Scale-out Systolic Arrays
ACM Transactions on Architecture and Code Optimization · 30.06.2023 · 20 (2) · 1-25
DOI: 10.1145/3572917 ↗Kurz gefasst
Fachartikel in der Zeitschrift ACM Transactions on Architecture and Code Optimization, 2023-03. Verfasst von Ahmet Caner Yüzügüler und 5 weiteren Autor:innen. Die Arbeit wurde bisher 28-mal zitiert (Crossref) und nennt 69 Literaturangaben.
Abstract (Originalsprache)
Multi-pod systolic arrays are emerging as the architecture of choice in DNN inference accelerators. Despite their potential, designing multi-pod systolic arrays to maximize effective throughput/Watt—i.e., throughput/Watt adjusted when accounting for array utilization—poses a unique set of challenges. In this work, we study three key pillars in multi-pod systolic array designs, namely array granularity, interconnect, and tiling. We identify optimal array granularity across workloads and show that state-of-the-art commercial accelerators use suboptimal array sizes for single-tenancy workloads. We, then evaluate the bandwidth/latency trade-offs in interconnects and show that Butterfly networks offer a scalable topology for accelerators with a large number of pods. Finally, we introduce a novel data tiling scheme with custom partition size to maximize utilization in optimally sized pods. We propose Scale-out Systolic Arrays , a multi-pod inference accelerator for both single- and multi-tenancy based on these three pillars. We show that SOSA exhibits scaling of up to 600 TeraOps/s in effective throughput for state-of-the-art DNN inference workloads, and outperforms state-of-the-art multi-pod accelerators by a factor of 1.5 ×. 1
Zitieren
APA
Yüzügüler, A. C., Sönmez, C., Drumond, M., Oh, Y., Falsafi, B., & Frossard, P. (2023). Scale-out Systolic Arrays. ACM Transactions on Architecture and Code Optimization, 20(2), 1-25. https://doi.org/10.1145/3572917
Vancouver
Yüzügüler AC, Sönmez C, Drumond M, Oh Y, Falsafi B, Frossard P. Scale-out Systolic Arrays. ACM Transactions on Architecture and Code Optimization. 2023;20(2):1-25. doi:10.1145/3572917
BibTeX
@article{Yzgler2023,
author = {Yüzügüler, Ahmet Caner and Sönmez, Canberk and Drumond, Mario and Oh, Yunho and Falsafi, Babak and Frossard, Pascal},
title = {Scale-out Systolic Arrays},
journal = {ACM Transactions on Architecture and Code Optimization},
year = {2023},
volume = {20},
number = {2},
pages = {1--25},
doi = {10.1145/3572917}
}
EUVIMED