The new math routines in script.cpp have direct
analogues in 3dnow & SSE, at least for single-precision
(which should be more than good enough). Should be
outlined clearly enough in the AMD & Intel whitepapers.
I'll do this myself as a short-n-sweet way to get back up
to speed on x86 ASM, once we get some architectural
issues talked out on the forum...
Logged In: YES
user_id=495068
Forgot to add -- once these are up to spec, they'll probably be
added to the core files (eg internal.h) so that any/all other
code can get their benefits for "free."
Logged In: YES
user_id=495068
Assigned to myself; I've started work
Logged In: YES
user_id=495068
I've decided this is dumb -- the new script primitives get
executed at most once per frame, and asm support outside
the AMD camp is weak.
Logged In: YES
user_id=495068
I've decided this is dumb -- the new script primitives get
executed at most once per frame, and asm support outside
the AMD camp is weak.