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.

Sign up for GitHub

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

added a workaround for issue 4604 #4631

Closed
wants to merge 1 commit into from
Closed

added a workaround for issue 4604 #4631

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions Source/Diagnostics/WarpXOpenPMD.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -709,6 +709,7 @@ WarpXOpenPMDPlot::DumpToFile (ParticleContainer* pc,
}

if (is_last_flush_to_step) {
SetupPos(currSpecies, NewParticleVectorSize, isBTD); // workaround for issue 4604
SetConstParticleRecordsEDPIC(currSpecies, NewParticleVectorSize, charge, mass);
}
Comment on lines 711 to 714
Copy link
Member

@ax3l ax3l Jan 24, 2024 •
edited
Loading

Choose a reason for hiding this comment

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

I am not sure this should go exactly like this here.

What do you think about this?

Suggested change
if (is_last_flush_to_step) {
SetupPos(currSpecies, NewParticleVectorSize, isBTD); // workaround for issue 4604
SetConstParticleRecordsEDPIC(currSpecies, NewParticleVectorSize, charge, mass);
}
if (is_last_flush_and_never_particles) {
SetupPos(currSpecies, NewParticleVectorSize, isBTD);
}
if (is_last_flush_to_step) {
SetConstParticleRecordsEDPIC(currSpecies, NewParticleVectorSize, charge, mass);
}

Copy link
Member

Choose a reason for hiding this comment

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

Admittedly, this is exactly what is done a few lines earlier, isn't it? Why is that not called, I wonder.

Copy link
Member

Choose a reason for hiding this comment

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

What are the values of num_dump_particles and ParticleFlushOffset for this problem?

Copy link
Contributor Author

Choose a reason for hiding this comment

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

num_dump_particles=0
ParticleFlushOffset=1000

so is_last_flush_and_never_particles (= is_last_flush_to_step && num_dump_particles == 0 && ParticleFlushOffset == 0) will be false.

Copy link
Member

@ax3l ax3l Jan 29, 2024 •
edited
Loading

Choose a reason for hiding this comment

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

Thanks, got it.

If ParticleFlushOffset > 0 that means at some point, at least one rank had particles to write.

We need to check if maybe in the call where we write particles (earlier), we have a logic mistake that forgets to call SetupPos collectively.

Copy link
Member

@ax3l ax3l Jan 29, 2024 •
edited
Loading

Choose a reason for hiding this comment

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

Hm, maybe we filter twice and forget num_fraction in one case? Let's take a look where num_fraction is applied and if the counter is done correctly afterwards.

Copy link
Contributor Author

Choose a reason for hiding this comment

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

My understanding is that the offsets are set in the BTD object, and when it calls flush(), the filters are then applied in the underlying flush formats during copyParticles(). Unlike other filters, the random filter affected the offsets of the caller BTD object.
I think the workaround is the easy and harmless. Otherwise, have to collect the particles written from all processes and update the offset in BTD.

Copy link
Member

@ax3l ax3l Jan 29, 2024 •
edited
Loading

Choose a reason for hiding this comment

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

I think the workaround is the easy and harmless.

I would rather like to fix the underlying issue in the BTD offset counter. We need consistent meta data.

cc @RevathiJambunathan :)

Copy link
Contributor Author

Choose a reason for hiding this comment

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

If it makes sense, can apply the random filter before BTD calls flush(). It will make things cleaning and easier.

Copy link
Member

Choose a reason for hiding this comment

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

Yes, we probably need to add this to:

  • full diags
  • BTD
  • scraping


Expand Down
Loading