#1795: PITTNullCode results depend on number of processors ------------------------------------+--------------------------------------- Reporter: rhaas | Owner: Type: defect | Status: confirmed Priority: major | Milestone: Component: EinsteinToolkit thorn | Version: development version Resolution: | Keywords: PITTNullCode ------------------------------------+---------------------------------------
Comment (by rhaas):
I used the parfile nad there are still differences. To try and see if the coordinates or the angular spacing were to blame, I added HDF5 output for "NullGrid::StCrd" and "NullGrid::NullGrParArrR" and used the attached python script to recombine the 2 processor data into a single chunk. There is no difference in either coordinates nor the angular grid spacing.
Are there other coordinates to consider? Eg the complex ones (though if the real ones agree I am not sure why the compex ones should not).