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
We (@pgrete and I) are seeing a render failure with the slice filter for GPU runs with our Parthenon-based code. It renders fine on GPU without the slice filter, and it always works on host runs. The simulation snapshots on disk are identical whether running on host or GPU.
I haven't seen this issue with my AMReX code, but I haven't run that with the latest (0.9.0) version of Ascent, so I am wondering if this is a recently introduced bug.
It should be precisely cut through the center of the mesh. Since it's an octree mesh, the individual mesh blocks should be on either side of the plane without intersecting any interior volume, which might explain what's going on.
We (@pgrete and I) are seeing a render failure with the slice filter for GPU runs with our Parthenon-based code. It renders fine on GPU without the slice filter, and it always works on host runs. The simulation snapshots on disk are identical whether running on host or GPU.
I haven't seen this issue with my AMReX code, but I haven't run that with the latest (0.9.0) version of Ascent, so I am wondering if this is a recently introduced bug.
A reproducer is here: parthenon-hpc-lab/athenapk#41.
Failure (on GPU run):
Success (on host only run):
The text was updated successfully, but these errors were encountered: