Is there a reason (speed or otherwise) that you don't see many fiadd instructions in floating point code?
Is it faster to use a fild/fadd pair instead of fiadd ?
Is it faster to use a fild/fadd pair instead of fiadd ?
I think the reason is that most people forget about them. ;)
1 ?: I use a lot of fi??? on my code since I calculate a lot of integer based numbers.
2 ?: I don't know about this, but using fiadd directly would save you 1 fld/fild but would compensate for the slow memory access.
Probably a test would clear any doubts... :)
2 ?: I don't know about this, but using fiadd directly would save you 1 fld/fild but would compensate for the slow memory access.
Probably a test would clear any doubts... :)