Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[NPU] Fix regression caused by layer_norm change #12627

Merged
merged 3 commits into from
Dec 27, 2024

Conversation

plusbang
Copy link
Contributor

@plusbang plusbang commented Dec 27, 2024

Description

Background: https://github.com/analytics-zoo/nano/issues/1798

Add a new env IPEX_LLM_NPU_DRIVER_VERSION:

  • For default setting, revert layernorm graph to fix perf regression.
  • To support special driver (such as 5716 driver), need to set IPEX_LLM_NPU_DRIVER_VERSION=5716 manually.

4. How to test?

  • Application test

@plusbang plusbang requested a review from rnwang04 December 27, 2024 05:51
Copy link
Contributor

@rnwang04 rnwang04 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

@plusbang plusbang merged commit 46eeab4 into intel-analytics:main Dec 27, 2024
1 check passed
@jason-dai
Copy link
Contributor

Change the value of IPEX_LLM_NPU_DRIVER_VERSION to the driver version, instead of 0 or 1

@plusbang
Copy link
Contributor Author

Change the value of IPEX_LLM_NPU_DRIVER_VERSION to the driver version, instead of 0 or 1

yeah, have already changed : )
For example, need to set IPEX_LLM_NPU_DRIVER_VERSION=5716 manually for 5716 driver support.

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.

3 participants