Jochen Küpper wrote:
Twice as many floats vectorize per operation as doubles. Thus it goes
twice as fast. How much more evidence do you require?
Personally I do not need any -- I have to use double for precision
needs. Matrices are just to big and calculation paths too long.
Sometimes I wish I had a 16-byte-double Cray...
long double on x86-64 is 16 bytes, if you _really_ feel the need for it.
long double on IA32 is 12 bytes.
Gordan
_______________________________________________
Help-gsl mailing list
[email protected]
http://lists.gnu.org/mailman/listinfo/help-gsl
