On May 20, 2021, the OLCF and Frontier COE held a Spock Training to help CAAR and ECP project teams get started on using OLCF’s new Spock system. During the training, the OLCF and our vendor partners gave an introduction to the system, including node architecture, available storage, programming environment, running jobs, available tools, and more. The training was recorded and the slides and recordings can be found below. This was an open session so no CORAL2 NDA content was covered.
The Q & A session was managed in a shared Google sheet, a copy of which can be found here.
Time | Title | Speaker |
---|---|---|
11:00 AM – 11:10 AM | Welcome & Logistics | Tom Papatheodore (OLCF) |
11:10 AM – 11:20 AM | COE Updates (slides) |
Noah Reddell (HPE) |
11:20 AM – 12:00 PM | Spock System Architecture (slides | recording) |
Joe Glenski (HPE) |
12:00 PM – 12:30 PM | MI100 GPU (slides | recording) |
Nick Malaya (AMD) |
12:30 PM – 12:45 PM | Available Storage Areas & NVMe (slides | recording) |
Tom Papatheodore (OLCF) |
12:45 PM – 1:00 PM | Break | |
1:00 PM – 1:20 PM | State of HIP (slides | recording) |
Nick Malaya (AMD) |
1:20 PM – 1:30 PM | Programming Environment (slides | recording) |
John Levesque (HPE) |
1:30 PM – 2:05 PM | Compilers (slides | recording) |
Jeff Sandoval (HPE) |
2:05 PM – 2:30 PM | HPE Cray MPICH & GPU-Aware MPI (slides | recording) |
Noah Reddell (HPE) |
2:30 PM – 3:00 PM | Running Jobs (slides | recording) |
Hong Liu (OLCF) & Matt Davis (OLCF) |
3:00 PM – 3:10 PM | Break | |
3:10 PM – 3:40 PM | Node-Level Profiling (slides | recording) |
Julio Maia (AMD) |
3:40 PM – 4:30 PM | Cray Performance & Correctness Tools (slides | recording) |
Kostas Makrides (HPE) |
4:30 PM – 5:00 PM | Spock Tips & Information (slides | recording) |
Tom Papatheodore (OLCF) |