[llvm] [NVPTX] Add TMA Bulk Copy Intrinsics (PR #138679)

[Alex MacLean via llvm-commits]

================
@@ -2720,28 +2720,46 @@ void NVPTXDAGToDAGISel::SelectCpAsyncBulkTensorReduceCommon(SDNode *N,
   ReplaceNode(N, CurDAG->getMachineNode(Opcode, DL, N->getVTList(), Ops));
 }
 
-void NVPTXDAGToDAGISel::SelectCpAsyncBulkS2G(SDNode *N) {
+void NVPTXDAGToDAGISel::SelectCpAsyncBulkS2GCommon(SDNode *N, bool HasMask) {
----------------
AlexMaclean wrote:

Hmm, that's annoying that table-gen doesn't seem to allow you to drop/ignore an operand. It looks like you could to something like `(int_nvvm_cp_async_bulk_shared_cta_to_global i64:$dst, addr:$src, i32:$size, i64:$ch, -1)` for the case where we want the cache hint but I can't seem to figure out if there is a way to just ignore the cache-hint operand when the final operand is 0. @Artem-B do you know?

https://github.com/llvm/llvm-project/pull/138679