Open tbhaxor opened 1 year ago
Hi @tbhaxor it looks like you're initializing the data on the hos
// initializing on host
auto u = u_dev.GetHostMirrorAndCopy();
but do not copy the data back to the device.
Either you update the initialization to be done on the device (i.e., using a par_for
) or you add a deep copy after the initialization loop to copy the data from the host to the device.
I recommend the former as the "init on host" is a leftover from refactoring Athena++ code and we should clean that up at some point.
Hi @pgrete
Could you guide me into direction with some snippets?
I recommend the former..
So I am using auto u = u_dev.GetDeviceMirror();
, but still same issue
Here's a reference to "init on host" with the deep copy back to the device at the end https://github.com/parthenon-hpc-lab/athenapk/blob/8f04e18f75f4e655eff8e289759d06af87f4fbd9/src/pgen/blast.cpp#L202 and here's one with the initialization on the device https://github.com/parthenon-hpc-lab/athenapk/blob/8f04e18f75f4e655eff8e289759d06af87f4fbd9/src/pgen/orszag_tang.cpp#L40
I tried both of them, neither of them are working in my case. Could it be some problem with problem setup?
Could you please open a PR with the exact (full) code that you tried, i.e., including the modification to main.cpp
etc? That' might help in narrowing down the issue.
Hi @pgrete, thanks for waiting. Here is the PR https://github.com/parthenon-hpc-lab/athenapk/pull/70
I am facing problem with output of the Rayleigh-Taylor problem.
C++ Code
Input file
Contents of the history file are changing but neither paraview, nor movie2d script is showing the evolution of fluid.
It created 10 image files, and all of them has same plot