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

Enhancement possibility to compile with -flto and -fprofile-generate for LTO and PGO compilation at kernel compiling level. #1523

Closed
osevan opened this issue Sep 21, 2021 · 1 comment

Comments

@osevan
Copy link

osevan commented Sep 21, 2021

PGO trained .gcda could boost tempestakernel maybe 30 % after profiles generated,because tempesta fw do frequently repeated operations, and are ideal for PGO (2 times compilations),to reduce context switch and cache misses,when training are done with real world benchmarks like wrk and ddos situations.

Thanks and

Best Regards

@krizhanovsky
Copy link
Contributor

We have played with PGO #1219 . The most performance crucial part is the HTTP parser and while PGO explores hot and cold code paths, it does not account code dependency. You can find more details on the subject in my SCALE x17 talk https://tempesta-tech.com/research/http_str.pdf , slide 27 is about PGO.

We ended up with manual compiler guidance on the proper code layout.

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

No branches or pull requests

2 participants