Microbenchmarks for GPU characteristics: the occupancy roofline and the pipeline model

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

7 Citations (Scopus)

Abstract

In this paper we present microbenchmarks in OpenCL to measure the most important performance characteristics of GPUs. Microbenchmarks try to measure individual characteristics that influence the performance. First, performance, in operations or bytes per second, is measured with respect to
the occupancy and as such provides an occupancy roofline curve. The curve shows at which occupancy level peak performance is reached. Second, when considering the cycles per instruction of each compute unit, we measure the two most important characteristics of an instruction: its issue and completion latency. This is based on modeling each compute unit as a pipeline
for computations and a pipeline for the memory access. We also measure some specific characteristics: the influence of independent instructions within a kernel and thread divergence. We argue that these are the most important characteristics for understanding the performance and predicting performance. The results for several Nvidia and AMD GPUs are provided. A free java application containing the microbenchmarks is available on www.gpuperformance.org.
Original languageEnglish
Title of host publicationProceedings of 24th Euromicro International Conference on Parallel, Distributed and Network-based Processing (PDP)
PublisherIEEE
Pages456-463
Number of pages8
ISBN (Electronic)978-1-4673-8776-7
ISBN (Print)978-1-4673-8775-0
Publication statusPublished - 18 Feb 2016
EventPDP 2016 - Heraklion, Greece
Duration: 17 Feb 201619 Feb 2016

Conference

ConferencePDP 2016
Country/TerritoryGreece
CityHeraklion
Period17/02/1619/02/16

Keywords

  • microbenchmarks
  • GPU
  • OpenCL
  • Performance Analysis

Fingerprint

Dive into the research topics of 'Microbenchmarks for GPU characteristics: the occupancy roofline and the pipeline model'. Together they form a unique fingerprint.

Cite this