[llvm] [NVVM][NVPTX] Add tensor map override support in S2G and Reduce intrinsics (PR #215503)

[Rajat Bajpai via llvm-commits]

================
@@ -858,13 +858,187 @@ foreach dim = 3...5 in {
 defm TMA_S2G_TILE_SCATTER4_2D : TMA_TENSOR_S2G_INTR<5, "tile_scatter4",
                                 [callSubtarget<"hasTMABlackwellSupport">]>;
 
+// Base class for TMA tensor override intrinsics
+class TMA_TENSOR_OVERRIDE_BASE<int dim, string mode, string override> {
+  // Dimension handling
+  dag dims_dag = !dag(ins, !listsplat(B32, dim), !foreach(i, !range(dim), "d" # i));
+  string dims_str = !interleave(!foreach(i, !range(dim), "$d" # i), ", ");
----------------
rajatbajpai wrote:

yes

https://github.com/llvm/llvm-project/pull/215503