High Performance Computing (HPC) Working Group
The NAFEMS High Performance Computing Working Group (HPC-WG) was formed in 2011, in response to increasing demand from NAFEMS members for activities related to High Performance Computing.
The mission of the High Performance Computing-Working Group is:
“To provide a vendor-neutral, end-user driven consortium that promotes the effective use of High Performance Computing in engineering simulation.”
This includes education, communication, promotion of standards, and development of requirements that will have general benefits to the simulation and analysis community.
The High Performance Computing Working Group provides a forum for everyone with an interest in High Performance Computing to discuss issues of mutual interest, while at the same time initiating and directing activities of benefit to NAFEMS members as a whole through publications, seminars, and workshops. The Group also acts as NAFEMS’ technology centre in this area and interacts with industry and academic experts in order to provide an authoritative response to technical issues within its remit. The Group works closely with other working groups and seeks to complement their activities.
Terms of Reference
At NAFEMS, High Performance Computing is used as an umbrella term for a range of technologies such as traditional Supercomputing, Grid Computing, Cloud Computing, High Throughput Computing, Hardware Acceleration (such as GPGPUs), Data Storage and Visualization.
Priority Issues Being Addressed
The current priorities are to:
Promote dialogue between hardware vendors, software vendors, industrial users and academics to improve the scalability of algorithms and data formats used in engineering simulation, particularly for computational structural mechanics.
There is a growing gap between the most powerful computing platforms and the capability of both commercial and academic software.
Develop an independent database of realistic test problems for performance benchmarking.
Hardware and software vendors have a tendency to select test problems that show their products in the best light. A selection of realistic problems will enable end users to procure the most appropriate selection of hardware and software for their needs.