testActivePair playing weird
Here is a strange one for you. I am completely baffled.
If I run on cosma with the GCC tool stack (swift/c5/gcc/intelmpi/5.1.3
) and configure with ./configure --disable-vec
, I can run the full test suite without any problems.
If I now do ./configure --disable-vec --with-chemistry=gear
, I crash in the testActivePair
and testPeriodicBC
. Adding the GEAR chemistry module adds more fields to the particles and adds more work in the interaction function but that extra work does not update any of the normal hydrodynamics fields.
I don't understand why that should make any difference to the code results. I hope this is not a symptom of something more vicious.