[clang] [llvm] [CIR][AMDGPU] Lower cir.offload.printf to AMDGPU printf (PR #231563)

Steffen Larsen via cfe-commits cfe-commits at lists.llvm.org
Wed Oct 28 23:48:47 PDT 2026


================
@@ -7,16 +7,19 @@
 //===----------------------------------------------------------------------===//
 //
 // This file lowers cir.offload.printf to the printf runtime interface of each
-// device target. The expansions follow the lowerings of the GPU dialect's
-// gpu.printf in mlir/lib/Conversion/GPUCommon/GPUOpsLowering.cpp.
+// device target.
----------------
steffenlarsen wrote:

I don't think that'd necessarily be a bad idea, but I would prefer to do it as an NFC later, mainly to avoid the PR growing in size and people mistakingly thinking the PTX side is changing.

https://github.com/llvm/llvm-project/pull/231563


More information about the cfe-commits mailing list