NIH HPC News & Announcements
V100 GPUs now available on NIH Biowulf Cluster
Date: 10 April 2018 09:04:40
From: steven fellini
The NIH HPC/Biowulf staff is pleased to announce the availability of 8 compute nodes which are each equipted with four NVIDIA V100 GPUs. The V100 was designed specifically to boost performance for deep learning frameworks such as Caffe2, TensorFlow and Pytorch. See https://www.nvidia.com/en-us/data-center/tesla-v100 for details on the V100 GPU.
Users are asked to restrict use of the V100 nodes to deep learning applications.
To allocate V100 nodes, specify the following partition and resource switches to the 'sbatch' command:
--partition=gpu --gres=gpu:v100:N
where N = 1, 2, 3 or 4.
For sinteractive sessions, use only
--gres=gpu:v100:N
########################################################################
Please contact staff@hpc.nih.gov with any questions about the NIH HPC Systems
[Last 12 months of HPC announcements]