The Daily Bulletin delivers news about CISL’s high-performance computing environment, training opportunities, scheduled maintenance, and other events of interest to the user community. To get the Daily Bulletin by email, use this link to subscribe.
Later this year, the National Center for Atmospheric Research (NCAR) will deploy Derecho, its next-generation HPC system, and NCAR will begin accepting proposals in July from university-led teams to participate in the Accelerated Scientific Discovery (ASD) program on the system. The ASD program will provide a unique opportunity for a small group of large-scale computational projects to have nea...
CISL recommends running small jobs that use only CPUs on the Casper cluster’s high-throughput computing (HTC) nodes. Casper has 64 HTC nodes specifically for running small batch jobs. It also has: • More shared resources than the Cheyenne share queue • A far higher concurrent-use limit for CPU cores than Cheyenne: 468 vs. 36. • More available memory, plus NVMe swap for overflow. Here’s an exa...
No downtime scheduled for Cheyenne, Casper, Campaign Storage, JupyterHub, HPSS, GLADE, Quasar, or Stratus.
Users creating conda environments may encounter an I/O bug on Cheyenne and Casper when using Python versions 3.8 or later, so CISL recommends specifying Python 3.7 unless a new version is needed. Users encountering the bug receive an error message saying “BlockingIOError: [Errno 11] Resource temporarily unavailable.” Specifying Python 3.7 or an earlier version will avoid the problem. When creat...
Researchers who need help visualizing data to demonstrate the results of their scientific computing can request expert assistance and collaboration from CISL. This service is available to researchers who want help using specialized applications on data analysis and visualization resources that CISL manages. CISL visualization staff have particular expertise in CISL-developed software such as VA...
During the previously announced scheduled downtime of CISL’s HPC systems in July, the PBS Pro job scheduler on both Cheyenne and Casper will be upgraded to version 21 from the current version 18 (Cheyenne) and 20 (Casper). Updating both to version 21 will address some compatibility issues between the two clusters and is expected to improve performance and stability. When both systems are runnin...