Re: 3D texture size limitations in mesa/r200?

Hi;

This is the combination that is consistently reproducible.

1. startx
2. Render 256x256x64 (will not render right)
3. 512x512x32 (will render right)
4. 512x512x16 (will render right)
5. 512x512x32 (will not render right)

To be able to render 512x512x32 correctly, do step 2 again.

When I say "will not render right", it looks like remnants of previous data
are still seen (??)

Is this supposed to make any sense?

Thanks,
Aditya


On Jan 5, 2008 9:46 AM, R. Aditya Kadambi <[email protected]> wrote:

>
>
> On Jan 5, 2008 9:28 AM, R. Aditya Kadambi <[email protected]> wrote:
>
> >
> >
> > On Jan 5, 2008 8:35 AM, Roland Scheidegger <[email protected]>
> > wrote:
> >
> > > R. Aditya Kadambi wrote:
> > > > I get these debug messages at start on setting R200_DEBUG=tex. Same
> > > > messages for the successful rendering and unsuccessful rendering!
> > > >
> > > > r200AllocTexObj( 0x802bf2800, 0x802ef0800 )
> > > Hmm that's not much debug output. Guess you'd need a version built for
> > > debugging...
> > >
> > > > Thanks. Looks like the patch is for drivers/dri/radeon/*. I thought
> > > > the driver was drivers/dri/r200/*. Am I missing something here?
> > > No this patch is for the kernel module. It is against the git drm, but
> > > it should apply to the kernel drm as well probably (but obviously the
> > > file will be in a different location then).
> > >
> > > > In the meantime, I will put in the debug flags and report back. Also
> > > > I will try out the different sizes mentioned.
> > > If the problem is what I think it is, going past a certain texture
> > > height * depth is what will trigger the problem.
> > >
> > > Roland
> >
> >
> > Hi;
> >
> > I did a little bit more experimentation yesterday.
> >
> > I increase the texture size in the r200 driver from 256 to 512.
> >
> > (a)I was able to render a 512x512x32 succesfully. Then I tried
> > 256x256x64. It rendered successfully.
> >
> > (b) I restarted X and then rendered 256x256x256. Only "half" volume
> > rendered successfully.
> >
> > I tried (a) again and it was a success. That sequence always works.
> >
> > (c) I tried 512x512x32. Successful. Then tried 512x512x16. Successful.
> > Then tried 256x256x64. Successful.
> >
> > I will experiment a bit more. It almost looks like graphics memory is
> > not "cleaned up". But it is "cleaned up" when I go from 512x512x* to
> > lower  volumes.
> >
> > I will report back on the sequences.
> >
> > Thanks,
> > Aditya
> >
>
> Minor correction to the post. Point (b) should be "I restarted X and then
> rendered 256x256x64 (or 32). Only "half" volume rendered successfully"
>
> Thanks,
> Aditya
>

-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace

--
_______________________________________________
Dri-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/dri-users