Closed olupton closed 1 month ago
LGTM. Did you file a bug to upstream?
Did you file a bug to upstream?
No, I don't think there's anyone that would want to fix this for us, so I'm not sure who I could file a bug with.
The problem is that (openxla/triton, bazel) typically works, and is used in XLA, and (openai/triton, cmake+wheel) typically works, but here we are trying to build (openxla/triton, cmake+wheel), and this sometimes breaks.
AMD headers are included unconditionally; the simplest approach is to make sure they exist.