X viewer speed

Quentin Stafford-Fraser quentin "at" orl.co.uk
Sat, 07 Mar 1998 16:17:12 +0000


Hi Cedric,

I'm a little surprised.  On the same machine under the same conditions I
would have expected the X viewer to be a little faster, because it writes
straight to the screen where the Windows one writes into a memory bitmap
which it blits onto the screen.  But there are too many other factors here
to do a real comparison.  As you say, the video drivers are different, and X
generally has more layers between the application and the screen, the
network drivers and memory usage will be different, and I'm assuming your
screen has the same resolution, number of colours and pixel format in each
case?

There's probably more room to speed up the Windows viewer than the X one,
but there may be more room for speeding up your Linux environment?
Quentin
----
Dr Quentin Stafford-Fraser
The Olivetti & Oracle Research Lab
http://www.orl.co.uk/~qsf


-----Original Message-----
From: Cedric Maion <Cedric.Maion "at" insa-lyon.fr>
To: vnc-list "at" orl.co.uk <vnc-list "at" orl.co.uk>
Date: 06 March 1998 23:38
Subject: X viewer speed


>Hello
>
>I'm playing for the first time with vnc, and I find it very very
>interesting. Thanks a lot for your good job (and free software),
>ORL folks.
>
>What I wanted to know was if you already optimized vnc* for speed
>(especially the X viewer).
>I compared on the same box the X and the Win32 viewer, and the speed
>difference is quite impressive (Windows 95 OSR2 against XFree86 3.3.1/
>Red Hat Linux 5.0, using an ATI XPert "at" Play).
>The Windows viewer is clearly far quicker than the X one on my box...
>My Windows video card driver is probably one of the cause, but I'm not
>sure it explains such a difference.
>Do you have hints, or informations regarding the X viewer ?
>
>I also compared the speed when connecting from a Linux box to another Linux
>box, first using xdm and then vnc. I understand that because X is an higher
>level protocol than vnc, it should be quicker. Using a pure X solution
gives
>indeed far better results :-) (over a LAN).
>
>Regards,
>
> Cedric Maion
>--
>Cedric Maion <Cedric.Maion "at" insa-lyon.fr>
>http://www.maion.ml.org/
>
>