Alain schrieb:

> I remenber a tread where it was said that "pow(x,2)" is faster than 
> "x*x". It may only be due to the fact that the variable is evaluated 
> only once, but it seems to be faster.

That may well be indeed. I never thought about it, but it seems 
plausible, as each reference to a variable requires a hashtable lookup, 
as well as checking whether the value is indeed a scalar value. With 
SSE2, a pow() computation may be able to beat that overhead.

Post a reply to this message