2 Commits
Author SHA1 Message Date
Rail ChabdarovandGitHub 340807273b hip: Avoid compiler bug in RDNA code generation during debug builds on Windows (#20655) 2026-03-19 19:14:08 +01:00
Rail ChabdarovandGitHub 5a32a9b8a5 Fix data race in CUDA's "cpy" kernel (influences GGML's DUP, CONT operations). (#20507)
* Fix datarace in CUDA's "cpy" kernel.

* Remove extra barrier by using more of shared memory.
2026-03-14 13:19:44 +08:00