How can one make "double" two times longer (more accurate) in C++ in fast way?

do you mean quad. precision?

Yes. I mean quad. precision.

Sorry, I try and stay away from quad precision :P I heard mathmatica can do quad precision fairly easily.

Anyway, I did a web search and get tons of references to c++ "precision" libraries.

Here's a link to one which looks good to me

http://www.nersc.gov/~dhbailey/mpdist/mpdist.html

Cheers

I don't know about C++, but 'double precision' in modern programming languages (e.g. Visual Basic) implies 64-bit, which is known in assembly as a 'quad word' (4*16-bit). The older Fortran codes, such as Plot3D, only use 32-bit precision ('double word') for floating point numbers. So 'double double precision' in Fortran may be equivalent to 'double precision' in C++.

Many thanks.