Synthesize x86 max/min instructions also for vectors (i.e. produce
maxps and maxpd). This broke the sse41-blend.ll testcase by causing
maxpd to be produced rather than a cmp+blend pair, which is the reason
I tweaked it. Gives a small speedup on doduc with dragonegg when the
GCC vectorizer is used.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@139986 91177308-0d34-0410-b5e6-96231b3b80d8
diff --git a/lib/Target/X86/X86ISelLowering.cpp b/lib/Target/X86/X86ISelLowering.cpp
index d279e04..42193c4 100644
--- a/lib/Target/X86/X86ISelLowering.cpp
+++ b/lib/Target/X86/X86ISelLowering.cpp
@@ -1129,6 +1129,7 @@
setTargetDAGCombine(ISD::VECTOR_SHUFFLE);
setTargetDAGCombine(ISD::EXTRACT_VECTOR_ELT);
setTargetDAGCombine(ISD::BUILD_VECTOR);
+ setTargetDAGCombine(ISD::VSELECT);
setTargetDAGCombine(ISD::SELECT);
setTargetDAGCombine(ISD::SHL);
setTargetDAGCombine(ISD::SRA);
@@ -12551,7 +12552,8 @@
return SDValue();
}
-/// PerformSELECTCombine - Do target-specific dag combines on SELECT nodes.
+/// PerformSELECTCombine - Do target-specific dag combines on SELECT and VSELECT
+/// nodes.
static SDValue PerformSELECTCombine(SDNode *N, SelectionDAG &DAG,
const X86Subtarget *Subtarget) {
DebugLoc DL = N->getDebugLoc();
@@ -12564,9 +12566,9 @@
// instructions match the semantics of the common C idiom x<y?x:y but not
// x<=y?x:y, because of how they handle negative zero (which can be
// ignored in unsafe-math mode).
- if (Subtarget->hasXMMInt() &&
- (LHS.getValueType() == MVT::f32 || LHS.getValueType() == MVT::f64) &&
- Cond.getOpcode() == ISD::SETCC) {
+ if (Subtarget->hasXMMInt() && Cond.getOpcode() == ISD::SETCC &&
+ (LHS.getValueType() == MVT::f32 || LHS.getValueType() == MVT::f64 ||
+ LHS.getValueType() == MVT::v4f32 || LHS.getValueType() == MVT::v2f64)) {
ISD::CondCode CC = cast<CondCodeSDNode>(Cond.getOperand(2))->get();
unsigned Opcode = 0;
@@ -13871,6 +13873,7 @@
default: break;
case ISD::EXTRACT_VECTOR_ELT:
return PerformEXTRACT_VECTOR_ELTCombine(N, DAG, *this);
+ case ISD::VSELECT:
case ISD::SELECT: return PerformSELECTCombine(N, DAG, Subtarget);
case X86ISD::CMOV: return PerformCMOVCombine(N, DAG, DCI);
case ISD::ADD: return OptimizeConditionalInDecrement(N, DAG);