[Mlir-commits] [mlir] [MLIR][NVVM] Add TMA prefetch Op (PR #141211)

Durgadoss R llvmlistbot at llvm.org
Fri May 23 03:03:33 PDT 2025


================
@@ -2344,6 +2344,49 @@ def NVVM_PrefetchTensorMapOp : NVVM_Op<"prefetch.tensormap",
   }];
 }
 
+def NVVM_CpAsyncBulkPrefetchOp : NVVM_Op<"cp.async.bulk.prefetch"> {
----------------
durga4github wrote:

This is the non-tensor variant.

The tensor variants usually support various modes and `tensor` forms like 1d,2d..5d etc.
The non-tensor is a simple linear copy/prefetch of bulk data.

(updated the title to state non-tensor explicitly)

https://github.com/llvm/llvm-project/pull/141211


More information about the Mlir-commits mailing list