Portable and transparent software managed scheduling on accelerators for fair resource sharing

Accelerators, such as Graphic Processing Units (GPUs), are popular components of modern parallel systems. Their energy-efficient performance make them attractive components for modern data center nodes. However, they lack control for fair resource sharing amongst multiple users. This paper presents...

Full description

Saved in:
Bibliographic Details
Published in2016 IEEE/ACM International Symposium on Code Generation and Optimization (CGO) pp. 82 - 93
Main Authors Margiolas, Christos, O'Boyle, Michael F. P.
Format Conference Proceeding
LanguageEnglish
Published ACM 01.03.2016
Subjects
accelerator sharing
accelerators
fair resource sharing
GPUs
Graphics processing units
Hardware
heterogeneous computing
Kernel
multi-tasking
OpenCL
Registers
Resource management
Throughput
Online AccessGet full text

Cover

More Information
Summary:Accelerators, such as Graphic Processing Units (GPUs), are popular components of modern parallel systems. Their energy-efficient performance make them attractive components for modern data center nodes. However, they lack control for fair resource sharing amongst multiple users. This paper presents a runtime and Just In Time compiler that enables resource sharing control and software managed scheduling on accelerators. It is portable and transparent, requiring no modification or recompilation of existing systems or user applications. We provide an extensive evaluation of our scheme with over 40,000 different workloads on 2 platforms and we deliver fairness improvements ranging from 6.8× to 13.66×. In addition, we also deliver system throughput speedups ranging from 1.13× to 1.31×.