mirror of
https://github.com/dogkeeper886/ollama37.git
synced 2025-12-11 00:07:07 +00:00
Release gpu discovery library after use
Leaving the cudart library loaded kept ~30m of memory pinned in the GPU in the main process. This change ensures we don't hold GPU resources when idle.
This commit is contained in:
@@ -191,4 +191,10 @@ void cudart_compute_capability(cudart_handle_t h, cudart_compute_capability_t *r
|
||||
}
|
||||
}
|
||||
|
||||
void cudart_release(cudart_handle_t h) {
|
||||
LOG(h.verbose, "releasing cudart library\n");
|
||||
UNLOAD_LIBRARY(h.handle);
|
||||
h.handle = NULL;
|
||||
}
|
||||
|
||||
#endif // __APPLE__
|
||||
Reference in New Issue
Block a user