X

Presentations

2019 - November 4: Cluster information from fall 2019 meeting. slides

2019 - October 11: Recent presentation of HPC usage. slides

News

2019 - December 5: We are draining gpuq nodes to update the CUDA driver. We are also draining nodes c075, c076, c077, and c078 to be placed into a shortq partition. This partition will have a 2 day time-limit and each job will be limited to single nodes.

2019 - November 4: Thank you for attending today's meeting. If you wanted to attend but didn't make it, we will be having the same meeting on November 7th in Administration building room 191 at 11:00 AM.

2019 - October 23: We will be having our semester meeting in room 191 of the administration building on Monday, November 4th at 11:00 AM. If unable to attend the Monday time, we will also be available to meet in the same time and place on Thursday, November 7th.

2019 - August 1: We have implemented x2go remote desktop with XFCE for cluster login in addition to the normal ssh login. Download x2go-client from here.

2019 - May 13-18 : We will be modifying SLURM's job scheduling. There will be no downtime.

2019 - May 13 : We will be decreasing the default memory per CPU from 2 gigabytes down to 1 gigabyte.

2019 - April - 24: hpcxfer.memphis.edu is no longer available. Additionally, /oldoldhome is no longer available.

2019 - April - 24 : Access to old storage from first penguin cluster, specifically hpcxfer.memphis.edu:/oldoldhome, will be decommissioned, but access to previous cluster storage, hpcxfer.memphis.edu:/oldhome and hpclogin.memphis.edu:/oldhome, will remain. Additionally, there is an HPC meeting at 11AM on Wednesday the 24th in FIT 203/205 for which discussion of HPC resources and policies is a part.

2019 - January - 30 : Old storage from both previous (2015-2018) and first (2012-2015) penguin cluster is available by ssh on hpcxfer.memphis.edu. Just use your campus user name and password like you do on the current cluster. Archive storage is available for those who are interested.

2019 - January - 3 : Cluster scheduler reconfigured.

2018 - September - 25 : The new cluster should be in place by the end of October 2018.

The current cluster will be shut down around the middle of October (exact date to be determined) and will be removed by the end of October See new cluster details at 2018 HPC upgrade.

Contact hpcadmins@memphis.edu with any issues or concerns.

Also visit memphis.edu/hpc for documentation.

Details

We are draining the gpuq node partitions and updating the CUDA driver. Additionally, this will provide updates for the cuda toolkit, blas, and fft modules (version 10.0 and 10.1). Once a node is drained, the driver is updated, and the node rebooted, it will be un-drained. This shouldn't affect any jobs, but jobs may show 'Resources' while in 'PENDING' status.

We are also draining nodes c075, c076, c077, and c078 to place in a new partition called shortq. You can use this partition by simply changing the sbatch partition option to '--partition shortq'. This partition will be limited to 2 day jobs and single nodes (aka shared memory). MPI will work on these nodes, but the single node rule will limit your execution to one node per job. We will not extend time-limits on these nodes.

Thank you for your attention.

The University of Memphis uses cookies in order to enhance your website experience. Visit our Website’s Cookie Policy for more information on how the UofM uses cookies. I understand that by clicking “I agree” and/or continuing to use this website, I agree to the UofM’s use of cookies. More information >