pmadd.ll

Removed: 
    


################################################################################
diff  --git a/llvm/test/CodeGen/X86/combine-pmadd.ll b/llvm/test/CodeGen/X86/combine-pmadd.ll
index ad3b9e645f628..c4a28fe2eda34 100644
--- a/llvm/test/CodeGen/X86/combine-pmadd.ll
+++ b/llvm/test/CodeGen/X86/combine-pmadd.ll
@@ -1,7 +1,7 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse4.1 | FileCheck %s --check-prefixes=CHECK,SSE
-; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx | FileCheck %s --check-prefixes=CHECK,AVX
-; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefixes=CHECK,AVX
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx | FileCheck %s --check-prefixes=CHECK,AVX,AVX1
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefixes=CHECK,AVX,AVX2
 
 declare <4 x i32> @llvm.x86.sse2.pmadd.wd(<8 x i16>, <8 x i16>) nounwind readnone
 declare <8 x i16> @llvm.x86.ssse3.pmadd.ub.sw.128(<16 x i8>, <16 x i8>) nounwind readnone
@@ -34,6 +34,33 @@ define <4 x i32> @combine_pmaddwd_zero_commute(<8 x i16> %a0, <8 x i16> %a1) {
   ret <4 x i32> %1
 }
 
+define <8 x i32> @combine_pmaddwd_concat(<8 x i16> %a0, <8 x i16> %a1, <8 x i16> %a2, <8 x i16> %a3) {
+; SSE-LABEL: combine_pmaddwd_concat:
+; SSE:       # %bb.0:
+; SSE-NEXT:    pmaddwd %xmm1, %xmm0
+; SSE-NEXT:    pmaddwd %xmm3, %xmm2
+; SSE-NEXT:    movdqa %xmm2, %xmm1
+; SSE-NEXT:    retq
+;
+; AVX1-LABEL: combine_pmaddwd_concat:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vpmaddwd %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpmaddwd %xmm3, %xmm2, %xmm1
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: combine_pmaddwd_concat:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vpmaddwd %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    vpmaddwd %xmm3, %xmm2, %xmm1
+; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
+; AVX2-NEXT:    retq
+  %1 = call <4 x i32> @llvm.x86.sse2.pmadd.wd(<8 x i16> %a0, <8 x i16> %a1)
+  %2 = call <4 x i32> @llvm.x86.sse2.pmadd.wd(<8 x i16> %a2, <8 x i16> %a3)
+  %3 = shufflevector <4 x i32> %1, <4 x i32> %2, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+  ret <8 x i32> %3
+}
+
 define i32 @combine_pmaddwd_constant() {
 ; CHECK-LABEL: combine_pmaddwd_constant:
 ; CHECK:       # %bb.0:
@@ -72,6 +99,33 @@ define <8 x i16> @combine_pmaddubsw_zero_commute(<16 x i8> %a0, <16 x i8> %a1) {
   ret <8 x i16> %1
 }
 
+define <16 x i16> @combine_pmaddubsw_concat(<16 x i8> %a0, <16 x i8> %a1, <16 x i8> %a2, <16 x i8> %a3) {
+; SSE-LABEL: combine_pmaddubsw_concat:
+; SSE:       # %bb.0:
+; SSE-NEXT:    pmaddubsw %xmm1, %xmm0
+; SSE-NEXT:    pmaddubsw %xmm3, %xmm2
+; SSE-NEXT:    movdqa %xmm2, %xmm1
+; SSE-NEXT:    retq
+;
+; AVX1-LABEL: combine_pmaddubsw_concat:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vpmaddubsw %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpmaddubsw %xmm3, %xmm2, %xmm1
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: combine_pmaddubsw_concat:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vpmaddubsw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    vpmaddubsw %xmm3, %xmm2, %xmm1
+; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
+; AVX2-NEXT:    retq
+  %1 = call <8 x i16> @llvm.x86.ssse3.pmadd.ub.sw.128(<16 x i8> %a0, <16 x i8> %a1)
+  %2 = call <8 x i16> @llvm.x86.ssse3.pmadd.ub.sw.128(<16 x i8> %a2, <16 x i8> %a3)
+  %3 = shufflevector <8 x i16> %1, <8 x i16> %2, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+  ret <16 x i16> %3
+}
+
 define i32 @combine_pmaddubsw_constant() {
 ; CHECK-LABEL: combine_pmaddubsw_constant:
 ; CHECK:       # %bb.0: