JOGL-demos ti

[shawnkendall]

The demos look great.
All built fine and worked great on my AMD/GeForce 4 Ti4200 Win2000.

Glad you guys put a shadow demo in there ;-)
Gives me some really get stuff to show.

Developer:
"Oh yeah, I've seen those nVidia demos"

Me:
"This is the Java OpenGL wrappers"

Developer:
"OMG, Hell froze over? Schweet!"
or
"What's Java?"
or
"JavaScript crashes I.E"

:-D

[shawnkendall]

Ahh, hey admin, can you remove the "ti" at the end of this forum.  Typo I guess  

[Ken Russell]

Shawn: thanks for the positive feedback. Glad they ran smoothly.

Games published by our own members! Check 'em out!

[Archimedes]

The demos are quite nice.

However many of them (or most? or all?) run with Nvidia cards only?

This is ... very strange for "Java" demos.

Before I knew that, I recently tried to show the demos (binary version downloaded) to someone whom I wanted to show that "Java and OpenGL" work fine... What I didn't know: he got an high end ATI card ... and has been confrontated with large Java error stacks. He then said: So you Java fan - ha!

Don't get me wrong please: it's better to have some demos compared to nothing. However platform independance is most important for me. That's why I like Java and OpenGL.

[gregorypierce]

I have thus far run Gears on OSX and it runs fast....very fast, and its not even optimized. On my Windows box with an FX 5200 the demos don't seem to run as fast as the ATI 7500 in my TiBook and many of the shader ones suffer from some serious tearing problems (running the latest release off the NVidia website).


There is a serious problem with what appears to be GC. I'm not getting any pauses so much as I'm getting slowdown from time to time. There should be a lot of memory allocation going on so I'm somewhat confused as to what the garbage collector is actually doing. Strange strange issue. Happens on both the Windows and OSX builds.

[Ken Russell]

Regarding the NVidia dependence of the demos: the reason they are mostly NVidia-dependent is that NVidia publishes the most source code for demonstrations. I would encourage you all to port the NV_register_combiners and NV_texture_shader portions of some of the demos to the analogous ARB-approved extensions. We did some of the work during the OS X ports (NV_vertex_program had to be translated to ARB_vertex_program because Apple's drivers don't support NV_vertex_program) but a significant chunk remains. This is on my own list of things to do but is not at the top.

Regarding slowdowns: try -verbose:gc and see if you get any output that correlates to the slowdowns. I doubt there will be any, however. The issues probably relate to thread scheduling. Other demonstrations than the Gears demo (which runs at hundreds of frames per second if sync-to-vertical-retrace is disabled) that run at ~70 FPS look very smooth.

[Archimedes]

Regarding the NVidia dependence of the demos: the reason they are mostly NVidia-dependent is that NVidia publishes the most source code for demonstrations. I would encourage you all to port the NV_register_combiners and NV_texture_shader portions of some of the demos to the analogous ARB-approved extensions.

So actually it's possible to make OpenGL-compatible versions of the demos? That's good news.

[gregorypierce]

Regarding slowdowns: try -verbose:gc and see if you get any output that correlates to the slowdowns. I doubt there will be any, however. The issues probably relate to thread scheduling. Other demonstrations than the Gears demo (which runs at hundreds of frames per second if sync-to-vertical-retrace is disabled) that run at ~70 FPS look very smooth.

Other than the fact that it seems to GC continuously?

... a lot of GC similar to the line below happening all the time ...
[GC 1913K->1405K(1984K), 0.0020791 secs]
[GC 1917K->1423K(1984K), 0.0026429 secs]
[GC 1935K->1399K(1984K), 0.0019733 secs]
[GC 1911K->1406K(1984K), 0.0021693 secs]
[GC 1918K->1399K(1984K), 0.0020983 secs]
[GC 1911K->1439K(1984K), 0.0031239 secs]
[GC 1951K->1398K(1984K), 0.0019469 secs]
[Full GC 1398K->427K(1984K), 0.2301223 secs]
[GC 939K->430K(1984K), 0.0032683 secs]
[GC 942K->429K(1984K), 0.0029342 secs]
[GC 941K->430K(1984K), 0.0019340 secs]
[GC 942K->442K(1984K), 0.0022335 secs]
[GC 954K->429K(1984K), 0.0021853 secs]
[GC 941K->429K(1984K), 0.0019332 secs]

Clearly at some point the OSX JVM cleans house something fierce

Anyways, so much for the idea that all Java applications consume a lot of memory   You could fit that demo on a PS2 for sure