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

Batches with threads introduce roundoff error, which can effect adaptive timestep #9

Open
slundqui opened this issue Aug 5, 2015 · 2 comments
Labels

Comments

@slundqui
Copy link
Contributor

slundqui commented Aug 5, 2015

Various testing with batches shows that there exists a 1e-3 error in timeScaleTrue between a run with no batches versus a run greater than 8 batches running on the same images. As the timeScaleTrue converges, this roundoff error can trigger the timeScale reset (due to timeScaleTrue decreasing), making the 2 runs that should be identical diverge. Keep this in mind when running with batches, as you may need to change your dtChangeMin accordingly.

Sheng

@dpaiton dpaiton added the bug label Aug 20, 2015
@dpaiton
Copy link
Contributor

dpaiton commented Aug 20, 2015

Marked as a bug. Although maybe just posting a warning to the user is a sufficient fix for now.

@slundqui slundqui added wontfix and removed bug labels Aug 20, 2015
@slundqui
Copy link
Contributor Author

Marked as wontfix, as roundoff errors is not a bug, but a property of being parallel.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

No branches or pull requests

2 participants