2022 IEEE International Symposium on Performance Analysis of Systems and Software (ISPASS) 2022
DOI: 10.1109/ispass55109.2022.00017
|View full text |Cite
|
Sign up to set email alerts
|

FASE: A Fast, Accurate and Seamless Emulator for Custom Numerical Formats

Abstract: Deep Neural Networks (DNNs) have become ubiquitous in a wide range of application domains. Despite their success, training DNNs is an expensive task that has motivated the use of reduced numerical precision formats to improve performance and reduce power consumption. Emulation techniques are a good fit to understand the properties of new numerical formats on a particular workload. However, current SoA techniques are not able to perform these tasks quickly and accurately on a wide variety of workloads.We propos… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1

Citation Types

0
0
0

Year Published

2023
2023
2023
2023

Publication Types

Select...
1
1

Relationship

0
2

Authors

Journals

citations
Cited by 2 publications
(1 citation statement)
references
References 7 publications
0
0
0
Order By: Relevance
“…Prior tools and methodologies to emulate reduced precision formats cannot deliver a fast, accurate, and seamless experience when training DNN workloads. In this section, we propose FASE, an emulation tool for custom numerical formats [74]. FASE is: (i) accurate by leveraging DBT techniques to emulate formats at instruction operand level; (ii) fast as it enables emulation of unmodified applications on large input sets thanks to a set of optimizations that lower its overheads significantly; and (iii) seamless as it works on any application or DNN framework without any language, compiler or source access restrictions and the guarantee that the instrumented binary matches the original one.…”
Section: Discussionmentioning
confidence: 99%
“…Prior tools and methodologies to emulate reduced precision formats cannot deliver a fast, accurate, and seamless experience when training DNN workloads. In this section, we propose FASE, an emulation tool for custom numerical formats [74]. FASE is: (i) accurate by leveraging DBT techniques to emulate formats at instruction operand level; (ii) fast as it enables emulation of unmodified applications on large input sets thanks to a set of optimizations that lower its overheads significantly; and (iii) seamless as it works on any application or DNN framework without any language, compiler or source access restrictions and the guarantee that the instrumented binary matches the original one.…”
Section: Discussionmentioning
confidence: 99%