|
Work item:
|
F.748.18 (ex F.AI-DLEMT)
|
|
Subject/title:
|
Metric and evaluation methods for AI-enabled multimedia application computing power benchmark
|
|
Status:
|
Approved on 2022-12-14 [Issued from previous study period]
|
|
Approval process:
|
AAP
|
|
Type of work item:
|
Recommendation
|
|
Version:
|
New
|
|
Equivalent number:
|
-
|
|
Timing:
|
-
|
|
Liaison:
|
-
|
|
Supporting members:
|
-
|
|
Summary:
|
Facing more and more diverse AI computing systems, users hope to have a unified evaluation metric for the system that provides AI computing power. The establishment of relevant real application performance evaluation benchmarks can objectively reflect the current state of the AI computing ability by providing objective metrics and comparison dimensions.
This Recommendation provides an AI computing power benchmark framework, evaluation metrics and methods, and a guideline for technical testing for AI clusters.
|
|
Comment:
|
-
|
|
Reference(s):
|
|
|
Historic references:
|
|
Contact(s):
|
|
| ITU-T A.5 justification(s): |
|
|
|
|
First registration in the WP:
2021-11-12 09:33:30
|
|
Last update:
2022-11-09 10:18:26
|
