[flang-commits] [lld] [lldb] [libcxx] [compiler-rt] [clang-tools-extra] [llvm] [libc] [clang] [flang] [NVPTX] Add support for -march=native in standalone NVPTX (PR #79373)

[Joseph Huber via flang-commits]

jhuber6 wrote:

> I...think I understand.
> 
> Is the output of this compilation step a cubin, then?

Yes, it will spit out a simple `cubin` instead of a fatbinary. The NVIDIA toolchain is much worse about this stuff than the AMD one, but in general it works. You can check with `-###` or whatever like in https://godbolt.org/z/zWf5jezYP.

https://github.com/llvm/llvm-project/pull/79373