Energy and power considerations of GPUs
J. Coplin; M. Burtscher Texas State University, San Marcos, TX, United States
Abstract
This chapter discusses the impact of different algorithm implementations, program inputs, core and memory clock frequencies, and source-code optimizations on the energy consumption, power draw, and performance of a modern compute GPU. We distinguish between memory- and compute-bound codes as well as regular and irregular codes and point out behavioral differences between these classes of programs. We present examples of software approaches to alter the energy, power, and runtime of GPU kernels and explain how they can be employed by the end user to improve energy efficiency.
Keywords
General-purpose ...
Get Advances in GPU Research and Practice 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.
