Intel is donating 15 Xeon Phi 5110P boards, $25K for host systems, and access to their software engineering tools in support of an effort to develop parallel computing curricular materials. The system will be used in the Fall 2015 offering of CS 5220, Applications of Parallel Computers, and material from the class will be made available online.
CS5220 is taken not only by CS majors, but by a wide variety of computationally-oriented graduate students from fields across campus who use high-performance computing in their research. Since the course is currently offered every other year, I know that the materials I’ve put online from previous offerings have been used locally for self-study by several graduate students.
The Xeon Phi accelerators are becoming more common in a variety of high-performance computing installations, both in clusters operated in academia and research and in the big supercomputers operated by NSF and DOE. Xeon Phi accelerators are available in the Stampede system at Texas and in the supercomputer at Minessota, and are a major part of the next-generation Aurora supercomputer that Intel recently contracted to build at Argonne National Lab. I’m very excited to be able to help train our students to be able to use these types of resources.
Thanks to a discount offered by Dell (since this is for a course, with another vendor also donating), the boards will be housed in an 8-node dual-CPU cluster, so the total number of CPUs available will be
Excerpted from an email sent in June when the grant came through.