UIT HPC Services is in the final stages of preparing to upgrade the Talon 2.0 HPC system to Talon 3.0. This upgrade will result in additional computational capacity while minimizing, as much as possible, complete interruption of HPC service. However, during the next two months of November and December, expect operations to continue with degraded compute capacity. This will likely increase queue wait times for submitted jobs.
On November 14, 2016, we plan to remove 96 of the c32 nodes (racks 1-3) and 16 GPU nodes (rack 9) from the Talon cluster to prepare for the installation of new computing hardware. Jobs currently executing on those nodes will be allowed to run to completion if possible up until said date. On that date, jobs still in execution on those nodes will have to be requeued.
By mid January, 2017 we anticipate the new hardware will be installed and returning to full operation again. The upgraded Talon 3.0 cluster will have 8,128 CPU cores, 39,936 GPU coprocessor cores, and a combined 24TB of system memory. The same high-performance Lustre scratch file system will continue to be available under the new configuration as well as additional object storage we will soon be revealing.
We apologize in advance for any inconvenience that may arise during this transition period. If you have any concerns or questions before, during, or after this upgrade period please address them to firstname.lastname@example.org. If you already have plans that require significant compute resources, please contact us to inquire about utilizing XSEDE resources.