High Performance Computing Services systems analyst Jackie Scoggins will be giving a talk about how we recently migrated our scheduling environment, which manages the job scheduling for all of our clusters, cluster condo users, and institutional computing at the Lab and for the UC Berkeley Research Computing program, to the SLURM job scheduler at the SLURM User Group Meeting being held this week. SLURM, an open-source job scheduler originally developed at LLNL and used on the largest systems in the world, is now available with commercial support from SchedMD and is fast becoming the preferred scheduler to use for academic research computing. Jackie will also be giving a tutorial on using Berkeley Lab NHC (Node Health Check) with SLURM.
Outage Details
Date
Impact
Researchers at the ALS used “soft” X-rays to image structures only five nanometers in size. This resolution is the highest ever achieved with X-ray microscopy and may have ushered in a new era for nanoscale imaging. Diffraction data is recorded by a high speed CCD camera and fed through a Data Pipeline consisting of a 43K GPU core infiniband LInux cluster, a high speed 10 gigabit network and a Data Transfer Node running Globus Online implemented by the High Performance Computing Services Group. Read more.