ClearML/clearml
1
0
Fork 0
mirror of https://github.com/clearml/clearml synced 2025-04-21 06:44:16 +00:00
Code Issues Actions 1 Packages Projects Releases Wiki Activity

Fix GPU info such as gpu_memory and gpu_type is not being collected in some cases

Browse Source
This commit is contained in:
allegroai 2024-03-01 22:39:52 +02:00
parent 1320927fdf
commit c600b05386
1 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
clearml/utilities/resource_monitor.py
View File

@ -52,7 +52,7 @@ class ResourceMonitor(BackgroundMonitor):
try: try:
active_gpus = os.environ.get('NVIDIA_VISIBLE_DEVICES', '') or \ active_gpus = os.environ.get('NVIDIA_VISIBLE_DEVICES', '') or \
os.environ.get('CUDA_VISIBLE_DEVICES', '') os.environ.get('CUDA_VISIBLE_DEVICES', '')
if active_gpus != "all": if active_gpus and active_gpus != "all":
self._active_gpus = [g.strip() for g in active_gpus.split(',')] self._active_gpus = [g.strip() for g in active_gpus.split(',')]
except Exception: except Exception:
pass pass
@ -389,7 +389,7 @@ class ResourceMonitor(BackgroundMonitor):
if self._gpustat: if self._gpustat:
gpu_stat = self._gpustat.new_query(shutdown=True, get_driver_info=True) gpu_stat = self._gpustat.new_query(shutdown=True, get_driver_info=True)
if gpu_stat.gpus: if gpu_stat.gpus:
gpus = [g for i, g in enumerate(gpu_stat.gpus) if not self._active_gpus or i in self._active_gpus] gpus = [g for i, g in enumerate(gpu_stat.gpus) if not self._skip_nonactive_gpu(i, g)]
specs.update( specs.update(
gpu_count=int(len(gpus)), gpu_count=int(len(gpus)),
gpu_type=', '.join(g.name for g in gpus), gpu_type=', '.join(g.name for g in gpus),