Skip to content

Conversation

@ChenyuZhu1
Copy link
Contributor

@ChenyuZhu1 ChenyuZhu1 commented Oct 29, 2025

The continuation work of PR #285 and PR #308 .

Results obtained on H20 device:
8b48a560-50e4-41c5-a47a-89a8d1ca277b
Compared with those obtained on H20 device without using scatter gather interface, the load/dump bandwidth increases by more than 80% (from about 25GB/s to about 45GB/s).
Results obtained on Bao De device:
[Coming soon]

@ChenyuZhu1 ChenyuZhu1 changed the title Develop klukowski scatter gather [feat] Call scatter gather interface in dramstore Oct 30, 2025
@ChenyuZhu1 ChenyuZhu1 changed the title [feat] Call scatter gather interface in dramstore [Feat] Call scatter gather interface in dramstore Oct 30, 2025
@mag1c-h mag1c-h self-requested a review October 31, 2025 02:19
@mag1c-h mag1c-h merged commit 1827635 into ModelEngine-Group:develop Oct 31, 2025
3 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants