XPM performance
Dave Craig
Resent-Date: Tue, 21 Oct 1997 20:00:56 -0400 (EDT)
Date: Tue, 21 Oct 1997 12:07:20 -0700
Message-ID: <62iugf$847$1...@news.aimnet.com>
Sender: xpert-...@opengroup.org
Resent-Message-ID: <"IVoL3vE0fPF.A.wUC.upTT0"@postman.opengroup.org>
X-Mailing-List: xpert:archive/latest/6300
X-Loop: xp...@opengroup.org
Precedence: list
Resent-Sender: xpert-...@opengroup.org
I'm a bit of a newbie to all things X and I have a question regarding the
performance of X in regards to pixmaps using XPM. When running some
performace tests displaying 2 color bitmaps I get around 40 million pixels
per second performance out of my system (VxWorks RTOS with RtX-Windows on
top). When I switch to 256 color pixmaps the performance drops to around 5
million pixels per second. Now there is a corellation here of going from 1
bit to 8 bit color causing a 8 times performance degradation, but is that
the reason for the performance decrease? I would of thought that because the
window color depth is still 8 bit then all 8 bit planes still have to be
written to?
Has anyone any explanations as to what is the most significant pixmapping
performance issue ?
Dave Craig.
Manfred Bartz
Resent-Date: Wed, 22 Oct 1997 02:02:24 -0400 (EDT)
Date: 22 Oct 1997 13:12:03 +1000
Message-ID: <m27mb6o...@reddwarf.aup.com.au>
Sender: xpert-...@opengroup.org
References: <62iugf$847$1...@news.aimnet.com>
Resent-Message-ID: <"3mbqK_-BUjD.A.xhF.I7YT0"@postman.opengroup.org>
X-Mailing-List: xpert:archive/latest/6304
"Dave Craig" <dcr...@euphonix.com> writes:
> I'm a bit of a newbie to all things X and I have a question regarding the
> performance of X in regards to pixmaps using XPM. When running some
> performace tests displaying 2 color bitmaps I get around 40 million pixels
> per second performance out of my system (VxWorks RTOS with RtX-Windows on
> top). When I switch to 256 color pixmaps the performance drops to around 5
> million pixels per second. Now there is a corellation here of going from 1
> bit to 8 bit color causing a 8 times performance degradation, but is that
> the reason for the performance decrease? I would of thought that because the
> window color depth is still 8 bit then all 8 bit planes still have to be
> written to?
>
> Has anyone any explanations as to what is the most significant pixmapping
> performance issue ?
My theory is: Black and white are always defined and don't need a
color table lookup. With 8 bit color (and 16 bit color) there is an
additional table lookup which would explain a performance drop.
24 and 32 bit color don't need lookups and should be faster, but then
you write more plains. Performance will depend a lot on the hardware
accellerations that may be available depending on your card and the
server.
Regards
--
Manfred Bartz <mba...@werple.net.au>
Dave Craig
Resent-Date: Wed, 22 Oct 1997 23:38:22 -0400 (EDT)
Date: Wed, 22 Oct 1997 16:24:25 -0700
Message-ID: <62m1u4$5ms$1...@news.aimnet.com>
Sender: xpert-...@opengroup.org
References: <62iugf$847$1...@news.aimnet.com>
Resent-Message-ID: <"tVYMt9FlYUK.A.-cH.u0rT0"@postman.opengroup.org>
X-Mailing-List: xpert:archive/latest/6332
For information....
The main problem I had seemed to be that I was loading the XPM format into
an XImage and not directly into a Pixmap. So instead of using XCopyArea, I
was using XPutImage which seems to be about 5 times slower as it has to do
the Image to Pixmap conversion each time it is called
Dave