Begin to support some vector operations for AVX 256-bit intructions. The long
term goal here is to be able to match enough of vector_shuffle and build_vector
so all avx intrinsics which aren't mapped to their own built-ins but to
shufflevector calls can be codegen'd. This is the first (baby) step, support
building zeroed vectors.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@110897 91177308-0d34-0410-b5e6-96231b3b80d8
diff --git a/lib/Target/X86/X86ISelLowering.cpp b/lib/Target/X86/X86ISelLowering.cpp
index 41f630f..15bbf07 100644
--- a/lib/Target/X86/X86ISelLowering.cpp
+++ b/lib/Target/X86/X86ISelLowering.cpp
@@ -883,7 +883,7 @@
setOperationAction(ISD::FDIV, MVT::v8f32, Legal);
setOperationAction(ISD::FSQRT, MVT::v8f32, Legal);
setOperationAction(ISD::FNEG, MVT::v8f32, Custom);
- //setOperationAction(ISD::BUILD_VECTOR, MVT::v8f32, Custom);
+ setOperationAction(ISD::BUILD_VECTOR, MVT::v8f32, Custom);
//setOperationAction(ISD::VECTOR_SHUFFLE, MVT::v8f32, Custom);
//setOperationAction(ISD::EXTRACT_VECTOR_ELT, MVT::v8f32, Custom);
//setOperationAction(ISD::SELECT, MVT::v8f32, Custom);
@@ -3412,18 +3412,27 @@
DebugLoc dl) {
assert(VT.isVector() && "Expected a vector type");
- // Always build zero vectors as <4 x i32> or <2 x i32> bitcasted to their dest
- // type. This ensures they get CSE'd.
+ // Always build zero vectors as <4 x i32> or <2 x i32> bitcasted
+ // to their dest type. This ensures they get CSE'd.
SDValue Vec;
if (VT.getSizeInBits() == 64) { // MMX
SDValue Cst = DAG.getTargetConstant(0, MVT::i32);
Vec = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v2i32, Cst, Cst);
- } else if (HasSSE2) { // SSE2
- SDValue Cst = DAG.getTargetConstant(0, MVT::i32);
- Vec = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v4i32, Cst, Cst, Cst, Cst);
- } else { // SSE1
+ } else if (VT.getSizeInBits() == 128) {
+ if (HasSSE2) { // SSE2
+ SDValue Cst = DAG.getTargetConstant(0, MVT::i32);
+ Vec = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v4i32, Cst, Cst, Cst, Cst);
+ } else { // SSE1
+ SDValue Cst = DAG.getTargetConstantFP(+0.0, MVT::f32);
+ Vec = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v4f32, Cst, Cst, Cst, Cst);
+ }
+ } else if (VT.getSizeInBits() == 256) { // AVX
+ // 256-bit logic and arithmetic instructions in AVX are
+ // all floating-point, no support for integer ops. Default
+ // to emitting fp zeroed vectors then.
SDValue Cst = DAG.getTargetConstantFP(+0.0, MVT::f32);
- Vec = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v4f32, Cst, Cst, Cst, Cst);
+ SDValue Ops[] = { Cst, Cst, Cst, Cst, Cst, Cst, Cst, Cst };
+ Vec = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v8f32, Ops, 8);
}
return DAG.getNode(ISD::BIT_CONVERT, dl, VT, Vec);
}
@@ -3437,9 +3446,9 @@
// type. This ensures they get CSE'd.
SDValue Cst = DAG.getTargetConstant(~0U, MVT::i32);
SDValue Vec;
- if (VT.getSizeInBits() == 64) // MMX
+ if (VT.getSizeInBits() == 64) // MMX
Vec = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v2i32, Cst, Cst);
- else // SSE
+ else // SSE
Vec = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v4i32, Cst, Cst, Cst, Cst);
return DAG.getNode(ISD::BIT_CONVERT, dl, VT, Vec);
}
@@ -3844,9 +3853,13 @@
SDValue
X86TargetLowering::LowerBUILD_VECTOR(SDValue Op, SelectionDAG &DAG) const {
DebugLoc dl = Op.getDebugLoc();
- // All zero's are handled with pxor, all one's are handled with pcmpeqd.
- if (ISD::isBuildVectorAllZeros(Op.getNode())
- || ISD::isBuildVectorAllOnes(Op.getNode())) {
+ // All zero's are handled with pxor in SSE2 and above, xorps in SSE1 and
+ // all one's are handled with pcmpeqd. In AVX, zero's are handled with
+ // vpxor in 128-bit and xor{pd,ps} in 256-bit, but no 256 version of pcmpeqd
+ // is present, so AllOnes is ignored.
+ if (ISD::isBuildVectorAllZeros(Op.getNode()) ||
+ (Op.getValueType().getSizeInBits() != 256 &&
+ ISD::isBuildVectorAllOnes(Op.getNode()))) {
// Canonicalize this to either <4 x i32> or <2 x i32> (SSE vs MMX) to
// 1) ensure the zero vectors are CSE'd, and 2) ensure that i64 scalars are
// eliminated on x86-32 hosts.