X86: optimized i64 vector multiply with constant

When we multiply two 64-bit vectors, we extract lower and upper part and use the PMULUDQ instruction.
When one of the operands is a constant, the upper part may be zero, we know this at compile time.
Example: %a = mul <4 x i64> %b, <4 x i64> < i64 5, i64 5, i64 5, i64 5>.
I'm checking the value of the upper part and prevent redundant "multiply", "shift" and "add" operations.

llvm-svn: 239802

1 files changed, 11 insertions, 5 deletions

diff --git a/llvm/lib/Target/X86/X86ISelLowering.cpp b/llvm/lib/Target/X86/X86ISelLowering.cpp
index 6228e4a9672..6b76fbc02ed 100644
--- a/llvm/lib/Target/X86/X86ISelLowering.cpp
+++ b/llvm/lib/Target/X86/X86ISelLowering.cpp
@@ -16530,6 +16530,8 @@ static SDValue LowerMUL(SDValue Op, const X86Subtarget *Subtarget,
   SDValue Ahi = getTargetVShiftByConstNode(X86ISD::VSRLI, dl, VT, A, 32, DAG);
   SDValue Bhi = getTargetVShiftByConstNode(X86ISD::VSRLI, dl, VT, B, 32, DAG);
 
+  SDValue AhiBlo = Ahi;
+  SDValue AloBhi = Bhi;
   // Bit cast to 32-bit vectors for MULUDQ
   EVT MulVT = (VT == MVT::v2i64) ? MVT::v4i32 :
                                   (VT == MVT::v4i64) ? MVT::v8i32 : MVT::v16i32;
@@ -16539,11 +16541,15 @@ static SDValue LowerMUL(SDValue Op, const X86Subtarget *Subtarget,
   Bhi = DAG.getBitcast(MulVT, Bhi);
 
   SDValue AloBlo = DAG.getNode(X86ISD::PMULUDQ, dl, VT, A, B);
-  SDValue AloBhi = DAG.getNode(X86ISD::PMULUDQ, dl, VT, A, Bhi);
-  SDValue AhiBlo = DAG.getNode(X86ISD::PMULUDQ, dl, VT, Ahi, B);
-
-  AloBhi = getTargetVShiftByConstNode(X86ISD::VSHLI, dl, VT, AloBhi, 32, DAG);
-  AhiBlo = getTargetVShiftByConstNode(X86ISD::VSHLI, dl, VT, AhiBlo, 32, DAG);
+  // After shifting right const values the result may be all-zero.
+  if (!ISD::isBuildVectorAllZeros(Ahi.getNode())) {
+    AhiBlo = DAG.getNode(X86ISD::PMULUDQ, dl, VT, Ahi, B);
+    AhiBlo = getTargetVShiftByConstNode(X86ISD::VSHLI, dl, VT, AhiBlo, 32, DAG);
+  }
+  if (!ISD::isBuildVectorAllZeros(Bhi.getNode())) {
+    AloBhi = DAG.getNode(X86ISD::PMULUDQ, dl, VT, A, Bhi);
+    AloBhi = getTargetVShiftByConstNode(X86ISD::VSHLI, dl, VT, AloBhi, 32, DAG);
+  }
 
   SDValue Res = DAG.getNode(ISD::ADD, dl, VT, AloBlo, AloBhi);
   return DAG.getNode(ISD::ADD, dl, VT, Res, AhiBlo);