-
Notifications
You must be signed in to change notification settings - Fork 395
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
weird ram usage. #1313
Comments
That's due to mmap, in future versions it will not be the default anymore. |
i have this behaviour with both mmap enabled or disabled. it is something low level thing? |
No, if mmap is disabled, when a layer is offloaded to GPU in cuda, it will be freed from RAM. |
im using clbalst since i have no avx2 also i noticed that some models behavior is different when using cuda or clblast |
Try the vulkan noavx2 version instead. Or use the cuda oldcpu exe. |
good day im trying to understand how the math is works for kobold cpp.
i have a modell arraund 10gb in size with q3 quentiuation. i have my ram usage before trying to load as 6gb. i trying to understand the follwing. kobold says it will load lets say 8 gb to gpu and 2 gb to cpu but in the end it loads 8gb to gpu and take an other 10gb on cpu ram. what makes it kinda weird for me why? shuld it be just eating 2gb as it promised it will?
The text was updated successfully, but these errors were encountered: