An N-Way Group Association Architecture and Sparse Data Group Association Load Balancing Algorithm for Sparse CNN Accelerators

In recent years, ASIC CNN Accelerators have attracted great attention among researchers for the high performance and energy efficiency. Some former works utilize the sparsity of CNN networks to improve the performance and the energy efficiency. However, these methods bring tremendous overhead to the...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:2019 24th Asia and South Pacific Design Automation Conference (ASP-DAC) s. 1 - 6
Hlavní autori: Wang, Jingyu, Yuan, Zhe, Liu, Ruoyang, Yang, Huazhong, Liu, Yongpan
Médium: Konferenčný príspevok..
Jazyk:English
Vydavateľské údaje: ACM 21.01.2019
Predmet:
Asia
C++ languages
CNN accelerator
Codes
Design automation
Energy efficiency
group association
load balancing
Load management
optimization algorithm
Scheduler
Simulation
ISSN:2153-697X
On-line prístup:Získať plný text
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Popis
Shrnutí:In recent years, ASIC CNN Accelerators have attracted great attention among researchers for the high performance and energy efficiency. Some former works utilize the sparsity of CNN networks to improve the performance and the energy efficiency. However, these methods bring tremendous overhead to the output memory, and the performance suffers from the hash collision. This paper presents: 1) an N-Way Group Association Architecture to reduce the memory overhead for Sparse CNN Accelerators; 2) a Sparse Data Group Association Load Balancing Algorithm which is implemented by the Scheduler module in the architecture to reduce the collision rate and improve the performance. Compared with the state-of-art accelerator, this work achieves either 1) 1.74x performance with 50% memory overhead reduction in the 4-way associated design or 2) 1.91x performance without memory overhead reduction in the 2-way associated design, which is close to the theoretical performance limit (without collision).
ISSN:2153-697X
DOI:10.1145/3287624.3287626