6502/llvm-6502
1
0
Fork 0
mirror of https://github.com/c64scene-ar/llvm-6502.git synced 2024-07-19 03:29:22 +00:00
Code Issues Projects Releases Wiki Activity

Fix a performance regression. Use {p}shuf* when there are only two distinct elements in a build_vector.

Browse Source 
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27945 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Evan Cheng 2006-04-22 06:21:46 +00:00
parent ee183471a6
commit a083af14c8
1 changed files with 17 additions and 13 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

30
lib/Target/X86/X86ISelLowering.cpp
View File

@ -3274,22 +3274,26 @@ SDOperand X86TargetLowering::LowerOperation(SDOperand Op, SelectionDAG &DAG) {
return DAG.getNode(ISD::VECTOR_SHUFFLE, VT, V[0], V[1], ShufMask);
}
// Expand into a number of unpckl*.
// e.g. for v4f32
// Step 1: unpcklps 0, 2 ==> X: <?, ?, 2, 0>
// : unpcklps 1, 3 ==> Y: <?, ?, 3, 1>
// Step 2: unpcklps X, Y ==> <3, 2, 1, 0>
SDOperand UnpckMask = getUnpacklMask(NumElems, DAG);
for (unsigned i = 0; i < NumElems; ++i)
V[i] = DAG.getNode(ISD::SCALAR_TO_VECTOR, VT, Op.getOperand(i));
NumElems >>= 1;
while (NumElems != 0) {
if (Values.size() > 2) {
// Expand into a number of unpckl*.
// e.g. for v4f32
// Step 1: unpcklps 0, 2 ==> X: <?, ?, 2, 0>
// : unpcklps 1, 3 ==> Y: <?, ?, 3, 1>
// Step 2: unpcklps X, Y ==> <3, 2, 1, 0>
SDOperand UnpckMask = getUnpacklMask(NumElems, DAG);
for (unsigned i = 0; i < NumElems; ++i)
V[i] = DAG.getNode(ISD::VECTOR_SHUFFLE, VT, V[i], V[i + NumElems],
UnpckMask);
V[i] = DAG.getNode(ISD::SCALAR_TO_VECTOR, VT, Op.getOperand(i));
NumElems >>= 1;
while (NumElems != 0) {
for (unsigned i = 0; i < NumElems; ++i)
V[i] = DAG.getNode(ISD::VECTOR_SHUFFLE, VT, V[i], V[i + NumElems],
UnpckMask);
NumElems >>= 1;
}
return V[0];
}
return V[0];
return SDOperand();
}
case ISD::EXTRACT_VECTOR_ELT: {
if (!isa<ConstantSDNode>(Op.getOperand(1)))