Chapter 28

Jacket

GPU Powered MATLAB Acceleration

Torben Larsen, Gallagher Pryor and James Malcolm

This chapter describes Jacket, which is a software platform for offloading MATLAB computation onto NVIDIA CUDA-capable GPUs. The objective of Jacket is to combine the productivity and simplicity of the MATLAB environment and M-language with the raw computational power of GPUs. We describe the ideas behind Jacket and its basic use, and we provide a simple example that compares MATLAB, CUDA C, and Jacket with respect to code and computational performance. Further, we include a performance evaluation of various Jacket features including CPU-GPU memory transfer, floating point performance, etc.

28.1 Introduction

Jacket is a software platform developed ...

Get GPU Computing Gems Jade Edition 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.