Low precision processing for high order stencil computations

Gagandeep Singh, Dionysios Diamantopoulos, Sander Stuijk, Christoph Hagleitner, Henk Corporaal

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

3 Downloads (Pure)

Abstract

Modern scientific workloads have demonstrated the inefficiency of using high precision formats. Moving to a lower bit format or even to a different number system can provide tremendous gains in terms of performance and energy efficiency. In this article, we explore the applicability of different number formats and exhaustively search for the appropriate bit width for 3D complex stencil kernels, which are one of the most widely used scientific kernels. Further, we demonstrate the achievable performance of these kernels on state-of-the-art hardware that includes CPU and FPGA, which is the only hardware supporting arbitrary fixed-point precision. Thus, this work fills the gap between current hardware capabilities and future systems for stencil-based scientific applications.

Original languageEnglish
Title of host publicationEmbedded Computer Systems
Subtitle of host publicationArchitectures, Modeling, and Simulation - 19th International Conference, SAMOS 2019, Proceedings
EditorsDionisios N. Pnevmatikatos, Maxime Pelcat, Matthias Jung
Place of PublicationCham
PublisherSpringer
Pages403-415
Number of pages13
ISBN (Electronic)978-3-030-27562-4
ISBN (Print)978-3-030-27561-7
DOIs
Publication statusPublished - 1 Jan 2019
Event19th International Conference on Embedded Computer Systems: Architectures, Modeling, and Simulation, SAMOS 2019 - Samos, Greece
Duration: 7 Jul 201911 Jul 2019

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume11733 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference19th International Conference on Embedded Computer Systems: Architectures, Modeling, and Simulation, SAMOS 2019
CountryGreece
CitySamos
Period7/07/1911/07/19

    Fingerprint

Cite this

Singh, G., Diamantopoulos, D., Stuijk, S., Hagleitner, C., & Corporaal, H. (2019). Low precision processing for high order stencil computations. In D. N. Pnevmatikatos, M. Pelcat, & M. Jung (Eds.), Embedded Computer Systems: Architectures, Modeling, and Simulation - 19th International Conference, SAMOS 2019, Proceedings (pp. 403-415). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11733 LNCS). Cham: Springer. https://doi.org/10.1007/978-3-030-27562-4_29