[X86] combineConcatVectorOps - concatenation of constant subvectors is free.

RKSimon · RKSimon · commit 85cf2e8286dd · 2023-08-29T14:15:24.000+01:00
diff --git a/llvm/lib/Target/X86/X86ISelLowering.cpp b/llvm/lib/Target/X86/X86ISelLowering.cpp
@@ -54378,15 +54378,19 @@ static SDValue combineConcatVectorOps(const SDLoc &DL, MVT VT,
       return DAG.getNode(ISD::CONCAT_VECTORS, DL, VT, Subs);
     };
     auto IsConcatFree = [](MVT VT, ArrayRef<SDValue> SubOps, unsigned Op) {
+      bool AllConstants = true;
+      bool AllSubVectors = true;
       for (unsigned I = 0, E = SubOps.size(); I != E; ++I) {
         SDValue Sub = SubOps[I].getOperand(Op);
         unsigned NumSubElts = Sub.getValueType().getVectorNumElements();
-        if (Sub.getOpcode() != ISD::EXTRACT_SUBVECTOR ||
-            Sub.getOperand(0).getValueType() != VT ||
-            Sub.getConstantOperandAPInt(1) != (I * NumSubElts))
-          return false;
-      }
-      return true;
+        SDValue BC = peekThroughBitcasts(Sub);
+        AllConstants &= ISD::isBuildVectorOfConstantSDNodes(BC.getNode()) ||
+                        ISD::isBuildVectorOfConstantFPSDNodes(BC.getNode());
+        AllSubVectors &= Sub.getOpcode() == ISD::EXTRACT_SUBVECTOR &&
+                         Sub.getOperand(0).getValueType() == VT &&
+                         Sub.getConstantOperandAPInt(1) == (I * NumSubElts);
+      }
+      return AllConstants || AllSubVectors;
     };
 
     switch (Op0.getOpcode()) {
diff --git a/llvm/test/CodeGen/X86/masked_store_trunc_usat.ll b/llvm/test/CodeGen/X86/masked_store_trunc_usat.ll
@@ -531,28 +531,28 @@ define void @truncstore_v8i64_v8i16(<8 x i64> %x, ptr %p, <8 x i32> %mask) {
 ;
 ; AVX1-LABEL: truncstore_v8i64_v8i16:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm3
-; AVX1-NEXT:    vmovddup {{.*#+}} xmm4 = [9223372036854775808,9223372036854775808]
-; AVX1-NEXT:    # xmm4 = mem[0,0]
-; AVX1-NEXT:    vpxor %xmm4, %xmm3, %xmm5
-; AVX1-NEXT:    vmovddup {{.*#+}} xmm6 = [9223372036854841343,9223372036854841343]
-; AVX1-NEXT:    # xmm6 = mem[0,0]
-; AVX1-NEXT:    vpcmpgtq %xmm5, %xmm6, %xmm5
-; AVX1-NEXT:    vmovddup {{.*#+}} xmm7 = [65535,65535]
-; AVX1-NEXT:    # xmm7 = mem[0,0]
-; AVX1-NEXT:    vblendvpd %xmm5, %xmm3, %xmm7, %xmm3
-; AVX1-NEXT:    vpxor %xmm4, %xmm0, %xmm5
-; AVX1-NEXT:    vpcmpgtq %xmm5, %xmm6, %xmm5
-; AVX1-NEXT:    vblendvpd %xmm5, %xmm0, %xmm7, %xmm0
-; AVX1-NEXT:    vpackusdw %xmm3, %xmm0, %xmm0
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm3
-; AVX1-NEXT:    vpxor %xmm4, %xmm3, %xmm5
-; AVX1-NEXT:    vpcmpgtq %xmm5, %xmm6, %xmm5
-; AVX1-NEXT:    vblendvpd %xmm5, %xmm3, %xmm7, %xmm3
-; AVX1-NEXT:    vpxor %xmm4, %xmm1, %xmm4
-; AVX1-NEXT:    vpcmpgtq %xmm4, %xmm6, %xmm4
-; AVX1-NEXT:    vblendvpd %xmm4, %xmm1, %xmm7, %xmm1
+; AVX1-NEXT:    vmovddup {{.*#+}} xmm3 = [9223372036854775808,9223372036854775808]
+; AVX1-NEXT:    # xmm3 = mem[0,0]
+; AVX1-NEXT:    vpxor %xmm3, %xmm0, %xmm4
+; AVX1-NEXT:    vmovddup {{.*#+}} xmm5 = [9223372036854841343,9223372036854841343]
+; AVX1-NEXT:    # xmm5 = mem[0,0]
+; AVX1-NEXT:    vpcmpgtq %xmm4, %xmm5, %xmm4
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm6
+; AVX1-NEXT:    vpxor %xmm3, %xmm6, %xmm7
+; AVX1-NEXT:    vpcmpgtq %xmm7, %xmm5, %xmm7
+; AVX1-NEXT:    vpxor %xmm3, %xmm1, %xmm8
+; AVX1-NEXT:    vpcmpgtq %xmm8, %xmm5, %xmm8
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm9
+; AVX1-NEXT:    vpxor %xmm3, %xmm9, %xmm3
+; AVX1-NEXT:    vpcmpgtq %xmm3, %xmm5, %xmm3
+; AVX1-NEXT:    vmovddup {{.*#+}} xmm5 = [65535,65535]
+; AVX1-NEXT:    # xmm5 = mem[0,0]
+; AVX1-NEXT:    vblendvpd %xmm3, %xmm9, %xmm5, %xmm3
+; AVX1-NEXT:    vblendvpd %xmm8, %xmm1, %xmm5, %xmm1
 ; AVX1-NEXT:    vpackusdw %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vblendvpd %xmm7, %xmm6, %xmm5, %xmm3
+; AVX1-NEXT:    vblendvpd %xmm4, %xmm0, %xmm5, %xmm0
+; AVX1-NEXT:    vpackusdw %xmm3, %xmm0, %xmm0
 ; AVX1-NEXT:    vpackusdw %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vextractf128 $1, %ymm2, %xmm1
 ; AVX1-NEXT:    vpxor %xmm3, %xmm3, %xmm3
@@ -1003,28 +1003,28 @@ define void @truncstore_v8i64_v8i8(<8 x i64> %x, ptr %p, <8 x i32> %mask) {
 ;
 ; AVX1-LABEL: truncstore_v8i64_v8i8:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm3
-; AVX1-NEXT:    vmovddup {{.*#+}} xmm4 = [9223372036854775808,9223372036854775808]
-; AVX1-NEXT:    # xmm4 = mem[0,0]
-; AVX1-NEXT:    vpxor %xmm4, %xmm3, %xmm5
-; AVX1-NEXT:    vmovddup {{.*#+}} xmm6 = [9223372036854776063,9223372036854776063]
-; AVX1-NEXT:    # xmm6 = mem[0,0]
-; AVX1-NEXT:    vpcmpgtq %xmm5, %xmm6, %xmm5
-; AVX1-NEXT:    vmovddup {{.*#+}} xmm7 = [255,255]
-; AVX1-NEXT:    # xmm7 = mem[0,0]
-; AVX1-NEXT:    vblendvpd %xmm5, %xmm3, %xmm7, %xmm3
-; AVX1-NEXT:    vpxor %xmm4, %xmm0, %xmm5
-; AVX1-NEXT:    vpcmpgtq %xmm5, %xmm6, %xmm5
-; AVX1-NEXT:    vblendvpd %xmm5, %xmm0, %xmm7, %xmm0
-; AVX1-NEXT:    vpackusdw %xmm3, %xmm0, %xmm0
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm3
-; AVX1-NEXT:    vpxor %xmm4, %xmm3, %xmm5
-; AVX1-NEXT:    vpcmpgtq %xmm5, %xmm6, %xmm5
-; AVX1-NEXT:    vblendvpd %xmm5, %xmm3, %xmm7, %xmm3
-; AVX1-NEXT:    vpxor %xmm4, %xmm1, %xmm4
-; AVX1-NEXT:    vpcmpgtq %xmm4, %xmm6, %xmm4
-; AVX1-NEXT:    vblendvpd %xmm4, %xmm1, %xmm7, %xmm1
+; AVX1-NEXT:    vmovddup {{.*#+}} xmm3 = [9223372036854775808,9223372036854775808]
+; AVX1-NEXT:    # xmm3 = mem[0,0]
+; AVX1-NEXT:    vpxor %xmm3, %xmm0, %xmm4
+; AVX1-NEXT:    vmovddup {{.*#+}} xmm5 = [9223372036854776063,9223372036854776063]
+; AVX1-NEXT:    # xmm5 = mem[0,0]
+; AVX1-NEXT:    vpcmpgtq %xmm4, %xmm5, %xmm4
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm6
+; AVX1-NEXT:    vpxor %xmm3, %xmm6, %xmm7
+; AVX1-NEXT:    vpcmpgtq %xmm7, %xmm5, %xmm7
+; AVX1-NEXT:    vpxor %xmm3, %xmm1, %xmm8
+; AVX1-NEXT:    vpcmpgtq %xmm8, %xmm5, %xmm8
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm9
+; AVX1-NEXT:    vpxor %xmm3, %xmm9, %xmm3
+; AVX1-NEXT:    vpcmpgtq %xmm3, %xmm5, %xmm3
+; AVX1-NEXT:    vmovddup {{.*#+}} xmm5 = [255,255]
+; AVX1-NEXT:    # xmm5 = mem[0,0]
+; AVX1-NEXT:    vblendvpd %xmm3, %xmm9, %xmm5, %xmm3
+; AVX1-NEXT:    vblendvpd %xmm8, %xmm1, %xmm5, %xmm1
 ; AVX1-NEXT:    vpackusdw %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vblendvpd %xmm7, %xmm6, %xmm5, %xmm3
+; AVX1-NEXT:    vblendvpd %xmm4, %xmm0, %xmm5, %xmm0
+; AVX1-NEXT:    vpackusdw %xmm3, %xmm0, %xmm0
 ; AVX1-NEXT:    vpackusdw %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vpackuswb %xmm0, %xmm0, %xmm0
 ; AVX1-NEXT:    vextractf128 $1, %ymm2, %xmm1
@@ -1578,19 +1578,19 @@ define void @truncstore_v4i64_v4i16(<4 x i64> %x, ptr %p, <4 x i32> %mask) {
 ; AVX1-LABEL: truncstore_v4i64_v4i16:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm3
-; AVX1-NEXT:    vmovddup {{.*#+}} xmm4 = [9223372036854775808,9223372036854775808]
-; AVX1-NEXT:    # xmm4 = mem[0,0]
-; AVX1-NEXT:    vpxor %xmm4, %xmm3, %xmm5
-; AVX1-NEXT:    vmovddup {{.*#+}} xmm6 = [9223372036854841343,9223372036854841343]
-; AVX1-NEXT:    # xmm6 = mem[0,0]
-; AVX1-NEXT:    vpcmpgtq %xmm5, %xmm6, %xmm5
-; AVX1-NEXT:    vmovddup {{.*#+}} xmm7 = [65535,65535]
-; AVX1-NEXT:    # xmm7 = mem[0,0]
-; AVX1-NEXT:    vblendvpd %xmm5, %xmm3, %xmm7, %xmm3
-; AVX1-NEXT:    vpxor %xmm4, %xmm0, %xmm4
-; AVX1-NEXT:    vpcmpgtq %xmm4, %xmm6, %xmm4
-; AVX1-NEXT:    vblendvpd %xmm4, %xmm0, %xmm7, %xmm0
+; AVX1-NEXT:    vmovddup {{.*#+}} xmm3 = [9223372036854775808,9223372036854775808]
+; AVX1-NEXT:    # xmm3 = mem[0,0]
+; AVX1-NEXT:    vpxor %xmm3, %xmm0, %xmm4
+; AVX1-NEXT:    vmovddup {{.*#+}} xmm5 = [9223372036854841343,9223372036854841343]
+; AVX1-NEXT:    # xmm5 = mem[0,0]
+; AVX1-NEXT:    vpcmpgtq %xmm4, %xmm5, %xmm4
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm6
+; AVX1-NEXT:    vpxor %xmm3, %xmm6, %xmm3
+; AVX1-NEXT:    vpcmpgtq %xmm3, %xmm5, %xmm3
+; AVX1-NEXT:    vmovddup {{.*#+}} xmm5 = [65535,65535]
+; AVX1-NEXT:    # xmm5 = mem[0,0]
+; AVX1-NEXT:    vblendvpd %xmm3, %xmm6, %xmm5, %xmm3
+; AVX1-NEXT:    vblendvpd %xmm4, %xmm0, %xmm5, %xmm0
 ; AVX1-NEXT:    vpackusdw %xmm3, %xmm0, %xmm0
 ; AVX1-NEXT:    vpackusdw %xmm0, %xmm0, %xmm0
 ; AVX1-NEXT:    vpcmpeqd %xmm2, %xmm1, %xmm1
@@ -1860,19 +1860,19 @@ define void @truncstore_v4i64_v4i8(<4 x i64> %x, ptr %p, <4 x i32> %mask) {
 ; AVX1-LABEL: truncstore_v4i64_v4i8:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm3
-; AVX1-NEXT:    vmovddup {{.*#+}} xmm4 = [9223372036854775808,9223372036854775808]
-; AVX1-NEXT:    # xmm4 = mem[0,0]
-; AVX1-NEXT:    vpxor %xmm4, %xmm3, %xmm5
-; AVX1-NEXT:    vmovddup {{.*#+}} xmm6 = [9223372036854776063,9223372036854776063]
-; AVX1-NEXT:    # xmm6 = mem[0,0]
-; AVX1-NEXT:    vpcmpgtq %xmm5, %xmm6, %xmm5
-; AVX1-NEXT:    vmovddup {{.*#+}} xmm7 = [255,255]
-; AVX1-NEXT:    # xmm7 = mem[0,0]
-; AVX1-NEXT:    vblendvpd %xmm5, %xmm3, %xmm7, %xmm3
-; AVX1-NEXT:    vpxor %xmm4, %xmm0, %xmm4
-; AVX1-NEXT:    vpcmpgtq %xmm4, %xmm6, %xmm4
-; AVX1-NEXT:    vblendvpd %xmm4, %xmm0, %xmm7, %xmm0
+; AVX1-NEXT:    vmovddup {{.*#+}} xmm3 = [9223372036854775808,9223372036854775808]
+; AVX1-NEXT:    # xmm3 = mem[0,0]
+; AVX1-NEXT:    vpxor %xmm3, %xmm0, %xmm4
+; AVX1-NEXT:    vmovddup {{.*#+}} xmm5 = [9223372036854776063,9223372036854776063]
+; AVX1-NEXT:    # xmm5 = mem[0,0]
+; AVX1-NEXT:    vpcmpgtq %xmm4, %xmm5, %xmm4
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm6
+; AVX1-NEXT:    vpxor %xmm3, %xmm6, %xmm3
+; AVX1-NEXT:    vpcmpgtq %xmm3, %xmm5, %xmm3
+; AVX1-NEXT:    vmovddup {{.*#+}} xmm5 = [255,255]
+; AVX1-NEXT:    # xmm5 = mem[0,0]
+; AVX1-NEXT:    vblendvpd %xmm3, %xmm6, %xmm5, %xmm3
+; AVX1-NEXT:    vblendvpd %xmm4, %xmm0, %xmm5, %xmm0
 ; AVX1-NEXT:    vpackusdw %xmm3, %xmm0, %xmm0
 ; AVX1-NEXT:    vpackusdw %xmm0, %xmm0, %xmm0
 ; AVX1-NEXT:    vpackuswb %xmm0, %xmm0, %xmm0
diff --git a/llvm/test/CodeGen/X86/vector-trunc-usat.ll b/llvm/test/CodeGen/X86/vector-trunc-usat.ll
