[flang-commits] [flang] [flang][cuda] Use NVVM op for barrier0 intrinsic (PR #140947)

[Valentin Clement バレンタイン クレメン via flang-commits]

clementval wrote:

> I should mention that in non strictly CUF flows (OpenAcc), I am considering using gpu.barrier as it can target different back-ends.

`syncthreads` is a cudadevice procedure so I don't think it needs to be portable. 

https://github.com/llvm/llvm-project/pull/140947