A fixed point exponential function accelerator for a neuromorphic many-core system

Johannes Partzsch; Sebastian Hoppner; Matthias Eberlein; Rene Schuffny; Christian Mayr; David R. Lester; Steve Furber

doi:10.1109/ISCAS.2017.8050528

A fixed point exponential function accelerator for a neuromorphic many-core system

Johannes Partzsch, Sebastian Hoppner, Matthias Eberlein, Rene Schuffny, Christian Mayr, David R. Lester, Steve Furber

Technische Universitat Dresden

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

700 Downloads (Pure)

Abstract

Many models of spiking neural networks heavily rely on exponential waveforms. On neuromorphic multiprocessor systems like SpiNNaker, they have to be approximated by dedicated algorithms, often dominating the processing load. Here we present a processor extension for fast calculation of exponentials, aimed at integration in the next-generation SpiNNaker system. Our implementation achieves single-LSB precision in a 32bit fixed-point format and 250Mexp/s throughput at 0.44nJ/exp for nominal supply (1.0V), or 0.21nJ/exp at 0.7V supply and 77Mexp/s, demonstrating a throughput multiplication of almost 50 and 98% energy reduction at 2% area overhead per processor on a 28nm CMOS chip.

Original language	English
Title of host publication	IEEE International Symposium on Circuits and Systems
Subtitle of host publication	From Dreams to Innovation, ISCAS 2017 - Conference Proceedings
Publisher	IEEE
ISBN (Electronic)	9781467368520
DOIs	https://doi.org/10.1109/ISCAS.2017.8050528
Publication status	Published - 28 Sept 2017
Event	50th IEEE International Symposium on Circuits and Systems, ISCAS 2017 - Baltimore, United States Duration: 28 May 2017 → 31 May 2017

Conference

Conference	50th IEEE International Symposium on Circuits and Systems, ISCAS 2017
Country/Territory	United States
City	Baltimore
Period	28/05/17 → 31/05/17

Keywords

exponential function
MPSoC
neuromorphic computing
SpiNNaker

Access to Document

10.1109/ISCAS.2017.8050528

PID4705479Accepted author manuscript, 680 KB

Cite this

@inproceedings{76104ec8c1b041b2a3ece404b0feba2e,

title = "A fixed point exponential function accelerator for a neuromorphic many-core system",

abstract = "Many models of spiking neural networks heavily rely on exponential waveforms. On neuromorphic multiprocessor systems like SpiNNaker, they have to be approximated by dedicated algorithms, often dominating the processing load. Here we present a processor extension for fast calculation of exponentials, aimed at integration in the next-generation SpiNNaker system. Our implementation achieves single-LSB precision in a 32bit fixed-point format and 250Mexp/s throughput at 0.44nJ/exp for nominal supply (1.0V), or 0.21nJ/exp at 0.7V supply and 77Mexp/s, demonstrating a throughput multiplication of almost 50 and 98% energy reduction at 2% area overhead per processor on a 28nm CMOS chip.",

keywords = "exponential function, MPSoC, neuromorphic computing, SpiNNaker",

author = "Johannes Partzsch and Sebastian Hoppner and Matthias Eberlein and Rene Schuffny and Christian Mayr and Lester, {David R.} and Steve Furber",

year = "2017",

month = sep,

day = "28",

doi = "10.1109/ISCAS.2017.8050528",

language = "English",

booktitle = "IEEE International Symposium on Circuits and Systems",

publisher = "IEEE",

address = "United States",

note = "50th IEEE International Symposium on Circuits and Systems, ISCAS 2017 ; Conference date: 28-05-2017 Through 31-05-2017",

}

Partzsch, J, Hoppner, S, Eberlein, M, Schuffny, R, Mayr, C, Lester, DR & Furber, S 2017, A fixed point exponential function accelerator for a neuromorphic many-core system. in IEEE International Symposium on Circuits and Systems: From Dreams to Innovation, ISCAS 2017 - Conference Proceedings., 8050528, IEEE, 50th IEEE International Symposium on Circuits and Systems, ISCAS 2017, Baltimore, United States, 28/05/17. https://doi.org/10.1109/ISCAS.2017.8050528

A fixed point exponential function accelerator for a neuromorphic many-core system. / Partzsch, Johannes; Hoppner, Sebastian; Eberlein, Matthias et al.
IEEE International Symposium on Circuits and Systems: From Dreams to Innovation, ISCAS 2017 - Conference Proceedings. IEEE, 2017. 8050528.

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

TY - GEN

T1 - A fixed point exponential function accelerator for a neuromorphic many-core system

AU - Partzsch, Johannes

AU - Hoppner, Sebastian

AU - Eberlein, Matthias

AU - Schuffny, Rene

AU - Mayr, Christian

AU - Lester, David R.

AU - Furber, Steve

PY - 2017/9/28

Y1 - 2017/9/28

N2 - Many models of spiking neural networks heavily rely on exponential waveforms. On neuromorphic multiprocessor systems like SpiNNaker, they have to be approximated by dedicated algorithms, often dominating the processing load. Here we present a processor extension for fast calculation of exponentials, aimed at integration in the next-generation SpiNNaker system. Our implementation achieves single-LSB precision in a 32bit fixed-point format and 250Mexp/s throughput at 0.44nJ/exp for nominal supply (1.0V), or 0.21nJ/exp at 0.7V supply and 77Mexp/s, demonstrating a throughput multiplication of almost 50 and 98% energy reduction at 2% area overhead per processor on a 28nm CMOS chip.

AB - Many models of spiking neural networks heavily rely on exponential waveforms. On neuromorphic multiprocessor systems like SpiNNaker, they have to be approximated by dedicated algorithms, often dominating the processing load. Here we present a processor extension for fast calculation of exponentials, aimed at integration in the next-generation SpiNNaker system. Our implementation achieves single-LSB precision in a 32bit fixed-point format and 250Mexp/s throughput at 0.44nJ/exp for nominal supply (1.0V), or 0.21nJ/exp at 0.7V supply and 77Mexp/s, demonstrating a throughput multiplication of almost 50 and 98% energy reduction at 2% area overhead per processor on a 28nm CMOS chip.

KW - exponential function

KW - MPSoC

KW - neuromorphic computing

KW - SpiNNaker

UR - http://www.scopus.com/inward/record.url?scp=85032674665&partnerID=8YFLogxK

U2 - 10.1109/ISCAS.2017.8050528

DO - 10.1109/ISCAS.2017.8050528

M3 - Conference contribution

AN - SCOPUS:85032674665

BT - IEEE International Symposium on Circuits and Systems

PB - IEEE

T2 - 50th IEEE International Symposium on Circuits and Systems, ISCAS 2017

Y2 - 28 May 2017 through 31 May 2017

ER -

A fixed point exponential function accelerator for a neuromorphic many-core system

Abstract

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this