Weekly Update: February 5, 2020
IN THIS MESSAGE
– OpenMPI Upgrade on Rhea (Feb 18)
Meetings & Workshops
– January OLCF User Conference Call (Jan 29)
– Scaling Up Deep Learning Applications on Summit (Feb 10)
– jsrun Tutorial (Feb 18)
Upcoming Scheduled Outages
– Rhea (Feb 18)
OPENMPI UPGRADE ON RHEA (FEB 18)
Rhea’s default OpenMPI will be upgraded to version 3.1.4 on February 18. Due to changes that will be made to underlying system libraries during the outage, all codes built against OpenMPI must be rebuilt once Rhea returns to service.
MEETINGS & WORKSHOPS
JANUARY OLCF USER CONFERENCE CALL (JAN 29)
Thanks for joining us for the January OLCF User Conference Call, which included a discussion of the new MyOLCF site led by OLCF’s Adam Carlyle. Links to the slide deck and the video will be made available on the event page if you would like to review the presentation.
SCALING UP DEEP LEARNING APPLICATIONS ON SUMMIT (FEB 10)
The Oak Ridge Leadership Computing Facility (OLCF) will hold a webinar titled “Scaling Up Deep Learning Applications on Summit” on Monday, February 10, 2020. During the event, presenters from OLCF and our vendor partners will offer recommendations for deploying and scaling deep learning (DL) frameworks on the Summit supercomputer. The intended audience is OLCF users who are already familiar with DL applications but are new to high-performance computing (i.e, someone who already knows how to run DL codes on a workstation but is not sure how to scale up on a cluster), although attendance is open to all users. In-person attendance will be available for participants who already have current ORNL badges.
JSRUN TUTORIAL (FEB 18)
OLCF will hold a jsrun tutorial on Tuesday, February 18, 2020. This event is intended to teach new and existing OLCF users how to use the jsrun job launcher on Summit. We will begin with an introduction to jsrun – to bring new users up to speed – followed by more advanced usage (e.g., using Explicit Resource Files (ERF)). After the presentations, participants will be encouraged to work through example problems during a hands-on session.
Both in-person and remote participation will be supported during the tutorial, but only in-person participants will be given temporary training accounts on Summit (for those who do not already have access). The connection details will be sent out to remote participants the day before the event.
UPCOMING SCHEDULED OUTAGES
Rhea will be unavailable from 8:00 AM until noon on Tuesday, February 18.