You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I am thinking of removing all the long double functions everywhere and bring everything to just use doubles. It would be a rather major change internally. But I think it would make the code quite a bit faster since the compiler can use the intrinsics better. Also, I find it extremely confusing to have some functions that want std::vector<double> and others that want std::vector<long double>.
What do you guys think?
The text was updated successfully, but these errors were encountered:
I know this is going to be an issue with VLE calculations with water due to
the complexity of its EOS. I think a bit of testing would be appropriate
before committing this change.
I think this is going to happen. I made the necessary changes in the CoolPropDbl_to_double branch. There's a single macro that changes this behavior, and a single typedef in python land. The only problems I have seen in the tests are a few composition derivatives not quite getting to tolerance. Overall this seems like a safe change.
@jowr @mikekaganski ,
I am thinking of removing all the long double functions everywhere and bring everything to just use doubles. It would be a rather major change internally. But I think it would make the code quite a bit faster since the compiler can use the intrinsics better. Also, I find it extremely confusing to have some functions that want
std::vector<double>
and others that wantstd::vector<long double>
.What do you guys think?
The text was updated successfully, but these errors were encountered: