mirror of
https://github.com/ggml-org/llama.cpp.git
synced 2025-07-31 22:53:52 -04:00
74d4f5b041
There seems to be a bubble waking up from waitForFences, which costs a few percent performance and also increased variance in performance. This change inserts an "almost_ready" fence when the graph is about 80% complete and we waitForFences for the almost_ready fence and then spin (with _mm_pauses) waiting for the final fence to be signaled.