IN THIS MESSAGE

Center Announcements
  – IBM Watson Machine Learning CE 1.7.0 Upgrade

Meetings & Workshops
  – OLCF User Conference Call (Feb 26)
  – NVIDIA Profiling Tools – Nsight Systems (Mar 9)
  – NVIDIA Profiling Tools – Nsight Compute (Mar 10)
  – OpenMP Hackathon at Georgia Tech (Apr 20-24)
  – Kokkos Bootcamp & Training (Apr 21-24)
  – GPU Hackathon Series
  – CUDA Training Series
  – OpenACC Training Series

Upcoming Scheduled Outages
  – Summit, Rhea, DTNs, Alpine, HPSS (Mar 10)
  – Home (Mar 18)

CENTER ANNOUNCEMENTS

IBM WATSON MACHINE LEARNING CE 1.7.0 UPGRADE
IBM Watson Machine Learning CE 1.7.0 will be made available on Summit on March 6 and will be set as the default version. This release will include the latest version of each framework, including TensorFlow 2.1, which may require code changes to existing TensorFlow 1.x scripts since the new version may break existing scripts. This does not affect users using custom conda environments. To continue using TensorFlow 1.15, load the previous IBM WML CE module with ‘module load ibm-wml-ce/1.6.2-3’. For more information on migrating to TensorFlow 2, see the migration guide. For more information on IBM WML CE on Summit, see https://docs.olcf.ornl.gov/software/analytics/ibm-wml-ce.html.

MEETINGS & WORKSHOPS

OLCF USER CONFERENCE CALL (FEB 26)
Thanks for joining us for the February OLCF User Conference Call, during which OLCF’s Bill Renaud presented “OLCF Overview for NewUsers”. To review the presentation, visit the event page. The slides and a link to the call video are available there.

NVIDIA PROFILING TOOLS – NSIGHT SYSTEMS (MAR 9)
On March 9, NVIDIA will present a webinar on how to use NVIDIA’s Nsight Systems – a statistical sampling profiler with tracing features – on Summit. Nsight Systems and Nsight Compute are NVIDIA’s next-generation profiling tools for understanding and optimizing the performance of CUDA, OpenACC, or OpenMP applications. NVIDIA recommends transitioning to these new tools since nvprof and Visual profiler will be deprecated in a future CUDA release. The presentation will be delivered remotely but there will be an in-person viewing of the webinar for participants with current ORNL badges. For more information or to register, please visit the event page.

NVIDIA PROFILING TOOLS – NSIGHT COMPUTE (MAR 10)
On March 10, NVIDIA will present a webinar on how to use NVIDIA’s Nsight Compute – a kernel-level analysis and performance metric tool – on Summit. Nsight Systems and Nsight Compute are NVIDIA’s next-generation profiling tools for understanding and optimizing the performance of CUDA, OpenACC, or OpenMP applications. NVIDIA recommends transitioning to these new tools since nvprof and Visual profiler will be deprecated in a future CUDA release. The presentation will be delivered remotely but there will be an in-person viewing of the webinar for participants with current ORNL badges. For more information or to register, please visit the event page.

OPENMP HACKATHON at GEORGIA TECH (APR 20-24)
The Electrical Engineering and Computer Science Department at Georgia Institute of Technology in conjunction with Oak Ridge National Laboratory (ORNL) is organizing an ECP OpenMP Hackathon on April 20–April 24, 2020. This event is sponsored by the Exascale Computing Project (ECP), and driven by the ECP SOLLVE Project. We encourage participation of teams especially interested in porting and optimizing their applications by using the latest OpenMP features. More information can be found at https://sites.google.com/view/omp-hack-atl/home.

KOKKOS BOOTCAMP & TRAINING (21-24 APR)
The OLCF will host a Kokkos training event organized by the Exascale Computing Project (ECP) on April 21-24, 2020. This workshop is intended to teach new Kokkos users how to get started and to help existing Kokkos users to further improve their codes. The training will cover the minimum required topics to get your application started on using Kokkos, and Kokkos experts will be on hand to help the more advanced users. The deadline to apply is March 20, 2020. Please note that ECP application teams will be given priority for the available seats until March 6th. Afterward, all registrations will be given equal priority on a first-come, first-served basis. If you would like more information, or to register, please visit the event page.

GPU HACKATHON SERIES
The following organizations, in conjunction with OLCF and NVIDIA, will be hosting GPU Hackathons as part of our annual GPU Hackathon Series. These hackathons are 5-day coding events in which teams of developers prepare their own application(s) to run on GPUs or focus on optimizing their application(s) that currently run on GPUs. Teams typically consist of three or more developers who are intimately familiar with (some part of) their application, and they work alongside two mentors with GPU programming expertise. These hackathons offer a unique opportunity for teams to set aside 5 days, surround themselves with experts in the field, and push toward their development goals. During the event, teams will have access to compute resources provided by OLCF. If you’re interested in more information, or to apply, you can visit the site-specific event pages below. If you have questions, please contact Tom Papatheodore.

  San Diego Supercomputer Center
   – https://hpc-training.sdsc.edu/gpuhackathon/
   – Event Dates: May 11-15, 2020
   – Deadline to apply: March 11, 2020

  Princeton University
   – https://researchcomputing.princeton.edu/events/gpu-hackathon-princeton
   – Event Dates: June 8-12, 2020
   – Deadline to apply: March 2, 2020

  NERSC
   – https://sites.google.com/lbl.gov/nersc-gpuhackathon
   – Event Dates: July 13-17, 2020
   – Deadline to apply: May 13, 2020

CUDA TRAINING SERIES
NVIDIA will present a 9-part CUDA training series intended to help new and existing GPU programmers understand the main concepts of the CUDA platform and its programming model. Each part will include a 1-hour presentation and example exercises. The exercises are meant to reinforce the material from the presentation and can be completed during a 1-hour hands-on session following each lecture (for in-person participants) or on your own (for remote participants). The full list of topics can be found on the series page. If you have any questions about this series, please contact Tom Papatheodore. For more information on the next 2 events in this series, or to register, please visit the links below:

  Fundamental CUDA Optimization – 1 of 2 (Wednesday, March 18)
  Fundamental CUDA Optimization – 2 of 2 (Thursday, April 16)

OPENACC TRAINING SERIES
NVIDIA will present a 3-part OpenACC training series intended to help new and existing GPU programmers learn to use the OpenACC API. Each part will include a 1-hour presentation and example exercises. The exercises are meant to reinforce the material from the presentation and can be completed during a 1-hour hands-on session following each lecture (for in-person participants) or on your own (for remote participants). The full list of events can be found on the series page. If you have any questions about this series, please contact Tom Papatheodore. For more information on the individual events in this series, or to register, please visit the links below:

  Introduction to OpenACC (Friday, April 17)
  OpenACC Data Management (Thursday, May 28)
  Loop Optimizations with OpenACC (Tuesday, June 23)

UPCOMING SCHEDULED OUTAGES

Summit, Rhea, the Data Transfer Nodes, the Alpine filesystem and HPSS will be unavailable beginning at 8:00 AM on Tuesday, March 10. HPSS will return to service at 8:00 PM, and all other systems will return at 11:59 PM on Tuesday, March 10.

The general-purpose login server (home.ccs.ornl.gov) will be rebooted between 6 PM and 9 PM on Wednesday, March 18. This will be a very brief outage.