OESF Portables Forum
Model Specific Forums => Sharp Zaurus => Zaurus - pdaXrom => Topic started by: iamasmith on January 23, 2005, 07:24:55 am
-
Hi,
I built Mesa with GLUT yesterday for PDAXROM and it seems to work reasonably well running at least some of the demos at more than a couple of FPS as long as they don't have any reflections or other serious environmental mapping requirements.
I'm debating the usefulness of releasing this and would like to find software that people would like to run but bear in mind it's not a good means of running things like Quake2 because there are far better ports that more effectively eliminate the need for OpenGL support.
I'm looking for OpenGL apps that are going to be useable in Zoomed mode particularly, that don't necessarily have a 3D component just a requirement for OpenGL and if I can find a couple I may then release an IPK and a developer bundle (with a static link SDL interface to give the OpenGL option back to SDL on SDL_Init).
Thinks like glHack may even run 640x480... it would be nice to have a graphical Nethack available wouldn't it!
I would particularly be interested in anything like adventure games that suitably use OpenGL for static scene render with no animation.
- Andy
-
Hey..
Sash has also build this yesterday - will update feeds today - we have glxgears running at approx 20 fps in 320x240 fullscreen.
Batallion is running okay too.
-
Oh, pipped at the post then.
If you are going to add it to the stable feed what about a rebuild of SDL so that it recognises it ?
A lot of apps use SDL to go full screen and then use OpenGL.
- Andy
-
Of course -its need to get Bzflag running :-) (Bzflag takes up all the little free time i have at work)...
-
Hello,
Was just wondering. How likely is it that something like blender (http://www.blender.org) could be eventually be ported to the zaurus? It's an opengl based 3d modeling package. Open source of course.
Harry.
-
I think blender may be pushing the envelope a little too far.
The scene rendering NEEDS hardware acceleration to do anything reasonable with blender and I think you would find the editor unusable in 640x480.
-
so, where are those compiled stuff to test?
c'mon...
-
ohhh... once again we have had a small delay - but try and look in feed :-)
-
I think blender may be pushing the envelope a little too far.
The scene rendering NEEDS hardware acceleration to do anything reasonable with blender and I think you would find the editor unusable in 640x480.
[div align=\"right\"][a href=\"index.php?act=findpost&pid=63712\"][{POST_SNAPBACK}][/a][/div]
I know that NaN managed to get blender 2.23(?) working on the iPaq under windows pocketpc 2002. They demonstrated it at SIGGRAPH running a game at about 4-5fps. Considering that my zaurus is about twice the speed of the demo system with a far better build environment (if you're using OZ or pdaXrom anyway) I dont believe it would be difficult to port nor would it be really slow and unusable. Although it wouldn't be ideal...
If I get time, I'll have a go compiling 2.4a from source. I'll have to try compiling python 2.4 as well first though.
-
How about KSudoku? I've been wanting to get around to making a build, but I just haven't had the time with school and work. It uses Mesa for 3D versions of the puzzle (I can't wait to try some of those out!). It does need some of the KDE libs, but if we can get a working package, my bus ride in the morning won't be nearly as boring...
-
iamasmith, youre runninh a c3k right, you might be able to get iwmmx up and running for opengl as the mmx instructions can just be converted to straght to iwmmx in most cases (backwards compatible) however if you found a softwaer implemntation of opengl ES you would see a large increse in performance due to it all bieng integer maths
as we all know floating point kills our zaurus', only problem is the game would have to bewritten for opengl ES support
that aside for the newer PDA's using mmx on the OS floating point routines and the opengl libary should see a large performance incresse, i dont like to throw numbers around that i havent tested myself but i could see up to a 4 times performance increse over non accelerated implementations