[Mlir-commits] [mlir] [MLIR][GPU] Support grid constant, byval, byref on gpu.func (PR #172037)

[Guray Ozen]

================
@@ -369,7 +370,10 @@ GPUFuncOpLowering::matchAndRewrite(gpu::GPUFuncOp gpuFuncOp, OpAdaptor adaptor,
     }
 
     if (lowersToPointer) {
+      copyPointerAttribute(mlir::NVVM::NVVMDialect::getGridConstantAttrName());
----------------
grypp wrote:

Sorry, I’m a bit late to the party here.

I don’t see a clean path to landing this PR. We can’t connect NVVM to GPU (or GPU to NVVM) directly, and that’s exactly why we introduced the `NVGPU` dialect as the modeling layer. Can we implement this in there?



https://github.com/llvm/llvm-project/pull/172037