You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
* Fix BTD/Scrape Flush Count with Filters
Move the counting of already flushed particles for writers that call the
I/O backends multiple time per data set, e.g., BTD and boundary scraping,
into the I/O backend.
Currently, filtering is done as the first step in I/O backends and thus
the previous count outside of the I/O backends was over-counting
particles that might still get filtered out.
Offset should be a `long`:
Overflow risk is very high for pure `int`.
Also, counter is `unsigned`, so `unsigned long` for now.
* Simplify: Remove `m_totalParticles_flushed_already`
Less state we can forget in checkpoint-restart and that we
have to transfer across API boundaries.
// Rescaling the box for plotfile after WriteToFile. This is because, for plotfiles, when writing particles, amrex checks if the particles are within the bounds defined by the box. However, in BTD, particles can be (at max) 1 cell outside the bounds of the geometry. So we keep a one-cell bigger box for plotfile when writing out the particle data and rescale after.
0 commit comments