Performance Portability with Data-Centric Parallel Programming
(Presentation - presented in Rio de Janeiro, Brasil, May 2019) Keynote talk at the The Ninth International Workshop on Accelerators and Hybrid Exascale Systems (AsHES) (delayed online)
Abstract
The ubiquity of accelerators in high-performance computing has driven programming complexity beyond the skill-set of the average domain scientist. To maintain performance portability in the future, it is imperative to decouple architecture-specific programming paradigms from the underlying scientific computations. We present the Stateful DataFlow multiGraph (SDFG), a data-centric intermediate representation that enables separating code definition from its optimization. We show how to tune several applications in this model and IR. Furthermore, we show a global, datacentric view of a state-of-the-art quantum transport simulator to optimize its execution on supercomputers. The approach yields coarse and fine-grained data-movement characteristics, which are used for performance and communication modeling, communication avoidance, and data-layout transformations. The transformations are tuned for the Piz Daint and Summit supercomputers, where each platform requires different caching and fusion strategies to perform optimally. We show that SDFGs deliver competitive performance, allowing domain scientists to develop applications naturally and port them to approach peak hardware performance without modifying the original scientific code.
Documents
download slides:
Recorded talk (best effort)
BibTeX
@misc{hoefler-isc-ashes19, author={Torsten Hoefler}, title={{Performance Portability with Data-Centric Parallel Programming}}, year={2019}, month={May}, location={Rio de Janeiro, Brasil}, source={http://www.unixer.de/~htor/publications/}, }
