matt/ollama37
1
0
Fork 0
mirror of https://github.com/dogkeeper886/ollama37.git synced 2025-12-10 15:57:04 +00:00
Code Issues Packages Projects Releases Wiki Activity

nvidia libs have inconsistent ordering (#7473)

Browse Source 
The runtime and management libraries may not always have
identical ordering, so use the device UUID to correlate instead of ID.
This commit is contained in:
Daniel Hiltgen
2024-11-02 16:35:41 -07:00
committed by GitHub
parent b8d5036e33
commit 29ab9fa7d7
3 changed files with 13 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
discover/gpu_info_nvml.c
View File

@@ -17,7 +17,7 @@ void nvml_init(char *nvml_lib_path, nvml_init_resp_t *resp) {
} l[] = {
{"nvmlInit_v2", (void *)&resp->ch.nvmlInit_v2},
{"nvmlShutdown", (void *)&resp->ch.nvmlShutdown},
{"nvmlDeviceGetHandleByIndex", (void *)&resp->ch.nvmlDeviceGetHandleByIndex},
{"nvmlDeviceGetHandleByUUID", (void *)&resp->ch.nvmlDeviceGetHandleByUUID},
{"nvmlDeviceGetMemoryInfo", (void *)&resp->ch.nvmlDeviceGetMemoryInfo},
{NULL, NULL},
};
@@ -67,20 +67,20 @@ void nvml_init(char *nvml_lib_path, nvml_init_resp_t *resp) {
}
void nvml_get_free(nvml_handle_t h, int device_id, uint64_t *free, uint64_t *total, uint64_t *used) {
void nvml_get_free(nvml_handle_t h, char *uuid, uint64_t *free, uint64_t *total, uint64_t *used) {
nvmlDevice_t device;
nvmlMemory_t memInfo = {0};
nvmlReturn_t ret;
ret = (*h.nvmlDeviceGetHandleByIndex)(device_id, &device);
ret = (*h.nvmlDeviceGetHandleByUUID)((const char *)(uuid), &device);
if (ret != NVML_SUCCESS) {
LOG(1, "unable to get device handle %d: %d", device_id, ret);
LOG(1, "unable to get device handle %s: %d", uuid, ret);
*free = 0;
return;
}
ret = (*h.nvmlDeviceGetMemoryInfo)(device, &memInfo);
if (ret != NVML_SUCCESS) {
LOG(1, "device memory info lookup failure %d: %d", device_id, ret);
LOG(1, "device memory info lookup failure %s: %d", uuid, ret);
*free = 0;
return;
}