vLLM v0.33.0rc2 fixes AITER GEMM issue on ROCm gfx950
AI summary
vLLM v0.33.0rc2 includes a ROCm bugfix that skips AITER tuned GEMM when M*N is at least 2^31 on gfx950. The release note says the change was cherry-picked from commit 6d01893.
In-depth insight
This targeted ROCm fix avoids using AITER tuned GEMM for very large matrix dimensions on gfx950, addressing a platform-specific execution issue in vLLM.
Signed-off-by: Micah Williamson [email protected] (cherry picked from commit 6d01893)