VLIW code generation for a Convolutional Network Accelerator

M.C.J. Peemen, W. Wisnu Pramadi, B. Mesman, H. Corporaal

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

2 Citations (Scopus)
283 Downloads (Pure)

Abstract

This paper presents a compiler flow to map Deep Convolutional Networks (ConvNets) to a highly specialized VLIW accelerator core targeting the low-power embedded market. Earlier works have focused on energy efficient accelerators for this class of algorithms, but none of them provides a complete and practical programming model. Due to the large parameter set of a ConvNet it is essential that the user can abstract from the accelerator architecture and does not have to rely on an error prone and ad-hoc assembly programming model. By using modulo scheduling for software pipelining we demonstrate that our automatic generated code achieves equal or within 5-20% less hardware utilization w.r.t. code written manually by experts. Our compiler removes the huge manual workload to efficiently map ConvNets to an energy-efficient core for the next-generation mobile and wearable devices.
Original languageEnglish
Title of host publicationProceedings of the 18th International Workshop on Software and Compilers for Embedded Systems, SCOPES 2015, 1-3 June 2015, St. Goar, Germany
EditorsS. Stuijk
Place of PublicationNew York
PublisherAssociation for Computing Machinery, Inc
Pages117-120
ISBN (Print)978-1-4503-3593-5
DOIs
Publication statusPublished - 2015
Event18th International Workshop on Software and Compilers for Embedded Systems (SCOPES 2015) - Schloss Rheinfels, St. Goar, Germany
Duration: 1 Jun 20153 Jun 2015
Conference number: 18
http://www.scopesconf.org/scopes-15/

Workshop

Workshop18th International Workshop on Software and Compilers for Embedded Systems (SCOPES 2015)
Abbreviated titleSCOPES 2015
Country/TerritoryGermany
CitySt. Goar
Period1/06/153/06/15
Internet address

Fingerprint

Dive into the research topics of 'VLIW code generation for a Convolutional Network Accelerator'. Together they form a unique fingerprint.

Cite this