================
@@ -2865,12 +2865,11 @@ void CodeGenModule::ConstructAttributeList(StringRef
Name,
NumElemsParam);
}
- // OpenCL v2.0 Work groups may be whether uniform or not.
- // '-cl-uniform-work-group-size' compile option gets a hint
- // to the compiler that the global work-size be a multiple of
- // the work-group size specified to clEnqueueNDRangeKernel
- // (i.e. work groups are uniform).
- if (getLangOpts().OffloadUniformBlock)
+ // Leftover work-groups are an OpenCL 2.0+ language feature. Host C/C++
+ // inherit OffloadUniformBlock but must not receive the GPU ABI attribute.
----------------
jhuber6 wrote:
I genuinely can't think of a clean way to do this without just using the
triple. What I had before was "CUDA / HIP / SYCL / OpenMP Device / C / C++ with
a GPU target" and I think that's perfectly reasonable.
https://github.com/llvm/llvm-project/pull/221279
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
