[libc-commits] [libc] [libc] Implement efficient 'malloc' on the GPU (PR #140156)

[Jon Chesterfield via libc-commits]

JonChesterfield wrote:

Alright, let's land this. I can't see anything obviously wrong with it and the state of the art on language tooling is we can't really do better than inspection and checking it doesn't fall over on the target.

I would really like to be able to build this to spirv and then execute that on a simd cpu under relacy or similar error detection tooling, or some variant of building this to run single step style on an emulator, but we don't have the tooling in place to do that at this time. There's a credible chance that passes unit tests on volta and amdgpu covers enough of the functionality that it'll be useful today, even if we discover and patch races in it in the future.

https://github.com/llvm/llvm-project/pull/140156