High Performance Python Offloading
Jussi Enkovaara*; Michael Klemm†; Freddie Witherden‡ * CSC, Finland† Intel Corporation, Germany‡ Imperial College of London, United Kingdom
Abstract
This chapter shows how to utilize the compute power of the Intel Xeon Phi coprocessor from Python HPC applications. The pyMIC module provides an easy-to-use, flexible way to offload application kernels to the coprocessor by supporting a streaming interface for asynchronous data transfers and kernel execution. The two example applications, GPAW and PyFR, show how pyMIC can be employed in application scenarios that have differing needs.
Keywords
Python
OpenMP
Offload
Quantum physics
CFD
Simulations
Acknowledgments
The work on GPAW is supported by ...
Get High Performance Parallelism Pearls Volume Two now with the O’Reilly learning platform.
O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.
