Performance Evaluation and Analysis Consortium (PEAC) End Station

PI Name: 
Leonid Oliker
Institution: 
Lawrence Berkeley National Laboratory
Allocation Program: 
INCITE
Allocation Hours at ALCF: 
25 Million
Year: 
2017
Research Domain: 
Computer Science

Understanding how to efficiently use DOE leadership class systems is important due to the challenges of effectively managing extreme levels of concurrency as well as architectural heterogeneity. The performance research community can provide critical tools, runtimes, and methodologies that scientists can use to exploit leadership class machines. The team will have a Performance Evaluation and Analysis Consortium (PEAC) End Station to complete this project.

To facilitate further understanding of leadership class systems, Oliker’s team will develop new programming models and runtime systems. Consortium members will conduct research into the programming models, runtime systems, tools, system evaluations, and application analysis that support computational science on leadership computing platforms. Ultimately, this will allow scientists to maximize the speed—and therefore the impact—of these large-scale platforms.