Hello Chia-Hui, > In addition , I would like to ask question about computation > consumption. For the BNS merger (using .par file in the gallery > file), it costs about 5000 core hours (48 cores for 100 hours) for > the completed simulation (to iteration=15000). Is it reasonable? And > is there some way to save the computational resources ? If you want to save overall resources you could reduce the number of cores used (assuming you do not run out of memory). For example running on 32 cores will make it run slower but not by a factor of 48/32 so that the total resources (number-of-cores * hours-used) goes down.
You could also try reducing (slightly!) the resolution eg by a factor of 1.25 which would make the simulation cheaper (but also loose significantly in accuracy). Finally you could try and make the simulation domain smaller, which safes a bit but not a whole lot usually. You can also try and see if reducing the amount of output produced makes any difference (though that is unlikely). This is the outXXX_every parameters and generally everything with an _every in its name. Yours, Roland -- My email is as private as my paper mail. I therefore support encrypting and signing email messages. Get my PGP key from http://pgp.mit.edu .
pgpd34WQIQfWb.pgp
Description: OpenPGP digital signature
_______________________________________________ Users mailing list [email protected] http://lists.einsteintoolkit.org/mailman/listinfo/users
