Efficiently removing sparsity for high-throughput stream processing
File(s)
Accepted version
Author(s)
Papaphilippou, Philippos
Que, Zhiqiang
Luk, Wayne
Type
Conference Paper
Abstract
Big data analytics and machine learning are increasingly targeted by FPGAs due to their significant amount
of computing capabilities and internal parallelism. Different
programming models are used to distribute the workload to
the internals of the FPGAs at different granularities. While
the memory bandwidth has been steadily increasing, there are
some challenges in the way system-on-chips use this bandwidth.
One way system-on-chip architects exploit the increasing memory
bandwidth is by widening the datapath width. This is reflected
at various points in the system including the widening of
vector instructions. On FPGAs, many analytics accelerators are
memory-bound, and would benefit from making the most of
the available bandwidth. In this paper we present a scalable
and highly-efficient building block for building high-throughput
streaming accelerators, which removes sparsity on-the-fly without
backpressure.
of computing capabilities and internal parallelism. Different
programming models are used to distribute the workload to
the internals of the FPGAs at different granularities. While
the memory bandwidth has been steadily increasing, there are
some challenges in the way system-on-chips use this bandwidth.
One way system-on-chip architects exploit the increasing memory
bandwidth is by widening the datapath width. This is reflected
at various points in the system including the widening of
vector instructions. On FPGAs, many analytics accelerators are
memory-bound, and would benefit from making the most of
the available bandwidth. In this paper we present a scalable
and highly-efficient building block for building high-throughput
streaming accelerators, which removes sparsity on-the-fly without
backpressure.
Date Issued
2024-02-01
Date Acceptance
2023-10-19
Citation
IEEE International Conference on Field-Programmable Technology (FPT), 2024
ISBN
979-8-3503-5911-4
ISSN
2837-0449
Publisher
IEEE
Journal / Book Title
IEEE International Conference on Field-Programmable Technology (FPT)
Copyright Statement
Copyright © 2024 IEEE. This accepted manuscript is licensed under a Creative Commons Attribution 4.0 International License (https://creativecommons.org/licenses/by/4.0/).
License URL
Identifier
https://ieeexplore.ieee.org/abstract/document/10416142
Source
2023 International Conference on Field-Programmable Technology
Publication Status
Published
Start Date
2023-12-11
Finish Date
2023-12-14
Coverage Spatial
Yokohama, Japan
Date Publish Online
2024-02-01