From: Ludvig Larsson <ludvig AT club-internet DOT fr>
Newsgroups: comp.os.msdos.djgpp
Subject: Re: Fastest Method of Virtual Buffer->Video Memory?
Date: Sat, 24 Oct 1998 04:11:43 +0200
Organization: Faas-Goldhart
Lines: 21
Message-ID: <3631375F.2328@club-internet.fr>
References: <3631349E DOT B70621ED AT geocities DOT com>
NNTP-Posting-Host: toulouse-camichel8-72.club-internet.fr
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Trace: front4.grolier.fr 909196719 5138 195.36.147.72 (24 Oct 1998 02:38:39 GMT)
NNTP-Posting-Date: 24 Oct 1998 02:38:39 GMT
X-Mailer: Mozilla 3.01C-CLUB  (Win95; I)
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Reply-To: djgpp AT delorie DOT com

Consider using movedata or _movedatal, AFAIK, with optimisations turned
on(-O2) they will be translated to a 
REP
STOSL
wich moves 4 bytes a time(if you want to move a number of bytes not
dividable by 4, use movedata, it will arrange things).
Also, copy from aligned a mempos, malloc returns aligned memory(adress
dividable by 4 I think), so don't do this:
buff=malloc(.....
move(buf+1,nr_of_bytes...

It will slow things down.

Otherwise, the transferspeed is due to memory speed and video-memory
speed, as a normal screen-buffer won't enter the cache(and won't be
there when you start to transfer probably anyway), the processorspeed is
not that important.

Hth

Ludvig Larsson