igfuw / UWLCM

University of Warsaw Lagrangian Cloud Model
GNU General Public License v3.0
6 stars 13 forks source link

Double precision in CPU computations? #159

Open pdziekan opened 1 year ago

pdziekan commented 1 year ago

Currently all computations are done on floats, except for those that fail if done on floats. Doubles on GPUs give much poorer performance. Doubles on CPUs also slow down the simulation, but not that much.