From: Mark <ma...@3e...> - 2002-12-29 11:07:41
|
So i've been sitting on the sidelines waiting for a fix for the rendering bugs in the R200 driver, but it doesn't seem like anyone is tackling it. I'm running a Radeon Mobility 9000. RTCW is playable but with significant artifacts, half life has same. UT2003 looks totally insane, crazy polygons everywhere. Has anyone looked at this? If someone could maybe point me at a troubled section of code, I could work on it. I would like to get the RTCW and Half-Life/Wine artifacts fixed at least. If someone knows of a very simple program which generates these same errors please let me know, it will make it much easier to track down. Mark |
From: Keith W. <ke...@tu...> - 2002-12-29 12:05:56
|
Mark wrote: > So i've been sitting on the sidelines waiting for a fix for the rendering bugs > in the R200 driver, but it doesn't seem like anyone is tackling it. I'm > running a Radeon Mobility 9000. RTCW is playable but with significant > artifacts, half life has same. UT2003 looks totally insane, crazy polygons > everywhere. > > Has anyone looked at this? If someone could maybe point me at a troubled > section of code, I could work on it. I would like to get the RTCW and > Half-Life/Wine artifacts fixed at least. > > If someone knows of a very simple program which generates these same errors > please let me know, it will make it much easier to track down. That might be a good place to start -- try and find or write (perhaps by modifying the mesa demos) a piece of code that breaks the r200 driver. Keith |
From: magenta <ma...@tr...> - 2002-12-30 01:40:55
|
On Sun, Dec 29, 2002 at 12:07:40PM +0000, Keith Whitwell wrote: > Mark wrote: > > So i've been sitting on the sidelines waiting for a fix for the rendering bugs > > in the R200 driver, but it doesn't seem like anyone is tackling it. I'm > > running a Radeon Mobility 9000. RTCW is playable but with significant > > artifacts, half life has same. UT2003 looks totally insane, crazy polygons > > everywhere. > > > > Has anyone looked at this? If someone could maybe point me at a troubled > > section of code, I could work on it. I would like to get the RTCW and > > Half-Life/Wine artifacts fixed at least. > > > > If someone knows of a very simple program which generates these same errors > > please let me know, it will make it much easier to track down. > > That might be a good place to start -- try and find or write (perhaps by > modifying the mesa demos) a piece of code that breaks the r200 driver. I know that my engine Solace (http://www.cs.nmsu.edu/~joshagam/Solace/) causes such artifacts... my guess is that it happens when playing back a displaylist which was created using glArrayElement(). (I'm guessing that the screenshot someone posted was using the default configurations, which does that by default.) Also, if whoever it was who posted that screenshot could change the Solace setting of GL_draw_method to 0 (which switches to immediate mode rendering rather than using vertex arrays) and see if that continues with the funkiness, that would be another useful data point. :) -- http://trikuare.cx |
From: D. H. <dha...@dr...> - 2002-12-30 02:09:29
|
On Sun, 29 Dec 2002, magenta wrote: > On Sun, Dec 29, 2002 at 12:07:40PM +0000, Keith Whitwell wrote: > > Mark wrote: > > > So i've been sitting on the sidelines waiting for a fix for the rendering bugs > > > in the R200 driver, but it doesn't seem like anyone is tackling it. I'm > > > running a Radeon Mobility 9000. RTCW is playable but with significant > > > artifacts, half life has same. UT2003 looks totally insane, crazy polygons > > > everywhere. > > > > > > Has anyone looked at this? If someone could maybe point me at a troubled > > > section of code, I could work on it. I would like to get the RTCW and > > > Half-Life/Wine artifacts fixed at least. > > > > > > If someone knows of a very simple program which generates these same errors > > > please let me know, it will make it much easier to track down. > > > > That might be a good place to start -- try and find or write (perhaps by > > modifying the mesa demos) a piece of code that breaks the r200 driver. > > I know that my engine Solace (http://www.cs.nmsu.edu/~joshagam/Solace/) > causes such artifacts... my guess is that it happens when playing back a > displaylist which was created using glArrayElement(). (I'm guessing that > the screenshot someone posted was using the default configurations, which > does that by default.) I got a couple of flickers on the right side of the middle sphere using your default radeon registry file. When I used just the plain default registry file I got quite a few more flickers in the cartoonish sphere and the thingie on the right side. > Also, if whoever it was who posted that screenshot could change the Solace > setting of GL_draw_method to 0 (which switches to immediate mode rendering > rather than using vertex arrays) and see if that continues with the > funkiness, that would be another useful data point. :) Switching the GL_draw_method to 1 gave me ... from what I can tell perfect rendering ... well ... at least no flickers. -- //========================================================\\ || D. Hageman <dha...@dr...> || \\========================================================// |
From: magenta <ma...@tr...> - 2002-12-30 03:07:41
|
On Sun, Dec 29, 2002 at 08:06:58PM -0600, D. Hageman wrote: > > > > I know that my engine Solace (http://www.cs.nmsu.edu/~joshagam/Solace/) > > causes such artifacts... my guess is that it happens when playing back a > > displaylist which was created using glArrayElement(). (I'm guessing that > > the screenshot someone posted was using the default configurations, which > > does that by default.) > > I got a couple of flickers on the right side of the middle sphere using > your default radeon registry file. When I used just the plain default > registry file I got quite a few more flickers in the cartoonish sphere and > the thingie on the right side. Cartoonish sphere? Do you mean the torus group? There's only one sphere in the default shader-test scene. :) Also, what do you mean by flickers? Keep in mind that there's a (somewhat crappy) CLOD algorithm going on too, so the objects are going to naturally change shape as you change distance from them. > > Also, if whoever it was who posted that screenshot could change the Solace > > setting of GL_draw_method to 0 (which switches to immediate mode rendering > > rather than using vertex arrays) and see if that continues with the > > funkiness, that would be another useful data point. :) > > Switching the GL_draw_method to 1 gave me ... from what I can tell perfect > rendering ... well ... at least no flickers. Huh. BTW, typing "d_set" from the controlling terminal will give online help for all of the configuration options (and you can also change them with that same command; for example, d_set GL_draw_method 0, though that won't really have any effect on cached displaylists). -- http://trikuare.cx |
From: D. H. <dha...@dr...> - 2002-12-30 05:46:23
|
On Sun, 29 Dec 2002, magenta wrote: > On Sun, Dec 29, 2002 at 08:06:58PM -0600, D. Hageman wrote: > > > > > > I know that my engine Solace (http://www.cs.nmsu.edu/~joshagam/Solace/) > > > causes such artifacts... my guess is that it happens when playing back a > > > displaylist which was created using glArrayElement(). (I'm guessing that > > > the screenshot someone posted was using the default configurations, which > > > does that by default.) > > > > I got a couple of flickers on the right side of the middle sphere using > > your default radeon registry file. When I used just the plain default > > registry file I got quite a few more flickers in the cartoonish sphere and > > the thingie on the right side. > > Cartoonish sphere? Do you mean the torus group? There's only one sphere in > the default shader-test scene. :) Yes, that would be the one. If you take all the torus together it reminds me of a cartoonish framework for what could be overall a sphere. Imagine stretching a piece of cloth around the whole grouping ... > Also, what do you mean by flickers? It was like the image that was supposed to be clipped because it was hidden became visible briefly as the light went by. It just happens briefly and then it is quickly corrected. This is probably not an accurate description, but there it is. > > > Also, if whoever it was who posted that screenshot could change the Solace > > > setting of GL_draw_method to 0 (which switches to immediate mode rendering > > > rather than using vertex arrays) and see if that continues with the > > > funkiness, that would be another useful data point. :) > > > > Switching the GL_draw_method to 1 gave me ... from what I can tell perfect > > rendering ... well ... at least no flickers. > > Huh. Oops - mistype. Switching the GL_draw method to 0 correct all of the anomalies I was seeing in the display. -- //========================================================\\ || D. Hageman <dha...@dr...> || \\========================================================// |
From: D. H. <dha...@dr...> - 2002-12-30 02:12:33
|
I have RPMS available of the CVS version of the main XFree86 CVS tree merged with the CVS version of the DRI tree dated 20021224. If anyone is interested in testing with these, then I can make the available. No guarantee they will work for you. I only use the radeon stuff out of them, but all drivers are included. They are made from a modified version of Mike Harris's SPEC file. Let me know. -- //========================================================\\ || D. Hageman <dha...@dr...> || \\========================================================// |
From: magenta <ma...@tr...> - 2002-12-30 09:02:27
|
On Sun, Dec 29, 2002 at 11:43:59PM -0600, D. Hageman wrote: > On Sun, 29 Dec 2002, magenta wrote: > > > On Sun, Dec 29, 2002 at 08:06:58PM -0600, D. Hageman wrote: > > > > > > > > I know that my engine Solace (http://www.cs.nmsu.edu/~joshagam/Solace/) > > > > causes such artifacts... my guess is that it happens when playing back a > > > > displaylist which was created using glArrayElement(). (I'm guessing that > > > > the screenshot someone posted was using the default configurations, which > > > > does that by default.) > > > > > > I got a couple of flickers on the right side of the middle sphere using > > > your default radeon registry file. When I used just the plain default > > > registry file I got quite a few more flickers in the cartoonish sphere and > > > the thingie on the right side. > > > > Cartoonish sphere? Do you mean the torus group? There's only one sphere in > > the default shader-test scene. :) > > Yes, that would be the one. If you take all the torus together it reminds > me of a cartoonish framework for what could be overall a sphere. Imagine > stretching a piece of cloth around the whole grouping ... > > > Also, what do you mean by flickers? > > It was like the image that was supposed to be clipped because it was > hidden became visible briefly as the light went by. It just happens > briefly and then it is quickly corrected. This is probably not an > accurate description, but there it is. Okay... could you set the value of testRender/capt to some directory with lots of free space and press 'k'? It'll record PPM-format screenshots... then send the appropriate frame(s) to me, preferrably as PNG or JPG. :) I *think* I know what you're talking about (I think I've seen something similar on some builds of r100) - sometimes it skips over a few vertices while playing the displaylist and so the surface gets "streched" out - but I'd like to see it myself to be sure. A good scene for testing that particular failure mode is scenes/test6, because the main object in it has a shape which makes those errors quite obvious. :) > > > > Also, if whoever it was who posted that screenshot could change the Solace > > > > setting of GL_draw_method to 0 (which switches to immediate mode rendering > > > > rather than using vertex arrays) and see if that continues with the > > > > funkiness, that would be another useful data point. :) > > > > > > Switching the GL_draw_method to 1 gave me ... from what I can tell perfect > > > rendering ... well ... at least no flickers. > > > > Huh. > > Oops - mistype. Switching the GL_draw method to 0 correct all of the > anomalies I was seeing in the display. That makes more sense. :) Sounds like it's a problem with displaylists and vertex arrays. Does it still happen if you set the value of GL_dlistMax to 0? -- http://trikuare.cx |
From: D. H. <dha...@dr...> - 2002-12-30 15:40:42
|
I have put up the RPMS from my 20021224 build of the XFree86 CVS Head and the DRI CVS Head merged at the following location since I have had a few requests for them: http://people.eecs.ku.edu/~dhageman/XFree86/ Remember, please don't bug Mike Harris about these RPMS ... he had nothing to do with except the creation of the original SPEC file. Actually, don't bug me privately if they don't work. They are provided in case someone wants to easily try to test things out. You will still need to build the kernel module for your card from DRI CVS. -- //========================================================\\ || D. Hageman <dha...@dr...> || \\========================================================// |
From: D. H. <dha...@dr...> - 2002-12-29 18:19:16
|
On Sun, 29 Dec 2002, Mark wrote: > So i've been sitting on the sidelines waiting for a fix for the rendering bugs > in the R200 driver, but it doesn't seem like anyone is tackling it. I'm > running a Radeon Mobility 9000. RTCW is playable but with significant > artifacts, half life has same. UT2003 looks totally insane, crazy polygons > everywhere. Are you running a recent version (CVS version) of the XFree86/DRI code? I have a Radeon Mobility 9000 and purchased RTCW to test it with and only on occasion did I notice some texture problems. I usually found that shutting down the program and restarting it fixed the issues. Other then that it looks great! I can't comment on the UT2003 though. I have seen some screenshots which makes me believe that patch to convert the textures isn't so ... nice. > Has anyone looked at this? If someone could maybe point me at a troubled > section of code, I could work on it. I would like to get the RTCW and > Half-Life/Wine artifacts fixed at least. Seriously, what type of artifacts are you seeing with RTCW? Do you have a demo that I could run to reproduce these on my system? -- //========================================================\\ || D. Hageman <dha...@dr...> || \\========================================================// |
From: Daniel V. <vo...@ep...> - 2002-12-30 03:36:19
|
> I can't comment on the UT2003 though. I have seen some screenshots which > makes me believe that patch to convert the textures isn't so ... nice. Can't comment on Linux though on Mac and Windows at least the textures get decompressed correctly and the code is the same on all platforms. Keep in mind that the game can use insane amounts of texture memory with decompression enabled. -- Daniel, Epic Games Inc. |