jhuber6 wrote: > > It should only be invoking `nvptx-arch` if the user passed `-march=native`. > > Sorry, didn't notice this sentence. Well, _I am_ building with > `-march=native` here — after all, other files are built for a CPU. If I > change it to, say, `-march=zen2`, then it indeed compiles, with a lot of > warnings: > > ``` > 'zen2' is not a recognized processor for this target (ignoring processor) > 'zen2' is not a recognized processor for this target (ignoring processor) > 'zen2' is not a recognized processor for this target (ignoring processor) > 'zen2' is not a recognized processor for this target (ignoring processor) > 'zen2' is not a recognized processor for this target (ignoring processor) > 'zen2' is not a recognized processor for this target (ignoring processor) > ``` > > To be honest, this only proves my point — you shouldn't be invoking > `CMAKE_CXX_COMPILER` when you're building for a GPU, since it's a compiler > configured to be building for the host CPU.
Hence why the follow-up to this will move it to a separate project where you pass https://cmake.org/cmake/help/latest/variable/CMAKE_LANG_COMPILER_TARGET.html. Right now I guess it's in an awful worst-of-both-worlds state, but I was hoping to get the functional change to the generated code done so I could just do the infrastructure change. https://github.com/llvm/llvm-project/pull/126143 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
