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

proposal: testing: reuse the iteration number of the first round when count > 1 #38000

Open
shawndx opened this issue Mar 22, 2020 · 0 comments

Comments

@shawndx
Copy link
Contributor

shawndx commented Mar 22, 2020

We often run all benchmarks of std packages with -count=5 or higher, it usually takes 5-6 hours to finish on our x86 and arm64 servers, the overall turnaround time might be reduced a lot if we can skip the auto adjustment from round #2 and reuse the iteration number computed before.

With an experimental change I got the following numbers:

                                    old                                 new
package math/big            2865.780s                              2242.640s
package runtime              5936.460s                              4926.073s
package encodng/json     486.691s                                321.370s

I realized same attempt was made before but then dropped due to benchmark regressions reported in #25622, a change to computing the 'grain' value of RunParallel proposed in #37996 shall avoid the regression. Furthermore this feature could be guarded by an option if necessary.

@gopherbot gopherbot added this to the Proposal milestone Mar 22, 2020
@ianlancetaylor ianlancetaylor added this to Incoming in Proposals (old) Jan 6, 2021
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
Status: Incoming
Development

No branches or pull requests

2 participants