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

Read dump parallel #1179

Closed
wants to merge 4 commits into
base: master
from

Conversation

Projects
None yet
3 participants
@sjplimp
Copy link
Contributor

sjplimp commented Oct 27, 2018

Purpose

Enable the read_dump command to read and process parallel dump files, for native format.
Parallel dump files means a single snapshot is spread across N files. This can now also
be used with the rerun command.

Author(s)

Steve

Backward Compatibility

For serial dump files, the behavior of the read_dump and rerun comands should not change.
One change was made to prevent a rerun from halting if a new snapshot has an
earlier (or same) timestep as the last one. Now rerun will simply skip such timesteps
until a larger timestep is found. This allows use of a a series of files where the last snapshot in
one and the first snapshot in the next are duplicates.

Implementation Notes

Provide any relevant details about how the changes are implemented, how correctness was verified, how other features - if any - in LAMMPS are affected

Post Submission Checklist

Please check the fields below as they are completed

  • The feature or features in this pull request is complete
  • Suitable new documentation files and/or updates to the existing docs are included
  • One or more example input decks are included
  • The source code follows the LAMMPS formatting guidelines

Further Information, Files, and Links

Put any additional information here, attach relevant text or image files, and URLs to external sites (e.g. DOIs or webpages)

@sjplimp

This comment has been minimized.

Copy link
Contributor

sjplimp commented Oct 27, 2018

@akohlmey We should wait on merging until Alex Stukowski has a chance to try it out on a huge use case. That should be a good stress test.

@akohlmey

This comment has been minimized.

Copy link
Member

akohlmey commented Oct 27, 2018

@sjplimp i have added the work_in_progress label to indicated that this must not be merged yet. please remove if the code has been tested successfully.

@akohlmey

This comment has been minimized.

Copy link
Member

akohlmey commented Oct 31, 2018

@sjplimp any news on this?

@stukowski

This comment has been minimized.

Copy link

stukowski commented Nov 21, 2018

Hi,
I've tested this addition and it seems to work as expected. Reading of parallel dump files appears to be much faster than reading a serial dump file of the same size. I've tested the rerun command too with a series of parallel dump files and it worked as expected in general. However, when I tried to apply my new analysis fix, which computes the atomic displacements with respect to the first timstep, I noticed a different issue. This issue doesn't seem to be related to this extension of the read_dump command since it also occurs when reading serial dump files with the original read_dump implementation. That's why I have created a separate bug report for it, please see issue #1218.

@sjplimp

This comment has been minimized.

Copy link
Contributor

sjplimp commented Nov 29, 2018

@stukowski I made the small change for the add keep atom count bug-fix. I think you are saying
this is otherwise ready-to-go. I.e. it works fine on your large-scale test. @akohlmey after the stable, this could be fully regression tested. The rerun2 and read_dump2 (old versions) should be removed before merging.

@sjplimp sjplimp assigned akohlmey and unassigned sjplimp Nov 29, 2018

@akohlmey akohlmey referenced this pull request Dec 3, 2018

Merged

Collected post stable patches #1242

4 of 4 tasks complete
@akohlmey

This comment has been minimized.

Copy link
Member

akohlmey commented Dec 29, 2018

@sjplimp FYI, this has been merged into PR #1242 and merge conflicts resolved.
This can be merged now into master via approving #1242 and no more work on this branch is required (this PR will be automatically closed as soon as the other #1242 is merged).

@akohlmey akohlmey deleted the read-dump-parallel branch Jan 8, 2019

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment