[llvm] [NVPTX] Add sparse MMA intrinsics (PR #150950)

[Durgadoss R via llvm-commits]

================
@@ -2001,6 +2183,48 @@ foreach layout_a = ["row", "col"] in {
   } // layout_b
 } // layout_a
 
+// MMA.SP
+class NVVM_MMA_SP<WMMA_REGS A, WMMA_REGS B, WMMA_REGS C, WMMA_REGS D>
+  : Intrinsic<D.regs,
+              !listconcat(A.regs, B.regs, C.regs, [llvm_i32_ty], [llvm_i32_ty])> {
+    int pos = !size(!listconcat(A.regs, B.regs, C.regs, [llvm_i32_ty]));
+
+    // The range [0;range) is for the sparsity selector that indicates the threads
+    // which contribute metadata.
+    int range = !if(!or(!and(!eq(A.geom, "m16n8k32"), !eq(A.ptx_elt_type, "bf16")),
----------------
durga4github wrote:

nit: can we name it num_threads or something like that?

https://github.com/llvm/llvm-project/pull/150950