|
Work item:
|
F.AI-CPP
|
|
Subject/title:
|
Technical specification for artificial intelligence cloud platform: Performance
|
|
Status:
|
[Carried to next study period]
|
|
Approval process:
|
AAP
|
|
Type of work item:
|
Recommendation
|
|
Version:
|
New
|
|
Equivalent number:
|
-
|
|
Timing:
|
-
|
|
Liaison:
|
-
|
|
Supporting members:
|
-
|
|
Summary:
|
With the continuous development and integration of artificial intelligence and cloud computing, cloud service providers have launched a variety of artificial intelligence cloud platforms, integrating the underlying resources with the upper deep learning framework to provide users with training, inference and other public services. At present, there are many AI platforms in the market, but the performance of these platforms is uneven. Lacking of comprehensive and unified performance evaluation methods is not conducive to users to make the most appropriate choice.
This recommendation specifies the performance of AI cloud platform from three levels of hardware level, platform level and service level. It aims to objectively measure the product performance of industry participants, thus providing direction for manufacturers to analyse and optimize the performance of AI cloud platform, and helping users choose an AI cloud platform that meets the performance requirements at the lowest possible cost, so as to promote the standardization and rapid development of the industry.
|
|
Comment:
|
-
|
|
Reference(s):
|
|
|
Historic references:
|
|
Contact(s):
|
|
| ITU-T A.5 justification(s): |
|
|
|
|
First registration in the WP:
2021-05-21 10:59:53
|
|
Last update:
2022-02-15 14:11:33
|
