Post date: May 6, 2020 7:19:09 PM
We have included 10 additional Volta 100 GPU nodes in our Rider cluster which use the same gpu partition (-p gpu). Four of them have 4 Volta GPUs in each server (-C gpu4v100) and the remaining six of them have 2 Volta GPUs in each server ( -C gpu2v100). For detail information about these GPU servers, please visit HPC Resource View (https://sites.google.com/a/case.edu/hpcc/servers-and-storage/cluster-resources) and for GPU job submission, refer to HPC Guide to CUDA/GPU - https://sites.google.com/a/case.edu/hpcc/home/Software-Guide/cuda.
We would like to request you to avoid making use of more than 1-2 nodes at a time. Also, make use of GPU nodes with 4 GPUs only if your job scales well with multiple GPUs.
Please visit our HPC Site where you can do keyword search for most of the topics of your interest.
Earlier, we had added 24 Compute nodes in Rider and 28 Compute nodes in Markov each with 40 CPUs. Check the details at https://sites.google.com/a/case.edu/hpcc/servers-and-storage/cluster-resources.