Weekly Update: July 17, 2019
IN THIS MESSAGE
- Center Announcements
- Atlas Becomes Read-Only on Aug 1; All data removed Aug 15
- Alpine Purge Policy Implementation (Jul 30)
- Rhea Now Mounting Alpine and Atlas
- Meetings & Workshops
- Profiling Tools Workshop (Aug 7-9)
- 2019 Petascale Computing Institute (Aug 19-23)
- OLCF GPU Hackathon Series (multiple dates)
- Want to Be a GPU Hackathon Mentor?
- Upcoming Scheduled Outages
- None through Jul 27
ATLAS BECOMES READ-ONLY ON AUG 1; ALL DATA REMOVED ON AUG 15
The Atlas filesystem will be made read-only on Thursday, August 1 and will be decommissioned (with all data permanently removed) on Thursday, August 15. Please begin moving data out of that filesystem and onto the newer Alpine filesystem or to some alternate location immediately. If you have questions about data movement, please contact email@example.com. For more information, please see the 2019 OLCF System Decommissions and Notable Changespage.
ALPINE PURGE POLICY IMPLEMENTATION (JUL 30)
The 90-day purge policy for the Alpine GPFS filesystem will be enabled on July 30, 2019. File purging will begin on that date. On that date, any file over the purge threshold will be eligible for deletion (even if that file was created before July 30).
RHEA NOW MOUNTING ALPINE AND ATLAS
During yesterday’s outage, the new center-wide GPFS filesystem, Alpine, was made available on Rhea. It is currently mounted in addition to the older center-wide Lustre filesystem (Atlas). Rhea will continue to mount both filesystems until August 15, at which time the Atlas filesystem will be unmounted due to its decommission.
NOTE: Since Alpine requires greater on-node resources than Atlas, between 8 and 16GB of memory on each node will be used by the filesystem and unavailable to users.
MEETINGS & WORKSHOPS
PROFILING TOOLS WORKSHOP (AUG 7-9)
The Oak Ridge Leadership Computing Facility (OLCF) will host a Profiling Tools Workshop from August 7-9, 2019. Participants will be guided through performance analysis of real applications on Summit, using well-known tools such as TAU, Score-P, Scalasca, HPCTookit, Extrae, NVprof, and Allinea MAP. Presentations will include using various profiling tools to apply performance analysis, including problems that may occur for specific applications and the required workflow for each tool. The goal of this training workshop is to educate users in the use of various profiling tools. We plan to present the performance analysis of 2-3 real applications with various profiling tools and there will be hands-on sessions.
2019 PETASCALE COMPUTING INSTITUTE (AUG 19-23)
The Oak Ridge Leadership Computing Facility (OLCF) will host a branch of the 2019 Petascale Computing Institute from August 19-23, 2019. The goal of the institute is to enable computational and data-enabled discovery in all fields of study by teaching the participants to scale their computational codes to leadership-class computing systems. The institute will be broadcast to host sites using full-duplex audio/video connections. Participants will receive training accounts, on-site mentoring, and the ability to ask questions of the presenters orally and via on-line chats. Presentations will be given by staff members of the organizing institutions (ALCF, NCSA, NERSC, OLCF, PSC, SciNet – University of Toronto, and TACC), and the keynote address will be given by Gordon Bell. For more information, or to register, please visit the event page.
OLCF GPU HACKATHON SERIES (OPEN CALLS)
OLCF is once again supporting several GPU hackathons. GPU hackathons are 5-day coding events in which teams of developers prepare their own applications(s) to run on GPUs or focus on optimizing their application(s) that currently run on GPUs. Teams should consist of three or more developers who are intimately familiar with (some part of) their application, and they will work alongside two mentors with GPU programming expertise. If you want/need to get your code running (or optimized) on a GPU-accelerated system, these hackathons offer a unique opportunity to set aside 5 days, surround yourself with experts in the field, and push toward your goals. If you are interested in more information, or would like to submit a proposal to one of the upcoming events, please visit the OLCF GPU Hackathons event page.
The OLCF Hackathon (October 21-25 in Knoxville, TN) is the only hackathon still accepting proposals; the deadline to apply is August 16.
WANT TO BE A GPU HACKATHON MENTOR?
As a mentor, you will work with a team of developers that intends to use a programming model (CUDA, OpenACC, OpenMP4.5+, Kokkos, etc.) you are familiar with. Typically, the mentors are paired (2 per team) so that one mentor has a strong GPU-programming background while the other has enough knowledge to help their team but is also there to learn. Regardless of your level of expertise, these events are great opportunities to sharpen your skills, learn about the latest technologies, and collaborate with others in your community. If you would like more information about being a mentor, or if you would like to sign up, please visit https://www.olcf.ornl.gov/gpu-hackathon-mentors/.
UPCOMING SCHEDULED OUTAGES
There are no outages scheduled through July 27.