Improving the Simulation of Biologically Accurate Neural Networks Using Data Flow HLS Transformations on Heterogeneous SoC-FPGA Platforms

Kaleb Alfaro-Badilla; Andrés Arroyo-Romero; Carlos Salazar-García; Luis G. León-Vega; Javier Espinoza-González; Franklin Hernández-Castro; Alfonso Chacón-Rodríguez; Georgios Smaragdos; Christos Strydis

doi:10.1007/978-3-030-41005-6_13

Improving the Simulation of Biologically Accurate Neural Networks Using Data Flow HLS Transformations on Heterogeneous SoC-FPGA Platforms

Kaleb Alfaro-Badilla^*, Andrés Arroyo-Romero, Carlos Salazar-García, Luis G. León-Vega, Javier Espinoza-González, Franklin Hernández-Castro, Alfonso Chacón-Rodríguez, Georgios Smaragdos, Christos Strydis

^*Corresponding author for this work

Research output: Chapter in Book/Conference proceedings/Edited volume › Conference contribution › Scientific › peer-review

Abstract

This work proposes a hardware performance-oriented design methodology aimed at generating efficient high-level synthesis (HLS) coded data multiprocessing on a heterogeneous platform. The methodology is tested on typical neuroscientific complex application: the biologically accurate modeling of a brain region known as the inferior olivary nucleus (ION). The ION cells are described using a multi-compartmental model based on the extended Hodgkin-Huxley membrane model (eHH), which requires the solution of a set of coupled differential equations. The proposed methodology is tested against alternative HPC implementations (multi-core CPU i7-7820HQ, and a Virtex7 FPGA) of the same ION model for different neural network sizes. Results show that the solution runs 10 to 4 times faster than our previous implementation using the same board and closes the gap between the performance against a Virtex7 implementation without using at full-capacity the AXI-HP channels.

Original language	English
Title of host publication	High Performance Computing - 6th Latin American Conference, CARLA 2019, Revised Selected Papers
Editors	Juan Luis Crespo-Mariño, Esteban Meneses-Rojas
Publisher	Springer
Pages	185-199
Number of pages	15
ISBN (Print)	9783030410049
DOIs	https://doi.org/10.1007/978-3-030-41005-6_13
Publication status	Published - 2020
Externally published	Yes
Event	6th Latin American High Performance Computing Conference, CARLA 2019 - Turrialba, Costa Rica Duration: 25 Sept 2019 → 27 Sept 2019

Publication series

Name	Communications in Computer and Information Science
Volume	1087 CCIS
ISSN (Print)	1865-0929
ISSN (Electronic)	1865-0937

Conference

Conference	6th Latin American High Performance Computing Conference, CARLA 2019
Country/Territory	Costa Rica
City	Turrialba
Period	25/09/19 → 27/09/19

Keywords

Dataflow
FPGA
HLS
HPC
Inferior olivary nucleus
Spiking neural networks

Access to Document

10.1007/978-3-030-41005-6_13

Cite this

Alfaro-Badilla, K., Arroyo-Romero, A., Salazar-García, C., León-Vega, L. G., Espinoza-González, J., Hernández-Castro, F., Chacón-Rodríguez, A., Smaragdos, G., & Strydis, C. (2020). Improving the Simulation of Biologically Accurate Neural Networks Using Data Flow HLS Transformations on Heterogeneous SoC-FPGA Platforms. In J. L. Crespo-Mariño, & E. Meneses-Rojas (Eds.), High Performance Computing - 6th Latin American Conference, CARLA 2019, Revised Selected Papers (pp. 185-199). (Communications in Computer and Information Science; Vol. 1087 CCIS). Springer. https://doi.org/10.1007/978-3-030-41005-6_13

Alfaro-Badilla, Kaleb ; Arroyo-Romero, Andrés ; Salazar-García, Carlos et al. / Improving the Simulation of Biologically Accurate Neural Networks Using Data Flow HLS Transformations on Heterogeneous SoC-FPGA Platforms. High Performance Computing - 6th Latin American Conference, CARLA 2019, Revised Selected Papers. editor / Juan Luis Crespo-Mariño ; Esteban Meneses-Rojas. Springer, 2020. pp. 185-199 (Communications in Computer and Information Science).

@inproceedings{f8a9194f9ab54b998f5aa4af80a5eb40,

title = "Improving the Simulation of Biologically Accurate Neural Networks Using Data Flow HLS Transformations on Heterogeneous SoC-FPGA Platforms",

abstract = "This work proposes a hardware performance-oriented design methodology aimed at generating efficient high-level synthesis (HLS) coded data multiprocessing on a heterogeneous platform. The methodology is tested on typical neuroscientific complex application: the biologically accurate modeling of a brain region known as the inferior olivary nucleus (ION). The ION cells are described using a multi-compartmental model based on the extended Hodgkin-Huxley membrane model (eHH), which requires the solution of a set of coupled differential equations. The proposed methodology is tested against alternative HPC implementations (multi-core CPU i7-7820HQ, and a Virtex7 FPGA) of the same ION model for different neural network sizes. Results show that the solution runs 10 to 4 times faster than our previous implementation using the same board and closes the gap between the performance against a Virtex7 implementation without using at full-capacity the AXI-HP channels.",

keywords = "Dataflow, FPGA, HLS, HPC, Inferior olivary nucleus, Spiking neural networks",

author = "Kaleb Alfaro-Badilla and Andr{\'e}s Arroyo-Romero and Carlos Salazar-Garc{\'i}a and Le{\'o}n-Vega, {Luis G.} and Javier Espinoza-Gonz{\'a}lez and Franklin Hern{\'a}ndez-Castro and Alfonso Chac{\'o}n-Rodr{\'i}guez and Georgios Smaragdos and Christos Strydis",

year = "2020",

doi = "10.1007/978-3-030-41005-6_13",

language = "English",

isbn = "9783030410049",

series = "Communications in Computer and Information Science",

publisher = "Springer",

pages = "185--199",

editor = "Crespo-Mari{\~n}o, {Juan Luis} and Esteban Meneses-Rojas",

booktitle = "High Performance Computing - 6th Latin American Conference, CARLA 2019, Revised Selected Papers",

note = "6th Latin American High Performance Computing Conference, CARLA 2019 ; Conference date: 25-09-2019 Through 27-09-2019",

}

Alfaro-Badilla, K, Arroyo-Romero, A, Salazar-García, C, León-Vega, LG, Espinoza-González, J, Hernández-Castro, F, Chacón-Rodríguez, A, Smaragdos, G & Strydis, C 2020, Improving the Simulation of Biologically Accurate Neural Networks Using Data Flow HLS Transformations on Heterogeneous SoC-FPGA Platforms. in JL Crespo-Mariño & E Meneses-Rojas (eds), High Performance Computing - 6th Latin American Conference, CARLA 2019, Revised Selected Papers. Communications in Computer and Information Science, vol. 1087 CCIS, Springer, pp. 185-199, 6th Latin American High Performance Computing Conference, CARLA 2019, Turrialba, Costa Rica, 25/09/19. https://doi.org/10.1007/978-3-030-41005-6_13

Improving the Simulation of Biologically Accurate Neural Networks Using Data Flow HLS Transformations on Heterogeneous SoC-FPGA Platforms. / Alfaro-Badilla, Kaleb; Arroyo-Romero, Andrés; Salazar-García, Carlos et al.
High Performance Computing - 6th Latin American Conference, CARLA 2019, Revised Selected Papers. ed. / Juan Luis Crespo-Mariño; Esteban Meneses-Rojas. Springer, 2020. p. 185-199 (Communications in Computer and Information Science; Vol. 1087 CCIS).

Research output: Chapter in Book/Conference proceedings/Edited volume › Conference contribution › Scientific › peer-review

TY - GEN

T1 - Improving the Simulation of Biologically Accurate Neural Networks Using Data Flow HLS Transformations on Heterogeneous SoC-FPGA Platforms

AU - Alfaro-Badilla, Kaleb

AU - Arroyo-Romero, Andrés

AU - Salazar-García, Carlos

AU - León-Vega, Luis G.

AU - Espinoza-González, Javier

AU - Hernández-Castro, Franklin

AU - Chacón-Rodríguez, Alfonso

AU - Smaragdos, Georgios

AU - Strydis, Christos

PY - 2020

Y1 - 2020

N2 - This work proposes a hardware performance-oriented design methodology aimed at generating efficient high-level synthesis (HLS) coded data multiprocessing on a heterogeneous platform. The methodology is tested on typical neuroscientific complex application: the biologically accurate modeling of a brain region known as the inferior olivary nucleus (ION). The ION cells are described using a multi-compartmental model based on the extended Hodgkin-Huxley membrane model (eHH), which requires the solution of a set of coupled differential equations. The proposed methodology is tested against alternative HPC implementations (multi-core CPU i7-7820HQ, and a Virtex7 FPGA) of the same ION model for different neural network sizes. Results show that the solution runs 10 to 4 times faster than our previous implementation using the same board and closes the gap between the performance against a Virtex7 implementation without using at full-capacity the AXI-HP channels.

AB - This work proposes a hardware performance-oriented design methodology aimed at generating efficient high-level synthesis (HLS) coded data multiprocessing on a heterogeneous platform. The methodology is tested on typical neuroscientific complex application: the biologically accurate modeling of a brain region known as the inferior olivary nucleus (ION). The ION cells are described using a multi-compartmental model based on the extended Hodgkin-Huxley membrane model (eHH), which requires the solution of a set of coupled differential equations. The proposed methodology is tested against alternative HPC implementations (multi-core CPU i7-7820HQ, and a Virtex7 FPGA) of the same ION model for different neural network sizes. Results show that the solution runs 10 to 4 times faster than our previous implementation using the same board and closes the gap between the performance against a Virtex7 implementation without using at full-capacity the AXI-HP channels.

KW - Dataflow

KW - FPGA

KW - HLS

KW - HPC

KW - Inferior olivary nucleus

KW - Spiking neural networks

UR - http://www.scopus.com/inward/record.url?scp=85081170196&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-41005-6_13

DO - 10.1007/978-3-030-41005-6_13

M3 - Conference contribution

AN - SCOPUS:85081170196

SN - 9783030410049

T3 - Communications in Computer and Information Science

SP - 185

EP - 199

BT - High Performance Computing - 6th Latin American Conference, CARLA 2019, Revised Selected Papers

A2 - Crespo-Mariño, Juan Luis

A2 - Meneses-Rojas, Esteban

PB - Springer

T2 - 6th Latin American High Performance Computing Conference, CARLA 2019

Y2 - 25 September 2019 through 27 September 2019

ER -

Alfaro-Badilla K, Arroyo-Romero A, Salazar-García C, León-Vega LG, Espinoza-González J, Hernández-Castro F et al. Improving the Simulation of Biologically Accurate Neural Networks Using Data Flow HLS Transformations on Heterogeneous SoC-FPGA Platforms. In Crespo-Mariño JL, Meneses-Rojas E, editors, High Performance Computing - 6th Latin American Conference, CARLA 2019, Revised Selected Papers. Springer. 2020. p. 185-199. (Communications in Computer and Information Science). doi: 10.1007/978-3-030-41005-6_13

Improving the Simulation of Biologically Accurate Neural Networks Using Data Flow HLS Transformations on Heterogeneous SoC-FPGA Platforms

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this