There are two types, float and double.

Is it enough to use float for CFD calculation? And does it really save memory, especially speed of the calculation?

Thanks

Junseok

Depends on the problem. I've had success with float for steady-state runs, for time-dependent ones I would use double.

If I want to do stability analysis on the results, then again double...

It does save memory, but how much, that depends on the code, environment, problem size and so on... not really my specialty.

-- Jarmo