[Mlir-commits] [mlir] [mlir][gpu] Add `gpu.subgroup_uniform` op (PR #157743)

[Ivan Butygin]

================
@@ -3255,4 +3255,37 @@ def GPU_SubgroupBroadcastOp : GPU_Op<"subgroup_broadcast",
   let hasVerifier = 1;
 }
 
+def GPU_SubgroupUniformOp : GPU_Op<"subgroup_uniform",
+    [Pure, AllTypesMatch<["result", "src"]>,
----------------
Hardcode84 wrote:

See my other comment on uniformity, but yeah, I think we can either make it speculatable if we require all lanes to be active at definition or make it non-speculatable but without requiring all lanes

https://github.com/llvm/llvm-project/pull/157743