[X86][AVX] Reuse existing VBROADCAST(x) for SCALAR_TO_VECTOR(x)

Similar to what we already do for BROADCASTs of different vector sizes - if we're going to broadcast it anyway might as well reuse it.

diff --git a/llvm/lib/Target/X86/X86ISelLowering.cpp b/llvm/lib/Target/X86/X86ISelLowering.cpp
index c965b7a..86052fa 100644 (file)
--- a/llvm/lib/Target/X86/X86ISelLowering.cpp
+++ b/llvm/lib/Target/X86/X86ISelLowering.cpp
@@ -49880,6 +49880,14 @@ static SDValue combineScalarToVector(SDNode *N, SelectionDAG &DAG) {
       Src.getOperand(0).getValueType() == MVT::x86mmx)
     return DAG.getNode(X86ISD::MOVQ2DQ, DL, VT, Src.getOperand(0));
 
+  // See if we're broadcasting the scalar value, in which case just reuse that.
+  // Ensure the same SDValue from the SDNode use is being used.
+  // TODO: Handle different vector sizes when we have test coverage.
+  for (SDNode *User : Src->uses())
+    if (User->getOpcode() == X86ISD::VBROADCAST && Src == User->getOperand(0) &&
+        User->getValueSizeInBits(0).getFixedSize() == VT.getFixedSizeInBits())
+      return SDValue(User, 0);
+
   return SDValue();
 }
 

diff --git a/llvm/test/CodeGen/X86/combine-bitselect.ll b/llvm/test/CodeGen/X86/combine-bitselect.ll
index 97a2ef2..d57bd87 100644 (file)
--- a/llvm/test/CodeGen/X86/combine-bitselect.ll
+++ b/llvm/test/CodeGen/X86/combine-bitselect.ll
@@ -253,29 +253,17 @@ define <2 x i64> @bitselect_v2i64_broadcast_rrm(<2 x i64> %a0, <2 x i64> %a1, i6
 ;
 ; XOP-LABEL: bitselect_v2i64_broadcast_rrm:
 ; XOP:       # %bb.0:
-; XOP-NEXT:    vmovsd {{.*#+}} xmm2 = mem[0],zero
-; XOP-NEXT:    vpermilps {{.*#+}} xmm2 = xmm2[0,1,0,1]
-; XOP-NEXT:    vandps %xmm2, %xmm0, %xmm0
-; XOP-NEXT:    vandnps %xmm1, %xmm2, %xmm1
-; XOP-NEXT:    vorps %xmm1, %xmm0, %xmm0
+; XOP-NEXT:    vmovddup {{.*#+}} xmm2 = mem[0,0]
+; XOP-NEXT:    vpcmov %xmm2, %xmm1, %xmm0, %xmm0
 ; XOP-NEXT:    retq
 ;
-; AVX1-LABEL: bitselect_v2i64_broadcast_rrm:
-; AVX1:       # %bb.0:
-; AVX1-NEXT:    vmovsd {{.*#+}} xmm2 = mem[0],zero
-; AVX1-NEXT:    vpermilps {{.*#+}} xmm2 = xmm2[0,1,0,1]
-; AVX1-NEXT:    vandps %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vandnps %xmm1, %xmm2, %xmm1
-; AVX1-NEXT:    vorps %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    retq
-;
-; AVX2-LABEL: bitselect_v2i64_broadcast_rrm:
-; AVX2:       # %bb.0:
-; AVX2-NEXT:    vmovddup {{.*#+}} xmm2 = mem[0,0]
-; AVX2-NEXT:    vandps %xmm2, %xmm0, %xmm0
-; AVX2-NEXT:    vandnps %xmm1, %xmm2, %xmm1
-; AVX2-NEXT:    vorps %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    retq
+; AVX-LABEL: bitselect_v2i64_broadcast_rrm:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vmovddup {{.*#+}} xmm2 = mem[0,0]
+; AVX-NEXT:    vandps %xmm2, %xmm0, %xmm0
+; AVX-NEXT:    vandnps %xmm1, %xmm2, %xmm1
+; AVX-NEXT:    vorps %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    retq
 ;
 ; AVX512F-LABEL: bitselect_v2i64_broadcast_rrm:
 ; AVX512F:       # %bb.0: