[llvm] [AMDGPU] Enable vectorization of i8 values. (PR #134934)

[Alexey Bataev via llvm-commits]

================
@@ -20927,7 +20934,11 @@ bool SLPVectorizerPass::tryToVectorizeList(ArrayRef<Value *> VL, BoUpSLP &R,
     // provided vectorization factor (i.e. the scalar type is used for vector
     // code during codegen).
     auto *VecTy = getWidenedType(ScalarTy, VF);
-    if (TTI->getNumberOfParts(VecTy) == VF)
+    unsigned NumParts = TTI->getNumberOfParts(VecTy);
+    if (TTI->canVectorizei8s() &&
+        VecTy->getElementType() == IntegerType::getInt8Ty(VecTy->getContext()))
+      NumParts = 1;
+    if (NumParts == VF)
----------------
alexey-bataev wrote:

Same here, it should not be in SLP vectorizer, it should be the part of getNumberOfParts logic in GCNTTIImpl

https://github.com/llvm/llvm-project/pull/134934