How can I remove the limitation that the CPU only calls half of the cores?
How can I remove the limitation that the CPU only calls half of the cores?
You can change it by setting an environment variable GGUF_THREADS=10
https://github.com/rupeshs/fastsdcpu/blob/ee0c2aba561f4fad6a747ef984477547b4a39741/src/constants.py#L24
You can change it by setting an environment variable
GGUF_THREADS=10https://github.com/rupeshs/fastsdcpu/blob/ee0c2aba561f4fad6a747ef984477547b4a39741/src/constants.py#L24
Thank you, I have found this in the previous issues. I would like to ask how to modify openvino and lcm.
You can change it by setting an environment variable
GGUF_THREADS=10https://github.com/rupeshs/fastsdcpu/blob/ee0c2aba561f4fad6a747ef984477547b4a39741/src/constants.py#L24
Thank you, My server's CPU has 96 cores, and by default, it should be using 48 cores. However, in practice, only 24 cores are being utilized. Setting the "GGUF_THREADS" environment variable also doesn't seem to have any effect.
You can change it by setting an environment variable
GGUF_THREADS=10https://github.com/rupeshs/fastsdcpu/blob/ee0c2aba561f4fad6a747ef984477547b4a39741/src/constants.py#L24
Thank you, My server's CPU has 96 cores, and by default, it should be using 48 cores. However, in practice, only 24 cores are being utilized. Setting the "GGUF_THREADS" environment variable also doesn't seem to have any effect.
which mode are you using? is it LCM-OpenVINO?
You can change it by setting an environment variable
GGUF_THREADS=10https://github.com/rupeshs/fastsdcpu/blob/ee0c2aba561f4fad6a747ef984477547b4a39741/src/constants.py#L24
Thank you, My server's CPU has 96 cores, and by default, it should be using 48 cores. However, in practice, only 24 cores are being utilized. Setting the "GGUF_THREADS" environment variable also doesn't seem to have any effect.
which mode are you using? is it LCM-OpenVINO?
Yes LCM-OpenVINO, the model uses rupeshs/FLUX.1-schnell-openvino-int4
Thank you, My server's CPU has 96 cores, and by default, it should be using 48 cores. However, in practice, only 24 cores are being utilized. Setting the "GGUF_THREADS" environment variable also doesn't seem to have any effect.