From:	"Bart Alewijnse" <scarfman AT geocities DOT com>
Newsgroups:	comp.os.msdos.djgpp
Subject:	Screen copy
Date:	20 Jun 1999 22:10:37 GMT
Organization:	WorldOnline News server
Lines:	21
Message-ID:	<01bebb6a$4346e640$e6b2f1c3@scarfboy.tip.nl>
NNTP-Posting-Host:	vp178-230.worldonline.nl
X-Trace:	news.worldonline.nl 929916637 2986 195.241.178.230 (20 Jun 1999 22:10:37 GMT)
X-Complaints-To:	abuse AT worldonline DOT nl
NNTP-Posting-Date:	20 Jun 1999 22:10:37 GMT
X-Newsreader:	Microsoft Internet News 4.70.1155
To:	djgpp AT delorie DOT com
DJ-Gateway:	from newsgroup comp.os.msdos.djgpp
Reply-To:	djgpp AT delorie DOT com

Damn.. My last message seems to have lost itself somewhere
between my news client and the server... I don't have the time to
retype it at the moment, but the question was how to copy a
double buffer to the screen the fastest, this being the A0000h in
dos memory or a LFB in 'regular' memory. There's rep movsl-ing,
which is what a memcpy would inline to with a constant as its
'amount of dwords to move' input, (is this true? I only read it
somewhere..) so that pretty much settles it for a LFB,
but what's the fastest for 0xA0000? dosmemput seems to be
faster than disabling memory protection (nearptr hack) and
rep movsl-ing it there for some reason. (though direct blitting this
wild-side way seems way faster..)  Just name me the fastest way,
I don't care too much if that means the nearptr hack... It's just
since I can blit a (compiled) sprite at 45000 fps using the nearptr
hack, but using the double buffer cuts that to about 300.. Still no
probs I sepose, but who knows what code in my programs won't
be optimized, slong the program down to slide-show speed?

I'm yakking/ranting/whatevering... Never mind that part:)

-Bart

- Raw text -