Anatomy of GPU memory system for multi-application execution

Adwait Jog, Onur Kayiran, Tuba Kesten, Ashutosh Pattnaik, Evgeny Bolotin, Niladrish Chatterjee, Stephen W. Keckler, Mahmut T. Kandemir, Chita R. Das

Research output: Chapter in Book/Report/Conference proceedingConference contribution

33 Scopus citations

Abstract

As GPUS make headway in the computing landscape span- ning mobile platforms, supercomputers, cloud and virtual desktop platforms, supporting concurrent execution of mul- Tiple applications in GPUS becomes essential for unlock- ing their full potential. However, unlike CPUs, multi- Application execution in GPUS is little explored. In this paper, we study the memory system of GPUS in a con- currently executing multi-application environment. We first present an analytical performance model for many-threaded architectures and show that the common use of misses-per- kilo-instruction (MPKI) as a proxy for performance is not accurate without considering the bandwidth usage of ap- plications. We characterize the memory interference of ap- plications and discuss the limitations of existing memory schedulers in mitigating this interference. We extend the analytical model to multiple applications and identify the key metrics to control various performance metrics. We conduct extensive simulations using an enhanced version of GPGPU-Sim targeted for concurrently executing multiple applications, and show that memory scheduling decisions based on MPKI and bandwidth information are more effec- Tive in enhancing throughput compared to the traditional FR-FCFS and the recently proposed RR FR-FCFS policies.

Original languageEnglish (US)
Title of host publicationMEMSYS 2015 - Proceedings of the 1st International Symposium on Memory Systems
PublisherAssociation for Computing Machinery
Pages223-234
Number of pages12
ISBN (Electronic)9781450336048
DOIs
StatePublished - Oct 5 2015
Event1st International Symposium on Memory Systems, MEMSYS 2015 - Washington, United States
Duration: Aug 14 2015Aug 15 2015

Publication series

NameACM International Conference Proceeding Series
Volume05-08-October-2015

Other

Other1st International Symposium on Memory Systems, MEMSYS 2015
CountryUnited States
CityWashington
Period8/14/158/15/15

All Science Journal Classification (ASJC) codes

  • Software
  • Human-Computer Interaction
  • Computer Vision and Pattern Recognition
  • Computer Networks and Communications

Fingerprint Dive into the research topics of 'Anatomy of GPU memory system for multi-application execution'. Together they form a unique fingerprint.

Cite this