nV News Forums

 
 

nV News Forums (http://www.nvnews.net/vbulletin/index.php)
-   NVIDIA Linux (http://www.nvnews.net/vbulletin/forumdisplay.php?f=14)
-   -   ut2k4 with EXT_framebuffer_object support (http://www.nvnews.net/vbulletin/showthread.php?t=54465)

skirst 08-01-05 06:27 PM

ut2k4 with EXT_framebuffer_object support
 
http://icculus.org/cgi-bin/finger/fi...l?user=icculus

Just got done fiddling with it, looks cool as hell :) Real shadows, hellbender license plates, you name it.

Spyke 08-01-05 07:23 PM

Re: ut2k4 with EXT_framebuffer_object support
 
- Realistic player shadows (or even blob shadows) won't render on terrain
on the Nvidia drivers, because of their 4 TMU limit for fixed-function
pipeline apps. There is no fix planned for this at this time. Go buy
an ATI card if it bothers you, and mail the receipt to Nvidia's
engineering department.

nVidia are you listening????????????

Spyke 08-01-05 08:23 PM

Re: ut2k4 with EXT_framebuffer_object support
 
Anyways for some reason it doesnt work for me, and yes I've added the stuff to the configs.
I get a black license plate and black scoreboards in morpheus, and shadows don't show.

Oh well.

Compholio 08-02-05 02:10 PM

Re: ut2k4 with EXT_framebuffer_object support
 
ignore me

evilghost 08-03-05 09:09 PM

Re: ut2k4 with EXT_framebuffer_object support
 
Very weird results. I installed it, followed the config directions.

When I'm in the Hellbender: Black License Plate
Bot in Hellbender: I see the name on the license plate.
I'm in HellBender, another one in sight (Bots only): NNNNPPPHRR on license plate.

Well, backing that patch off. Not that the license plate thing really mattered.

And as far as the "buy an ATI card", that might be a viable option for people who don't want decent Linux performance. I dumped my ATI Gaydeon 9800 Pro for an NVIDIA 6800GT because of grossly poor performance and the shoddy Linux support/allocation of resources.

LubosD 08-04-05 12:22 PM

Re: ut2k4 with EXT_framebuffer_object support
 
We must start terrorizing nVidia otherwise they won't give us those TMUs.
Another thing which would be great if fixed is Software Suspend 2 support.

Bugzilla would be great - all bugs and their fix progess could be easily tracked.
This forum is great for chatting but horrible for bug tracking.

--

I'd like to test ut2004 with EXT_framebuffer_object too, but I had a HDD crash. I'm currently running Knoppix and I am chrooted in a recovered (recompiled) Gentoo on a FireWire external HDD -> only nv driver. :thumbdwn:

LubosD 08-05-05 06:36 PM

Re: ut2k4 with EXT_framebuffer_object support
 
Quote:

Originally Posted by Spyke
Anyways for some reason it doesnt work for me, and yes I've added the stuff to the configs.
I get a black license plate and black scoreboards in morpheus, and shadows don't show.

I've managed to make nvidia driver work on Knoppix & chroot.

Scoreboard and license plate work like a charm (on my system). But I haven't seen any shadow (yet).

d4rk74m4 08-06-05 01:34 PM

Re: ut2k4 with EXT_framebuffer_object support
 
I have a 64-bit ut2004 installed currently so I haven't been able to test this, I'll just wait for an official patch :p

Question about the TMU problem: Does nvidia expose more than 4 TMU's in fixed-function D3D? If not, it's possible that the hardware doesn't actually support it. I have no idea though. On recent ATI cards (>=9550), fixed-function GL is actually done with vertex and fragment programs, so the full TMU set would be available.

Spyke 08-06-05 09:20 PM

Re: ut2k4 with EXT_framebuffer_object support
 
UT2004 is actually a newer game that for some reason still uses the older fixed function GL multitexturing stuff. NVIDIA wont change the number of pipelines seen in the drivers because they blame this on the application itself. Yes, UT2004 should use vertex and fragment programs, but it doesn't, it was never programmed this way. The cards actually have 16 TMU's as they are able to render 16 textures per rendering pass on most models. I still hope NVIDIA will allow the drivers someday to use 6 TMU's instead of 4, so UT2004 can at least render properly, as doing this wont break anything. If NVIDIA engineers played UT2004 in Linux they would of fixed this long ago as they would care about the rendering problems but apparently they do not.

Spyke 08-06-05 10:14 PM

Re: ut2k4 with EXT_framebuffer_object support
 
http://www.spykes.net/spike/rttnowork1.jpg
http://www.spykes.net/spike/rttnowork2.jpg

Spyke 08-06-05 10:23 PM

Re: ut2k4 with EXT_framebuffer_object support
 
ScriptLog: START MATCH
Log: OpenGL: Failed to create render target!
Log: OpenGL: Failed to create render target!
Log: OpenGL: Failed to create render target!
Log: OpenGL: Failed to create render target!
Log: OpenGL: Failed to create render target!
Log: OpenGL: Failed to create render target!
Log: OpenGL: Failed to create render target!
Log: OpenGL: Failed to create render target!

d4rk74m4 08-07-05 03:37 AM

Re: ut2k4 with EXT_framebuffer_object support
 
Quote:

Originally Posted by Spyke
UT2004 is actually a newer game that for some reason still uses the older fixed function GL multitexturing stuff. NVIDIA wont change the number of pipelines seen in the drivers because they blame this on the application itself. Yes, UT2004 should use vertex and fragment programs, but it doesn't, it was never programmed this way. The cards actually have 16 TMU's as they are able to render 16 textures per rendering pass on most models. I still hope NVIDIA will allow the drivers someday to use 6 TMU's instead of 4, so UT2004 can at least render properly, as doing this wont break anything. If NVIDIA engineers played UT2004 in Linux they would of fixed this long ago as they would care about the rendering problems but apparently they do not.

Sorry, I wasn't too clear about what I meant. Recent ATI cards don't have a fixed-function unit (afaik), OpenGL fixed-function is emulated using vertex/fragment programs. Thus, ATI is able to provide 8 TMU's for OGL fixed-function.

Perhaps, NVIDIA hardware has both fixed-function and programmable units. And the fixed-function unit is unable to access more than 4 TMU's. Hence the question about whether or not D3D fixed-function supports >4 TMU's.

Anyhow, it's just speculation.. Perhaps nvidia just doesn't see any point in extending the TMU count for fixed-function. I don't entirely blame them though, as shaders are much better way of doing things.


All times are GMT -5. The time now is 02:20 AM.

Powered by vBulletin® Version 3.7.1
Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Copyright 1998 - 2014, nV News.