[ET Trac] #2890: CarpetX: crash/failed assertion in Z4c_RHS

Miren Radia trac-noreply at einsteintoolkit.org
Thu Oct 9 07:03:37 CDT 2025


#2890: CarpetX: crash/failed assertion in Z4c_RHS

 Reporter: Miren Radia
   Status: open
Milestone: 
  Version: development version
     Type: bug
 Priority: major
Component: CarpetX

Comment (by Miren Radia):

> The problem is likely that the grid is too small for the given number of GPUs.

I’m not sure about this. I only ran on a single GPU \(initially I started on 8 but realised that I could reproduce on just 1\) and also started with a more production-sized parameter file with the following grid configuration that reproduces the same problem:

```
$nlevels = 10

CarpetX::xmin = -512.0
CarpetX::ymin = -512.0
CarpetX::zmin = -512.0
CarpetX::xmax = 512.0
CarpetX::ymax = 512.0
CarpetX::zmax = 512.0

CarpetX::ncells_x = 128
CarpetX::ncells_y = 128
CarpetX::ncells_z = 128

CarpetX::max_num_levels = $nlevels
CarpetX::regrid_every = 128
CarpetX::regrid_error_threshold = 0.03125

BoxInBox::num_regions = 2
# Region 1
BoxInBox::shape_1 = "cube"
BoxInBox::num_levels_1 = $nlevels
BoxInBox::radius_1 = [256.0,128.0,64.0,42.535384,21.267692,10.633846,5.316923,2.658462,1.329231,0.664615]
BoxInBox::position_x_1 = 4.461538461538462

# Region 2
BoxInBox::shape_2 = "cube"
BoxInBox::num_levels_2 = $nlevels
BoxInBox::radius_2 = [256.0,128.0,64.0,34.264616,17.132308,8.566154,4.283077,2.141538,1.070769,0.535385]
BoxInBox::position_x_2 = -5.538461538461538
```

I just made the grid configuration smaller for the purposes of reporting in this ticket as it’s quicker to run.

--
Ticket URL: https://bitbucket.org/einsteintoolkit/tickets/issues/2890/carpetx-crash-failed-assertion-in-z4c_rhs
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.einsteintoolkit.org/pipermail/trac/attachments/20251009/fb6fbb12/attachment.htm>


More information about the Trac mailing list