mirror of
https://github.com/ggml-org/llama.cpp.git
synced 2025-07-26 11:13:53 -04:00
There have been reports of failure to compile on systems with <= 32KB of shared memory (e.g. #10037). This change makes the large tile size fall back to a smaller size if necessary, and makes mul_mat_id fall back to CPU if there's only 16KB of shared memory.