mirror of
https://github.com/dogkeeper886/ollama37.git
synced 2025-12-15 18:27:08 +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:
@@ -51,6 +51,7 @@ typedef struct nvml_compute_capability {
|
||||
void nvml_init(char *nvml_lib_path, nvml_init_resp_t *resp);
|
||||
void nvml_check_vram(nvml_handle_t ch, mem_info_t *resp);
|
||||
void nvml_compute_capability(nvml_handle_t ch, nvml_compute_capability_t *cc);
|
||||
void nvml_release(nvml_handle_t ch);
|
||||
|
||||
#endif // __GPU_INFO_NVML_H__
|
||||
#endif // __APPLE__
|
||||
Reference in New Issue
Block a user