First post, by ih8registrations
Has there been any thought of optimizing the decode?
Like:
mov ax, cx
shr cx, 1
rep movsw
mov cx, ax
and cx, 1
rep movsb
or
mov ax, cx
shr cx, 1
rep movsw
and ax,1
jz itseven
movsb
itseven:
etc.
instead of rep movsb.