[llvm] [LV, VP]VP intrinsics support for the Loop Vectorizer (PR #76172)
Florian Hahn via llvm-commits
llvm-commits at lists.llvm.org
Fri Jan 26 08:36:33 PST 2024
================
@@ -9445,14 +9552,31 @@ void VPWidenMemoryInstructionRecipe::execute(VPTransformState &State) {
}
}
+ auto MaskValue = [&](unsigned Part) -> Value * {
+ if (isMaskRequired)
+ return BlockInMaskParts[Part];
+ return nullptr;
+ };
+
// Handle Stores:
if (SI) {
State.setDebugLocFrom(SI->getDebugLoc());
for (unsigned Part = 0; Part < State.UF; ++Part) {
Instruction *NewSI = nullptr;
Value *StoredVal = State.get(StoredValue, Part);
- if (CreateGatherScatter) {
+ if (State.EVL) {
----------------
fhahn wrote:
Still accessing EVL (as per https://reviews.llvm.org/D99750?id=558054#inline-1551722).
To move forward and make progress, I think this would suggest to leave as is now, with a TODO to model this explicitly in the recipe.
I am planning on splitting up the memory recipes soon now that address generation has been moved.
https://github.com/llvm/llvm-project/pull/76172
More information about the llvm-commits
mailing list