[X86][AVX] Fold extract_subvector(truncate(x),0) -> truncate(extract_subvector(x),0)

This is currently only supported for VLX targets where the op should be legal.

diff --git a/llvm/lib/Target/X86/X86ISelLowering.cpp b/llvm/lib/Target/X86/X86ISelLowering.cpp
index a807c51..6abe6c6 100644 (file)
--- a/llvm/lib/Target/X86/X86ISelLowering.cpp
+++ b/llvm/lib/Target/X86/X86ISelLowering.cpp
@@ -48435,6 +48435,14 @@ static SDValue combineExtractSubvector(SDNode *N, SelectionDAG &DAG,
       SDValue Ext2 = extractSubVector(InVec.getOperand(2), 0, DAG, DL, 128);
       return DAG.getNode(InOpcode, DL, VT, Ext0, Ext1, Ext2);
     }
+    if (InOpcode == ISD::TRUNCATE && Subtarget.hasVLX() &&
+        (VT.is128BitVector() || VT.is256BitVector())) {
+      SDLoc DL(N);
+      SDValue InVecSrc = InVec.getOperand(0);
+      unsigned Scale = InVecSrc.getValueSizeInBits() / InSizeInBits;
+      SDValue Ext = extractSubVector(InVecSrc, 0, DAG, DL, Scale * SizeInBits);
+      return DAG.getNode(InOpcode, DL, VT, Ext);
+    }
   }
 
   return SDValue();

diff --git a/llvm/test/CodeGen/X86/vector-reduce-mul.ll b/llvm/test/CodeGen/X86/vector-reduce-mul.ll
index f7e1a72..6e9bfbe 100644 (file)
--- a/llvm/test/CodeGen/X86/vector-reduce-mul.ll
+++ b/llvm/test/CodeGen/X86/vector-reduce-mul.ll
@@ -2008,24 +2008,24 @@ define i8 @test_v32i8(<32 x i8> %a0) {
 ; AVX512BWVL-NEXT:    vextracti128 $1, %ymm0, %xmm1
 ; AVX512BWVL-NEXT:    vpmovzxbw {{.*#+}} zmm1 = ymm1[0],zero,ymm1[1],zero,ymm1[2],zero,ymm1[3],zero,ymm1[4],zero,ymm1[5],zero,ymm1[6],zero,ymm1[7],zero,ymm1[8],zero,ymm1[9],zero,ymm1[10],zero,ymm1[11],zero,ymm1[12],zero,ymm1[13],zero,ymm1[14],zero,ymm1[15],zero,ymm1[16],zero,ymm1[17],zero,ymm1[18],zero,ymm1[19],zero,ymm1[20],zero,ymm1[21],zero,ymm1[22],zero,ymm1[23],zero,ymm1[24],zero,ymm1[25],zero,ymm1[26],zero,ymm1[27],zero,ymm1[28],zero,ymm1[29],zero,ymm1[30],zero,ymm1[31],zero
 ; AVX512BWVL-NEXT:    vpmovzxbw {{.*#+}} zmm0 = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero,ymm0[16],zero,ymm0[17],zero,ymm0[18],zero,ymm0[19],zero,ymm0[20],zero,ymm0[21],zero,ymm0[22],zero,ymm0[23],zero,ymm0[24],zero,ymm0[25],zero,ymm0[26],zero,ymm0[27],zero,ymm0[28],zero,ymm0[29],zero,ymm0[30],zero,ymm0[31],zero
-; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
-; AVX512BWVL-NEXT:    vpmovwb %zmm0, %ymm1
+; AVX512BWVL-NEXT:    vpmullw %ymm1, %ymm0, %ymm0
+; AVX512BWVL-NEXT:    vpmovwb %ymm0, %xmm1
 ; AVX512BWVL-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[2,3,2,3]
 ; AVX512BWVL-NEXT:    vpmovzxbw {{.*#+}} zmm1 = ymm1[0],zero,ymm1[1],zero,ymm1[2],zero,ymm1[3],zero,ymm1[4],zero,ymm1[5],zero,ymm1[6],zero,ymm1[7],zero,ymm1[8],zero,ymm1[9],zero,ymm1[10],zero,ymm1[11],zero,ymm1[12],zero,ymm1[13],zero,ymm1[14],zero,ymm1[15],zero,ymm1[16],zero,ymm1[17],zero,ymm1[18],zero,ymm1[19],zero,ymm1[20],zero,ymm1[21],zero,ymm1[22],zero,ymm1[23],zero,ymm1[24],zero,ymm1[25],zero,ymm1[26],zero,ymm1[27],zero,ymm1[28],zero,ymm1[29],zero,ymm1[30],zero,ymm1[31],zero
-; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
-; AVX512BWVL-NEXT:    vpmovwb %zmm0, %ymm1
+; AVX512BWVL-NEXT:    vpmullw %ymm1, %ymm0, %ymm0
+; AVX512BWVL-NEXT:    vpmovwb %ymm0, %xmm1
 ; AVX512BWVL-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[1,1,2,3]
 ; AVX512BWVL-NEXT:    vpmovzxbw {{.*#+}} zmm1 = ymm1[0],zero,ymm1[1],zero,ymm1[2],zero,ymm1[3],zero,ymm1[4],zero,ymm1[5],zero,ymm1[6],zero,ymm1[7],zero,ymm1[8],zero,ymm1[9],zero,ymm1[10],zero,ymm1[11],zero,ymm1[12],zero,ymm1[13],zero,ymm1[14],zero,ymm1[15],zero,ymm1[16],zero,ymm1[17],zero,ymm1[18],zero,ymm1[19],zero,ymm1[20],zero,ymm1[21],zero,ymm1[22],zero,ymm1[23],zero,ymm1[24],zero,ymm1[25],zero,ymm1[26],zero,ymm1[27],zero,ymm1[28],zero,ymm1[29],zero,ymm1[30],zero,ymm1[31],zero
-; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
-; AVX512BWVL-NEXT:    vpmovwb %zmm0, %ymm1
+; AVX512BWVL-NEXT:    vpmullw %ymm1, %ymm0, %ymm0
+; AVX512BWVL-NEXT:    vpmovwb %ymm0, %xmm1
 ; AVX512BWVL-NEXT:    vpsrld $16, %xmm1, %xmm1
 ; AVX512BWVL-NEXT:    vpmovzxbw {{.*#+}} zmm1 = ymm1[0],zero,ymm1[1],zero,ymm1[2],zero,ymm1[3],zero,ymm1[4],zero,ymm1[5],zero,ymm1[6],zero,ymm1[7],zero,ymm1[8],zero,ymm1[9],zero,ymm1[10],zero,ymm1[11],zero,ymm1[12],zero,ymm1[13],zero,ymm1[14],zero,ymm1[15],zero,ymm1[16],zero,ymm1[17],zero,ymm1[18],zero,ymm1[19],zero,ymm1[20],zero,ymm1[21],zero,ymm1[22],zero,ymm1[23],zero,ymm1[24],zero,ymm1[25],zero,ymm1[26],zero,ymm1[27],zero,ymm1[28],zero,ymm1[29],zero,ymm1[30],zero,ymm1[31],zero
-; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
-; AVX512BWVL-NEXT:    vpmovwb %zmm0, %ymm1
+; AVX512BWVL-NEXT:    vpmullw %ymm1, %ymm0, %ymm0
+; AVX512BWVL-NEXT:    vpmovwb %ymm0, %xmm1
 ; AVX512BWVL-NEXT:    vpsrlw $8, %xmm1, %xmm1
 ; AVX512BWVL-NEXT:    vpmovzxbw {{.*#+}} zmm1 = ymm1[0],zero,ymm1[1],zero,ymm1[2],zero,ymm1[3],zero,ymm1[4],zero,ymm1[5],zero,ymm1[6],zero,ymm1[7],zero,ymm1[8],zero,ymm1[9],zero,ymm1[10],zero,ymm1[11],zero,ymm1[12],zero,ymm1[13],zero,ymm1[14],zero,ymm1[15],zero,ymm1[16],zero,ymm1[17],zero,ymm1[18],zero,ymm1[19],zero,ymm1[20],zero,ymm1[21],zero,ymm1[22],zero,ymm1[23],zero,ymm1[24],zero,ymm1[25],zero,ymm1[26],zero,ymm1[27],zero,ymm1[28],zero,ymm1[29],zero,ymm1[30],zero,ymm1[31],zero
-; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
-; AVX512BWVL-NEXT:    vpmovwb %zmm0, %ymm0
+; AVX512BWVL-NEXT:    vpmullw %ymm1, %ymm0, %ymm0
+; AVX512BWVL-NEXT:    vpmovwb %ymm0, %xmm0
 ; AVX512BWVL-NEXT:    vmovd %xmm0, %eax
 ; AVX512BWVL-NEXT:    # kill: def $al killed $al killed $eax
 ; AVX512BWVL-NEXT:    vzeroupper