| Date & Time | 7 June 2017 (Wednesday) 13:00 – 15:00 |
|---|---|
| Title | High Performance Computing Research: Using the Loop Chain Programming Abstraction to Schedule Across Loops in Existing Code |
| Venue | “Kobo Workshop” on the Graduate Program for Embodiment Informatics, 3F, Shinjuku Lambdax Building （Map） |
| Target participants |
All Waseda students, faculty members, and the general public |
| Participant fee |
None |
Professor Michelle Strout
Department of Computer Science University of Arizona USA “High Performance Computing Research: Using the Loop Chain Programming Abstraction to Schedule Across Loops in Existing Code”
Abstract:
High Performance Computing (HPC) research involves many topics such as computational science applications, programming languages, compilers, run-time libraries, operating systems, and computer architecture. In this talk, I will focus on research problems in programming languages and compilers. Specifically, I will present a parallel programming approach called loop chaining that enables a compiler to effectively balance parallelism and data locality. The programmer inserts loop chain pragmas into C/C++ code to control loop schedules at a high level of abstraction. I will show example usage of the pragmas, review attempts to automate the transformation of a legacy scientific code written with specific language constraints to loop chain codes, and present performance results for a computational fluid dynamics benchmark.
