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

add wait() to make code stable #5895

Merged
merged 1 commit into from Mar 6, 2024

Conversation

NeoZhangJianyu
Copy link
Collaborator

@NeoZhangJianyu NeoZhangJianyu commented Mar 6, 2024

  1. add wait() to make code stable.
  2. use fp32 on oneMKL gemm_batch for better performance.
  3. add debug function.

Current performance reference:

GPU: 1 Arc 770
OS: ubuntu 22.04
Param: -mg 0 -sm none
Model: llama-2-7b.Q4_0.gguf

Avg: 30.66 tokens per second

Copy link
Collaborator

@airMeng airMeng left a comment

Choose a reason for hiding this comment

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

better leave some performance data for future reference

@NeoZhangJianyu
Copy link
Collaborator Author

better leave some performance data for future reference

Yes, update it in first comment.

@airMeng
Copy link
Collaborator

airMeng commented Mar 6, 2024

better leave some performance data for future reference

Yes, update it in first comment.

comparisons before and after?

@NeoZhangJianyu NeoZhangJianyu merged commit 8ced9f7 into ggerganov:master Mar 6, 2024
60 checks passed
hazelnutcloud pushed a commit to hazelnutcloud/llama.cpp that referenced this pull request Mar 10, 2024
NeoZhangJianyu added a commit to NeoZhangJianyu/llama.cpp that referenced this pull request Mar 12, 2024
jordankanter pushed a commit to jordankanter/llama.cpp that referenced this pull request Mar 13, 2024
hodlen pushed a commit to hodlen/llama.cpp that referenced this pull request Apr 1, 2024
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.

None yet

2 participants