Jefferson Lab Improves Application Performance Case Study
Jefferson Lab more than doubles application performance running optimized code on the Intel® Xeon Phi™ coprocessor.
Researchers from across the United States and around the world use resources at the Thomas Jefferson National Accelerator Facility (Jefferson Lab) to conduct nuclear physics experiments that enhance our understanding of matter. As part of a continuous effort to improve the performance of scientific applications, the lab implemented a new cluster that integrates Intel® Xeon Phi™ coprocessors alongside the Intel® Xeon® processor E5 family. By optimizing software for the Intel Xeon Phi coprocessor, developers have increased performance by more than two times over the most optimized code on Intel Xeon processors, enabling researchers to accelerate projects and address complex scientific questions.
• Accelerate application performance. Adopt technologies that will help improve performance of research applications.
• Maximize the benefit of optimization work. Make the most of programming work by enabling developers to use the same software development models for both coprocessors and industry-standard x86 processors.
• Seneca servers based on the Intel Xeon processor E5 family and Intel Xeon Phi coprocessors. Jefferson Lab installed 16 nodes equipped with Intel Xeon processors E5-2650 and Intel Xeon Phi coprocessors 5110P running on the Red Hat Enterprise Linux* operating system.