Improving efficiency of hybrid flexible streaming accelerator systems via memory and interrupt optimizations

Alexander Strachan; Nigel Topham

doi:10.1109/ICFSP67350.2025.11353742

Improving efficiency of hybrid flexible streaming accelerator systems via memory and interrupt optimizations

Alexander Strachan, Nigel Topham

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

Abstract

Hybrid accelerator systems have the potential to increase the useful lifespan of hardware designs. These systems allow parts of a hardware accelerator to be replaced with software running on a general purpose CPU. This allows the design to be extended or fixed in the field, after the fact, compared to traditional fixed function hardware accelerators.However, when implemented naively, such systems have significant overheads due to cache and interrupt management, limited memory bandwidth, and single threaded operation. This paper details three mechanisms that can reduce these overheads, and evaluates their impact using full-system simulation of an ARM-based embedded multicore architecture, extended with a cycle-accurate hardware accelerator. We show that efficiency can be increased by 5-20%, leading to decreased power and energy consumption, extending the potential applications of such systems.

Original language	English
Title of host publication	2025 10th International Conference on Frontiers of Signal Processing (ICFSP)
Publisher	Institute of Electrical and Electronics Engineers
Pages	49-54
Number of pages	6
ISBN (Electronic)	9798331574420
ISBN (Print)	9798331574437
DOIs	https://doi.org/10.1109/ICFSP67350.2025.11353742
Publication status	Published - 29 Jan 2026
Event	The 10th International Conference on Frontiers of Signal Processing - ECE Paris, Paris, France Duration: 10 Sept 2025 → 12 Sept 2025 Conference number: 10 https://icfsp.org/

Publication series

Name	International Conference on Frontiers of Signal Processing
Publisher	Institute of Electrical and Electronics Engineers
ISSN (Print)	3066-1544
ISSN (Electronic)	3066-1587

Conference

Conference	The 10th International Conference on Frontiers of Signal Processing
Abbreviated title	ICFSP 2025
Country/Territory	France
City	Paris
Period	10/09/25 → 12/09/25
Internet address	https://icfsp.org/

Keywords / Materials (for Non-textual outputs)

computer architecture
digital signal processing
embedded systems
wireless radio

Access to Document

10.1109/ICFSP67350.2025.11353742

StrachanTophamICFSP2025ImprovingEfficiencyofHybridAccepted author manuscript, 293 KBLicence: Creative Commons: Attribution (CC-BY)

Cite this

Strachan, A., & Topham, N. (2026). Improving efficiency of hybrid flexible streaming accelerator systems via memory and interrupt optimizations. In 2025 10th International Conference on Frontiers of Signal Processing (ICFSP) (pp. 49-54). (International Conference on Frontiers of Signal Processing). Institute of Electrical and Electronics Engineers. https://doi.org/10.1109/ICFSP67350.2025.11353742

@inproceedings{08b98f9990e244229253d9debb106b49,

title = "Improving efficiency of hybrid flexible streaming accelerator systems via memory and interrupt optimizations",

abstract = "Hybrid accelerator systems have the potential to increase the useful lifespan of hardware designs. These systems allow parts of a hardware accelerator to be replaced with software running on a general purpose CPU. This allows the design to be extended or fixed in the field, after the fact, compared to traditional fixed function hardware accelerators.However, when implemented naively, such systems have significant overheads due to cache and interrupt management, limited memory bandwidth, and single threaded operation. This paper details three mechanisms that can reduce these overheads, and evaluates their impact using full-system simulation of an ARM-based embedded multicore architecture, extended with a cycle-accurate hardware accelerator. We show that efficiency can be increased by 5-20\%, leading to decreased power and energy consumption, extending the potential applications of such systems.",

keywords = "computer architecture, digital signal processing, embedded systems, wireless radio",

author = "Alexander Strachan and Nigel Topham",

year = "2026",

month = jan,

day = "29",

doi = "10.1109/ICFSP67350.2025.11353742",

language = "English",

isbn = "9798331574437",

series = "International Conference on Frontiers of Signal Processing",

publisher = "Institute of Electrical and Electronics Engineers",

pages = "49--54",

booktitle = "2025 10th International Conference on Frontiers of Signal Processing (ICFSP)",

address = "United States",

note = "The 10th International Conference on Frontiers of Signal Processing, ICFSP 2025 ; Conference date: 10-09-2025 Through 12-09-2025",

url = "https://icfsp.org/",

}

Strachan, A & Topham, N 2026, Improving efficiency of hybrid flexible streaming accelerator systems via memory and interrupt optimizations. in 2025 10th International Conference on Frontiers of Signal Processing (ICFSP). International Conference on Frontiers of Signal Processing, Institute of Electrical and Electronics Engineers, pp. 49-54, The 10th International Conference on Frontiers of Signal Processing, Paris, France, 10/09/25. https://doi.org/10.1109/ICFSP67350.2025.11353742

Improving efficiency of hybrid flexible streaming accelerator systems via memory and interrupt optimizations. / Strachan, Alexander; Topham, Nigel.
2025 10th International Conference on Frontiers of Signal Processing (ICFSP). Institute of Electrical and Electronics Engineers, 2026. p. 49-54 (International Conference on Frontiers of Signal Processing).

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

TY - GEN

T1 - Improving efficiency of hybrid flexible streaming accelerator systems via memory and interrupt optimizations

AU - Strachan, Alexander

AU - Topham, Nigel

N1 - Conference code: 10

PY - 2026/1/29

Y1 - 2026/1/29

N2 - Hybrid accelerator systems have the potential to increase the useful lifespan of hardware designs. These systems allow parts of a hardware accelerator to be replaced with software running on a general purpose CPU. This allows the design to be extended or fixed in the field, after the fact, compared to traditional fixed function hardware accelerators.However, when implemented naively, such systems have significant overheads due to cache and interrupt management, limited memory bandwidth, and single threaded operation. This paper details three mechanisms that can reduce these overheads, and evaluates their impact using full-system simulation of an ARM-based embedded multicore architecture, extended with a cycle-accurate hardware accelerator. We show that efficiency can be increased by 5-20%, leading to decreased power and energy consumption, extending the potential applications of such systems.

AB - Hybrid accelerator systems have the potential to increase the useful lifespan of hardware designs. These systems allow parts of a hardware accelerator to be replaced with software running on a general purpose CPU. This allows the design to be extended or fixed in the field, after the fact, compared to traditional fixed function hardware accelerators.However, when implemented naively, such systems have significant overheads due to cache and interrupt management, limited memory bandwidth, and single threaded operation. This paper details three mechanisms that can reduce these overheads, and evaluates their impact using full-system simulation of an ARM-based embedded multicore architecture, extended with a cycle-accurate hardware accelerator. We show that efficiency can be increased by 5-20%, leading to decreased power and energy consumption, extending the potential applications of such systems.

KW - computer architecture

KW - digital signal processing

KW - embedded systems

KW - wireless radio

U2 - 10.1109/ICFSP67350.2025.11353742

DO - 10.1109/ICFSP67350.2025.11353742

M3 - Conference contribution

SN - 9798331574437

T3 - International Conference on Frontiers of Signal Processing

SP - 49

EP - 54

BT - 2025 10th International Conference on Frontiers of Signal Processing (ICFSP)

PB - Institute of Electrical and Electronics Engineers

T2 - The 10th International Conference on Frontiers of Signal Processing

Y2 - 10 September 2025 through 12 September 2025

ER -

Strachan A, Topham N. Improving efficiency of hybrid flexible streaming accelerator systems via memory and interrupt optimizations. In 2025 10th International Conference on Frontiers of Signal Processing (ICFSP). Institute of Electrical and Electronics Engineers. 2026. p. 49-54. (International Conference on Frontiers of Signal Processing). doi: 10.1109/ICFSP67350.2025.11353742

Improving efficiency of hybrid flexible streaming accelerator systems via memory and interrupt optimizations

Abstract

Publication series

Conference

Keywords / Materials (for Non-textual outputs)

Access to Document

Fingerprint

Cite this