[llvm] [NVPTX] Vectorize and lower 256-bit global loads/stores for sm_100+/ptx88+ (PR #139292)

[Drew Kersnar via llvm-commits]

================
@@ -199,6 +207,15 @@ getVectorLoweringShape(EVT VectorEVT) {
   switch (VectorVT.SimpleTy) {
   default:
     return std::nullopt;
+  case MVT::v4i64:
----------------
dakersnar wrote:

Yeah, I think you're right, but I would prefer if we did this in a separate future change.

https://github.com/llvm/llvm-project/pull/139292