I have a machine using qnx rtp with a pci voodoo3 3000. The banshee
driver is being used, but I’m not getting any performance gains with
opengl programs that I’m writing. Today I did a quick comparison test.
I’m using a slightly modified copy of the tk.c file that was in the
mesa demos directory.
The test I did was to load an obj model of a truck (using glm.c from a
glut demo) that is about 27000 triangles. I had it spin, and found the
average frame rate over 100 frames. It was run in fullscreen mode.
I tried it with the banshee driver at 640x480x8 and 640x480x16, and with
the vesa driver at 640x480x8. The results at 8bit color were identical
for both drivers (about 4.7fps) and the 16bit banshee was slower (about
2.9).
Is there something specific that needs to be done when compiling
software to get 3d acceleration? Or maybe some other system setting?
Here are the compile commands I used:
Two things you’ll need with the current mesa/glide. You need to be in
fullscreen direct mode (tk.c shows how to do this, and it sounds like you are doing this already), and the app needs to be
running with super user permissions (since glide accesses hardware directly).
Running it while logged in as root, or making it setuid root should solve this.
I have a machine using qnx rtp with a pci voodoo3 3000. The banshee
driver is being used, but I’m not getting any performance gains with
opengl programs that I’m writing. Today I did a quick comparison test.
I’m using a slightly modified copy of the tk.c file that was in the
mesa demos directory.
The test I did was to load an obj model of a truck (using glm.c from a
glut demo) that is about 27000 triangles. I had it spin, and found the
average frame rate over 100 frames. It was run in fullscreen mode.
I tried it with the banshee driver at 640x480x8 and 640x480x16, and with
the vesa driver at 640x480x8. The results at 8bit color were identical
for both drivers (about 4.7fps) and the 16bit banshee was slower (about
2.9).
Is there something specific that needs to be done when compiling
software to get 3d acceleration? Or maybe some other system setting?
Here are the compile commands I used:
Two things you’ll need with the current mesa/glide. You need to be in
fullscreen direct mode (tk.c shows how to do this, and it sounds like you are doing this already), and the app needs to be
running with super user permissions (since glide accesses hardware directly).
Running it while logged in as root, or making it setuid root should solve this.
I have a machine using qnx rtp with a pci voodoo3 3000. The banshee
driver is being used, but I’m not getting any performance gains with
opengl programs that I’m writing. Today I did a quick comparison test.
I’m using a slightly modified copy of the tk.c file that was in the
mesa demos directory.
The test I did was to load an obj model of a truck (using glm.c from a
glut demo) that is about 27000 triangles. I had it spin, and found the
average frame rate over 100 frames. It was run in fullscreen mode.
I tried it with the banshee driver at 640x480x8 and 640x480x16, and with
the vesa driver at 640x480x8. The results at 8bit color were identical
for both drivers (about 4.7fps) and the 16bit banshee was slower (about
2.9).
Is there something specific that needs to be done when compiling
software to get 3d acceleration? Or maybe some other system setting?
Here are the compile commands I used:
I’m logged in as root and I’m in fullscreen direct mode, just like tk.c had it. I don’t know why I’m not getting any
promising results!
Is there any chance that someone could send me a makefile and source code of an opengl program that you’ve been able to run
with hardware acceleration (and maybe some frame rates that you’ve gotten)? Then I’d have a benchmark to compare my machine
to, and some code to compare to see if it’s something dumb that I’m doing. We really need hardware acceleration here in the
lab to reduce some of the cpu usage for this research project…
Two things you’ll need with the current mesa/glide. You need to be in
fullscreen direct mode (tk.c shows how to do this, and it sounds like you are doing this already), and the app needs to be
running with super user permissions (since glide accesses hardware directly).
Running it while logged in as root, or making it setuid root should solve this.
I have a machine using qnx rtp with a pci voodoo3 3000. The banshee
driver is being used, but I’m not getting any performance gains with
opengl programs that I’m writing. Today I did a quick comparison test.
I’m using a slightly modified copy of the tk.c file that was in the
mesa demos directory.
The test I did was to load an obj model of a truck (using glm.c from a
glut demo) that is about 27000 triangles. I had it spin, and found the
average frame rate over 100 frames. It was run in fullscreen mode.
I tried it with the banshee driver at 640x480x8 and 640x480x16, and with
the vesa driver at 640x480x8. The results at 8bit color were identical
for both drivers (about 4.7fps) and the 16bit banshee was slower (about
2.9).
Is there something specific that needs to be done when compiling
software to get 3d acceleration? Or maybe some other system setting?
Here are the compile commands I used: