quic-zhanweiw
quic-zhanweiw
@ReinForce-II It seems the performance issue is due to the function 'ggml_compute_forward()' was called twice in function 'ggml_graph_compute_thread()'. It was just called once in previous code. May we avoid this?...
Thanks @ReinForce-II for your update! In the latest Llama.cpp code, after add the patch below, the performance issue disappeared. May you help check the reason? ``` diff --git a/ggml.c b/ggml.c...
Are there any solution for this issue?
I met the same issue with langflow 1.2.0, but langflow 1.1.1 doesn't have this issue.
即使设置成20个线程,也是一样的结果。
I can compile openssl v3.4.0 to ARM64EC successfully with the configuration 'no-asm'. But I think it will impact the openssl performance obviously. Hope can support ARM64EC officially with assembly code...