[Mlir-commits] [mlir] [mlir][amdgpu] Add amdgpu.make_dma_descriptor (PR #169407)

[Jakub Kuderski]

================
@@ -1192,4 +1227,75 @@ def AMDGPU_ScaledMFMAOp :
   }];
   let hasCanonicalizer = 1;
 }
+
+def AMDGPU_MakeDmaBaseOp :
+    AMDGPU_Op<"make_dma_base", [Pure, AttrSizedOperandSegments]>,
+    Arguments<(ins
+                   Arg<AnyMemRef, "buffer to read from">:$src,
+                   Variadic<Index>:$srcIndices,
+                   Arg<AnyMemRef, "buffer to write to">:$dst,
+                   Variadic<Index>:$dstIndices)>,
+    Results<(outs AMDGPU_TDMBaseType: $base)> {
+
+  // TODO:
+  // * Add verifiers such that one of the memrefs is from LDS and the other global.
+  // * Add verifiers to make sure that the number of indices do not exceed the number of dimensions.
+
+  let summary = "Pair of based addresses used when moving tiles between LDS and global memory.";
+  let description = [{
+    This operation creates a pair of addresses that will be used by tensor_load_to_lds
+    and tensor_store_from_lds.
+
+    This operation creates a value corresponding roughly to the descriptor group 0
+    found in TensorLoadToLDSOp and TensorStoreFromLDSOp in the rocdl dialect.
+  }];
----------------
kuhar wrote:

ping on this

https://github.com/llvm/llvm-project/pull/169407