[ODE] ODE Digest, Vol 49, Issue 31

[Jon Watte (ODE)]

Oleh Derevenko wrote:
> Why do you think operations with float are faster than operations with 
> double? This is correct for emulations only. In FPU everything is converted 
> to 80-bit and there is no difference in source float format you load numbers 
>   

While that is true on x86, other CPUs (such as the PPCs found in all the 
current consoles) are faster on floats. Also, even the x86 runs faster 
if you turn the precision mode down to 23-bit mantissa. Last, storing 
something in double means that it's twice the size of something in 
float, which means twice the cache pollution, and thus an overall slower 
program.

Cheers,

             / h+