Is it possible to test NaN before writing the next *.dmp's ? If so, the time
used for that could be negligle with regard to the time lost to continue
simulations having NaN.
Error must be returned if so, to know that there is a problem. The *.dmp's
could then be used to continue the simulation and trace the problem, to correct
it.
Original issue reported on code.google.com by ccos...@gmail.com on 7 Apr 2014 at 9:05
Original issue reported on code.google.com by
ccos...@gmail.com
on 7 Apr 2014 at 9:05