A new implementation of fully programmable complete CNN processor core on FPGA

Volkan Meriç; Müştak E. Yalçın

A new implementation of fully programmable complete CNN processor core on FPGA

Volkan Meriç, Müştak E. Yalçın

Department of Electronics and Communication Engineering

Istanbul Technical University

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

2 Citations (Scopus)

Abstract

In this paper, a new Cellular Neural Network (CNN) processing core architecture for digital emulation of discrete time CNN is proposed. The introduced CNN core processor is capable of executing 3 × 3 template operation which is considered as instruction. The instruction performs on input and initial images where are stored on DDR RAM. A result of the instruction can be stored at different memory segments. Data transfer mechanism between DDR memory and CNN Core is design to have maximum throughput. The architecture has been combined the core unit with camera and camera control units. CNN Processor can process up to 1600 × 900 video @15 fps. System performance is depend on iteration count n which can be also defined in the instruction. Furthermore, FPGA implementation of CNN processor core with camera control units is given and the introduced system is experimentally confirmed on Spartan 6 XC6SLX45 FPGA.

Original language	English
Title of host publication	CNNA 2016 - 15th International Workshop on Cellular Nanoscale Networks and Their Applications
Editors	Ronald Tetzlaff
Publisher	IEEE Computer Society
Pages	31-32
Number of pages	2
ISBN (Electronic)	9783800742523
Publication status	Published - 2016
Event	15th International Workshop on Cellular Nanoscale Networks and Their Applications, CNNA 2016 - Dresden, Germany Duration: 23 Aug 2016 → 25 Aug 2016

Publication series

Name	International Workshop on Cellular Nanoscale Networks and their Applications
Volume	2016-August
ISSN (Print)	2165-0160
ISSN (Electronic)	2165-0179

Conference

Conference	15th International Workshop on Cellular Nanoscale Networks and Their Applications, CNNA 2016
Country/Territory	Germany
City	Dresden
Period	23/08/16 → 25/08/16

Bibliographical note

Publisher Copyright:
© VDE VERLAG GMBH.

Cite this

@inproceedings{a5844f852643461d955e4d16cd1e300b,

title = "A new implementation of fully programmable complete CNN processor core on FPGA",

abstract = "In this paper, a new Cellular Neural Network (CNN) processing core architecture for digital emulation of discrete time CNN is proposed. The introduced CNN core processor is capable of executing 3 × 3 template operation which is considered as instruction. The instruction performs on input and initial images where are stored on DDR RAM. A result of the instruction can be stored at different memory segments. Data transfer mechanism between DDR memory and CNN Core is design to have maximum throughput. The architecture has been combined the core unit with camera and camera control units. CNN Processor can process up to 1600 × 900 video @15 fps. System performance is depend on iteration count n which can be also defined in the instruction. Furthermore, FPGA implementation of CNN processor core with camera control units is given and the introduced system is experimentally confirmed on Spartan 6 XC6SLX45 FPGA.",

author = "Volkan Meri{\c c} and Yal{\c c}ın, {M{\"u}{\c s}tak E.}",

note = "Publisher Copyright: {\textcopyright} VDE VERLAG GMBH.; 15th International Workshop on Cellular Nanoscale Networks and Their Applications, CNNA 2016 ; Conference date: 23-08-2016 Through 25-08-2016",

year = "2016",

language = "English",

series = "International Workshop on Cellular Nanoscale Networks and their Applications",

publisher = "IEEE Computer Society",

pages = "31--32",

editor = "Ronald Tetzlaff",

booktitle = "CNNA 2016 - 15th International Workshop on Cellular Nanoscale Networks and Their Applications",

address = "United States",

}

Meriç, V & Yalçın, ME 2016, A new implementation of fully programmable complete CNN processor core on FPGA. in R Tetzlaff (ed.), CNNA 2016 - 15th International Workshop on Cellular Nanoscale Networks and Their Applications. International Workshop on Cellular Nanoscale Networks and their Applications, vol. 2016-August, IEEE Computer Society, pp. 31-32, 15th International Workshop on Cellular Nanoscale Networks and Their Applications, CNNA 2016, Dresden, Germany, 23/08/16.

A new implementation of fully programmable complete CNN processor core on FPGA. / Meriç, Volkan; Yalçın, Müştak E.
CNNA 2016 - 15th International Workshop on Cellular Nanoscale Networks and Their Applications. ed. / Ronald Tetzlaff. IEEE Computer Society, 2016. p. 31-32 (International Workshop on Cellular Nanoscale Networks and their Applications; Vol. 2016-August).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - A new implementation of fully programmable complete CNN processor core on FPGA

AU - Meriç, Volkan

AU - Yalçın, Müştak E.

PY - 2016

Y1 - 2016

N2 - In this paper, a new Cellular Neural Network (CNN) processing core architecture for digital emulation of discrete time CNN is proposed. The introduced CNN core processor is capable of executing 3 × 3 template operation which is considered as instruction. The instruction performs on input and initial images where are stored on DDR RAM. A result of the instruction can be stored at different memory segments. Data transfer mechanism between DDR memory and CNN Core is design to have maximum throughput. The architecture has been combined the core unit with camera and camera control units. CNN Processor can process up to 1600 × 900 video @15 fps. System performance is depend on iteration count n which can be also defined in the instruction. Furthermore, FPGA implementation of CNN processor core with camera control units is given and the introduced system is experimentally confirmed on Spartan 6 XC6SLX45 FPGA.

AB - In this paper, a new Cellular Neural Network (CNN) processing core architecture for digital emulation of discrete time CNN is proposed. The introduced CNN core processor is capable of executing 3 × 3 template operation which is considered as instruction. The instruction performs on input and initial images where are stored on DDR RAM. A result of the instruction can be stored at different memory segments. Data transfer mechanism between DDR memory and CNN Core is design to have maximum throughput. The architecture has been combined the core unit with camera and camera control units. CNN Processor can process up to 1600 × 900 video @15 fps. System performance is depend on iteration count n which can be also defined in the instruction. Furthermore, FPGA implementation of CNN processor core with camera control units is given and the introduced system is experimentally confirmed on Spartan 6 XC6SLX45 FPGA.

UR - http://www.scopus.com/inward/record.url?scp=85073197045&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:85073197045

T3 - International Workshop on Cellular Nanoscale Networks and their Applications

SP - 31

EP - 32

BT - CNNA 2016 - 15th International Workshop on Cellular Nanoscale Networks and Their Applications

A2 - Tetzlaff, Ronald

PB - IEEE Computer Society

T2 - 15th International Workshop on Cellular Nanoscale Networks and Their Applications, CNNA 2016

Y2 - 23 August 2016 through 25 August 2016

ER -

A new implementation of fully programmable complete CNN processor core on FPGA

Abstract

Publication series

Conference

Bibliographical note

Other files and links

Fingerprint

Cite this