Weekly Update: April 6, 2022
In This Message
– Annual GPU Hackathon Series
– Building Containers on Summit
– User Assistance Center Schedule
Meetings & Workshops
– March OLCF User Conference Call is Today (Mar 30)
– Coding for GPUs Using Standard C++ (Apr 7)
– HPC Best Practices Webinar (Apr 13)
– Appentra Codee Training (Apr 26-27)
Upcoming Scheduled Outages
– HPSS (Apr 12)
– Slate (Onyx on April 18, Marble on April 20-21)
Annual GPU Hackathon Series
Do you have an application that needs to be ported to (or optimized on) GPUs? If so, why not work toward your GPU development goals surrounded by experts in the field? Each year, the OLCF works with NVIDIA and other GPU hackathon partners to deliver a series of hackathons around the world. During the events, teams (3 or more members) work alongside mentors with GPU programming expertise to port or optimize their codes on GPUs.
For more information about these events, please visit https://docs.olcf.ornl.gov/training/olcf_gpu_hackathons.html or, if you are already familiar with the events, you can visit https://gpuhackathons.org/events and click “Hackathons” to see the list of upcoming events. Look for the OLCF logo for the OLCF-sponsored hackathons.
Building Containers on Summit
Building container images is now possible on Summit! Users can build a container image with Podman, convert it to a Singularity Image File (SIF) format, and run the container using the Singularity runtime. The full workflow is detailed at https://docs.olcf.ornl.gov/software/containers_on_summit.html.
User Assistance Center Schedule
The OLCF User Assistance Center will be closed on Friday, April 15 due to an ORNL-observed holiday. The normal user assistance hours will resume on Monday, April 18. The Computer Operations staff will remain available for limited troubleshooting 24 hours a day.
Meetings & Workshops
March OLCF User Conference Call
Thanks for joining us for our March OLCF User Conference Call, whihc included a talk from HPE’s Sam Partee entitled Machine Learning for HPC Simulations: Using PyTorch, TensorFlow in Fortran, C, and C++ with SmartSim. The call video has been uploaded to the event page.
Coding for GPUs Using Standard C++ (Apr 7)
Parallelism features available in ISO C++ and ISO Fortran enable developers to write their codes such that the baseline code is parallel and ready to run on any parallel platform they encounter. Join us from 2 PM – 3 PM (EDT) on Thursday, April 7 for the first part of a series that covers how ISO C++ and ISO Fortran can be used as a portable solution to parallel programming targeting both GPUs and CPUs. This training, given by NVIDIA, will be delivered by Zoom and is targeted for OLCF and NERSC users. For this session we’ll demonstrate the current state-of-the-art in writing application code that is parallel and ready to run on GPUs, CPUs, and more, using only C++. For more information or to register, visit https://www.olcf.ornl.gov/calendar/coding-iso-c/.
HPC Best Practices Webinar (Apr 13)
The IDEAS Productivity project, in partnership with the DOE Computing Facilities of the ALCF, OLCF, and NERSC and the DOE Exascale Computing Project (ECP) run a webinar series on Best Practices for HPC Software Developers. The April webinar will be presented by JaeHyuk Kwack of Argonne National Laboratory and is titled Evaluating Performance Portability of HPC Applications and Benchmarks Across Diverse HPC Architectures. The webinar will take place on Wednesday, April 13, 2022 at 1:00 pm ET. The webinar is free and open to the public, but registration is required and is available at https://exascaleproject.zoomgov.com/meeting/register/vJIsdOGprDMjE4X9dnLIY1DleDme3cYyFpI. For more information, see https://www.exascaleproject.org/event/performance-portability-evaluation/.
Appentra Codee Training (Apr 26-27)
NERSC will host training on Appentra’s Codee Analyzer (formerly known as Parallelware Analyzer) on Tuesday, April 26 and Wednesday, April 27 from noon until 3:00 PM (EDT) each day. Appentra Codee Analyzer is a programming development tool for C/C++/Fortran parallel codes on multicore CPUs and GPUs using OpenMP and OpenACC. It provides a systematic, more predictable approach that leverages parallel programming best practices, and enables the novice to write codes at the level of experts, including inserting OpenMP/OpenACC directives to offloading to accelerator devices like GPUs. This training is focused on how the Codee tool provides automated code inspection capabilities for assisting OpenMP/OpenACC programming on GPUs. OLCF and ALCF users are invited and NERSC training accounts will be provided if needed. To register, see https://docs.google.com/forms/d/e/1FAIpQLSfId0XTR5TcLWnaSuHeQbRZCZX4ilwLIw6pycNgWzcgUI2kvg/viewform.
Upcoming Scheduled Outages
- HPSS will be unavailable from 8:00 AM until 4:00 PM on Tuesday, April 12.
- Work will be performed on Onyx from 8:00 AM until 8:00 PM on Monday, April 18 and on Marble from 8:00 AM on Wednesday, April 20 until 8:00 AM on Thursday, April 21. During these times, Jupyter notebooks will be restarted, so in-flight processing will be halted but the notebook state will be stored. Otherwise, workflows should migrate between servers in the respective clusters and user impact is not anticipated.