RiverDave wrote: > Do you plan to add the LLVM lowering next? Might not be too hard to consider > as part of this PR already?
`CUDAKernelNameAttr` is meant to be a CIR-only bookkeeping attribute used during LoweringPrepare to associate kernel symbol names with registration entries (FuncOp on CUDA, GlobalOp on HIP). It is not explicitly lowered to LLVM IR today. https://github.com/llvm/llvm-project/pull/180051 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
