HPC Tunathon for CPU and GPU Optimization and Scaling on JUWELS

A five-day workshop will be held on 25–29 May 2020 at JSC for qualified teams of application developers to optimize and scale their code on the CPU and GPU nodes of JUWELS in preparation for forthcoming exascale systems.

Each team will be paired with and mentored by experts from JSC, Intel, NVIDIA, or the Performance Optimization and Productivity (POP) Centre of Excellence, who will be available to assist with a variety of tools and advice for executions with large numbers of nodes (up to the full system). The focus is on hands-on practical training in a cooperative atmosphere.

All users of JSC computer systems can apply to participate in the HPC Tunathon. However, only a limited number of teams of two or more developers can be accommodated and will be selected based on demonstrated and potential scaling to use large clusters of CPU and GPU nodes such as JUWELS (and JURECA). For further information, including a Tunathon Guide and registration, please visit the webpage https://fz-juelich.de/ias/jsc/2020/HPC-tuna.

Contact: Dr. Brian Wylie, b.wylie@fz-juelich.de

from JSC News, No. 270, 13 March 2020