Cluster Upgrade (Summer 2026)¶

Important!

SC Cluster Scheduled Maintenance - Fri July 3rd (midnight), early completion 2pm Sat July 4th

/sailhome performance enhancement - increased disk/network IOPS with SSD-backed storage arrays, new network uplinks along with other tweaks
SLURM 25.11 upgrade (from 25.05) - all CS maintained cluster will be aligned with the same version
Up-to-date Security patches and OS maintenance on all infrastructure node and compute node (Ubuntu 24.04)
Nvidia GPU driver update across the cluster (both sc/ilc) - Nodes with GPU architecture Turing (TitanRTX, 2080) and newer are upgraded to Nvidia driver version 610.43.02 along with CUDA 13.3 (12.8 still being the system default). Nodes with older GPU such as Pascal/Volta (Titan Xp/V) will continue to stay on driver version 580.82.07, CUDA 13.0 is the last supported version from Nvidia
/vision storage server was upgraded to conform with minimum security requirement, other older storage servers will also be done at a later time and will be scheduled with respective group separately

Thank You Everyone!

Much appreciated everyone's patience during the outage, we get next to no inquiries during the maintenance window, which contributed for a smooth operation. As mentioned previously, we typically only do this once a year to minimize impact for various deadlines throughout the academic year. As always please contact us at action@cs.stanford.edu if there's anything we can help!