|
Work item:
|
L.DLEE
|
|
Subject/title:
|
Deep Learning Computation Energy Efficiency Evaluation Framework and Metrics
|
|
Status:
|
[Carried to next study period]
|
|
Approval process:
|
AAP
|
|
Type of work item:
|
Recommendation
|
|
Version:
|
New
|
|
Equivalent number:
|
-
|
|
Timing:
|
-
|
|
Liaison:
|
ISO/IEC JTC1/SC42, SG16
|
|
Supporting members:
|
China, China Telecom, State Grid Corporation of China (China), Korea (Republic of)
|
|
Summary:
|
The increasing energy demands of deep learning computing pose significant challenges to energy sustainability, cost-effectiveness, and the long-term evolution of AI technologies. To effectively address this issue, it is imperative to adopt a holistic approach aimed at optimizing energy efficiency across all stages of deep learning computing. Addressing the computing efficiency of deep learning computing requires the development of standardized metrics and evaluation methods to assess energy consumption across diverse models and tasks. By doing so, we can better manage energy resources, optimize computing efficiency, and ensure the sustainable evolution of AI technologies.
|
|
Comment:
|
-
|
|
Reference(s):
|
|
|
Historic references:
|
|
Contact(s):
|
|
| ITU-T A.5 justification(s): |
|
|
|
|
First registration in the WP:
2024-07-03 16:30:54
|
|
Last update:
2024-07-04 16:04:53
|
