Rennie Allen <RAllen@csical.com> wrote:
As if hosting X applications isn’t inherently fat (and slow) enough, you
fat doesn’t necessarily mean slow, see my comments below.
propose an additional layer of translation as a satisfactory solution ?
VNC is very efficient. The VNC server is very “small” and runs as
a daemon. We have a server with more than ten copies running, serving
as a X application server for MS Windows clients. This is similiar
to the MS Windows Terminal Server, serving Unix/Citrix clients.
If this “additional layer” gives better performance and better support,
I would say this is a better solution.
To verify my suspicion, I did some performance tests today. “xbench”
is the program I used. The results are
XFree4.1 on RTP is a bit slower than on Linux, but close enough (within 10%).
This is probably due to the fact that there is no UDS support in RTP.
I have trouble getting Nexware’s VNC photon client to work (maybe they
can release the source, and other Photon experts can improve it) so
I ran my VNC X client for the test. I ran XFree4.1 as main GUI, one VNC
server in background, and then start X vnc client to talk to it and
run the xbench in this vnc window. The result is, for simple graphic
drawings, the speed is comparable to the XFree4.1 itself, but for
complex ones, it is getting slower. I know Xphoton is slower, but
am surprised to find the xbench in Xphoton is average 100X slower
than VNC tests, with some 10X slower, and other 200-300X slower.
I had to kill the xbench test in Xphoton when it goes to the complex
graphics, watching the slow drawing on the screen is a pain.
I can send you the test result files if anyone is interested, or
you can do the test yourself. It will be intesting if someone
can port xbench to Photon so that we can compare the result of
XFree4.1 and Photon itself. My guess is X will be faster. Photon
is designed to have a small footprint (that’s why you can run it
on a single floppy while this is impossible for X). I will
definitely recommend Photon if your application is for small/embedded.
On a desktop system, X will be a better choice: you will have more
readily available applications and it is IMO faster, maybe better
hardware support.
BTW, my test was on my IBM Thinkpad T21 laptop, Savage graphics
card, configured for 16bit color.
Frank
-----Original Message-----
From: <> fliu@bb.vipstage.com> > [mailto:> fliu@bb.vipstage.com> ]
Posted At: Thursday, October 18, 2001 9:29 PM
Posted To: advocacy
Conversation: XPhoton OpenSrc anytime soon?
Subject: Re: XPhoton OpenSrc anytime soon?
QSSL should concentrate their limited resources on phinx I think.
I don’t see any reasons to have Xphoton. You can achieve Xphoton
by running a VNC server (> http://www.sourceforge.net/projects/openqnx> )
and a Photon based VNC viewer talking to the VNC server locally.
http://www.nexwarecorp.com/Developers/repository.html#VNC
All the remaining Xphoton issues should probably be gone then.
thoughts?
Igor Kovalenko <> kovalenko@home.com> > wrote:
Yes Armin, new Xphoton source is basically a patch to Xfree 4.1 CVS.
There are still some issues to be resolved though…
- igor
“Armin” <> a-steinhoff@web.de> > wrote in message
news:> 3BCBFB07.8070109@web.de> …
Igor Kovalenko wrote:
I am not speaking for QNX, but I happen to know that they are working
inboth directions (improving code and preparing it to open source). All
good things come to those who wait > 
Wow … XPhoton compatible with 4.1 … that would be great!
Armin
If they don’t, we will have to find better alternatives