10 Apr
2024
10 Apr
'24
4:36 a.m.
External Email - Use Caution
Hi,
I'm using NextBrain segmentation tool on our cluster which has NVIDIA H100 GPUs and I'm getting this error:
NVIDIA H100 with CUDA capability sm_90 is not compatible with the current PyTorch installation. The current PyTorch install supports CUDA capabilities sm_37 sm_50 sm_60 sm_70 sm_75 sm_80 sm_86. If you want to use the NVIDIA H100 GPU with PyTorch, please check the instructions at https://secure-web.cisco.com/1dXRtbKFFWRnAd7DYiR89C9QyPNuu24f97dyifrR_gDGXsQ...
Should I contact our cluster managers in this regard or is there a specific workaround that can overcome this issue?
Best, Amir