A1 Refereed original research article in a scientific journal
Accelerating Image Processing Using Reduced Precision Calculation Convolution Engines
Authors: Pokhrel Narayan, Snäll Sakari, Heimo Olli I, Sarwar Uruj, Airola Antti, Säntti Tero
Publisher: Springer New York LLC
Publication year: 2023
Journal: Journal of Signal Processing Systems
Journal name in source: JOURNAL OF SIGNAL PROCESSING SYSTEMS FOR SIGNAL IMAGE AND VIDEO TECHNOLOGY
Journal acronym: J SIGNAL PROCESS SYS
Number of pages: 12
ISSN: 1939-8018
eISSN: 1939-8115
DOI: https://doi.org/10.1007/s11265-023-01869-5
Web address : https://doi.org/10.1007/s11265-023-01869-5
Self-archived copy’s web address: https://research.utu.fi/converis/portal/detail/Publication/179722677
In this paper a method of accelerating image processing using convolution engines with reduced precision calculation is presented. The convolution engines are designed to be used with the Pulpissimo platform with RISC-V System-on-Chip. The aim is to move the calculation to the edge. The proposed linear convolution engines operate on 8-bit data set and the logarithmic convolution engine operates on 4-bit reduced precision data. The data reduction is done by using a logarithmic number space. Diminishing the size of the data to be processed reduces the amount of required memory, requirement for memory bandwidth, required computation, and required hardware area while simultaneously increasing the performance. This performance could benefit modern AI and image processing applications, especially in mobile and other battery-operated devices. The results show that the computation in the linear convolution engine is 91 times faster and computation in the logarithmic convolution engine is 122 times faster than in the RISC-V core with plain RISC-V instructions.
Downloadable publication This is an electronic reprint of the original article. |  |
