ASC23: LINPACK Results

By Dan Olds

 

 

With ISC23 now in the rearview mirror, let’s get back to the results from the ASC23 Student Cluster Competition. In our last articles, we looked at the competition and applications, plus introduced the teams, now it’s time to reveal some of the results and the award winners.

LINPACK is a staple of every cluster competition going back to the first one in 2007. It measures how fast a computer can solve a dense matrix of linear equations. It’s highly parallel and will generally show the maximum performance of a system.

The student teams were scored on their best LINPACK score, with the individual scores normalized to assign competition points. The team with the highest score received 9 points, the maximum, with the trailing scores normalized.

Here are the key results from the ASC23 Highest LINPACK competition:

First Place: Zhejiang University, with their unconventional single-node, eight GPU configuration took home the Highest LINPACK prize and netted nine points. Their system was tailor made for LINPACK, which scales very well on speedy GPUs.

Second Place: University of Science & Technology of China. The home team, USTC, grabbed second place with a normalized score of 8.28 points, which isn’t too far off from the winner. They did well to control the power draw of their four-node, eight GPU, configuration – keeping it just under the 3,000 watt power cap.

Third Place: Shanxi University, who was also running a four-node, eight GPU cluster. The team pulled in a score of 8.10 points, very close to the second-place competitor, but only .07 ahead of the fourth-place ShanghaiTech University.

In addition to a shiny trophy, the winning team Zhejiang also receives prize money of 10,000 CNY (about $1,425 US), which is definitely better than just a trophy alone, right?

 

Contact Us
Technical Support Yu Liu techsupport@asc-events.org
Media Jie He media@asc-events.org
Collaboration Vangel Bojaxhi executive.director@asc-events.org
General Information info@asc-events.org

 

Partners      Follow us
Copyright 2020 Asia Supercomputer Community. All Rights Reserved