[llvm] [NVPTX][NFC] Move more TMA intrinsics lowering to tablegen (PR #147576)

[Durgadoss R via llvm-commits]

================
@@ -560,6 +560,27 @@ defm CP_ASYNC_BULK_PREFETCH_CH : CP_ASYNC_BULK_PREFETCH_INTR<has_ch = 1>;
 // TMA Async Bulk Tensor Copy Functions
 //-------------------------------------
 
+class TMA_DIMS_UTIL<int dim> {
+  dag ins_dag = !dag(ins, !listsplat(B32, dim), !foreach(i, !range(dim), "d" # i));
+  string base_str = !interleave(!foreach(i, !range(dim), "$d" # i), ", ");
+}
----------------
durga4github wrote:

Sure. Added an example of a dim=3 case here.

Resolving this,

https://github.com/llvm/llvm-project/pull/147576