Windows+ATI star and nebula display bug solution.

Solas · Post by **Solas** » Sat Jan 31, 2009 1:28 am

Windows players whose ATI video cards display stars and nebulae as ugly translucent squares
there seems to be an OpenGL driver error in ATI's EXT_texture_lod_bias implementation.
this has not been fixed in Catalyst 9.1 drivers. ( edit .. ) or 9.2

edit: Ahruman has made a good code workaround for Oolite version 1.73

if you want to compile oolite.exe from source edit trunk/src/Core/Materials/OOTexture.m changing ..
if (sTextureLODBiasAvailable) glTexEnvf(GL_TEXTURE_FILTER_CONTROL_EXT, GL_TEXTURE_LOD_BIAS_EXT, _lodBias);
to
if (sTextureLODBiasAvailable) glTexEnvf(GL_TEXTURE_ENV, GL_TEXTURE_LOD_BIAS_EXT, _lodBias);
the resulting oolite.exe should work for Windows + ATI video cards. even with recent Catalyst drivers ( including 9.1 )
it is not necessary to update drivers first.

this apparent OpenGL driver error has been reported to ATI recently.
glTexEnvf(0x8500,0x8501,_lodBias); ( should work .. but does not )
glTexEnvf(0x2300,0x8501,_lodBias); ( should not work .. but does )
is there anybody out there with experience reporting graphics bugs to ATI ?

JensAyton · Post by **JensAyton** » Sat Jan 31, 2009 1:56 am

It should be noted that the change Solas made above has not been added to “official” Oolite because it goes against the EXT_texture_lod_bias specification.

For 1.73, I’ve added a setting to disable all use of EXT_texture_lod_bias, which should avoid the glitch. (See Hidden Settings in Oolite.)

Solas · Post by **Solas** » Fri Feb 27, 2009 1:32 am

currently there are only reports of this ATI Bug on Vista 32 and 64 bit.

Screet · Post by **Screet** » Fri Feb 27, 2009 2:13 am

Solas wrote:
currently there are only reports of this ATI Bug on Vista 32 and 64 bit.

Add XP X64 with a X800XTPE.

Screet

Baza · Post by **Baza** » Sun Mar 01, 2009 8:32 pm

Theres one thing I cannot understand..why is it that I get the nebula and point stars etc occassionally. Surely if this is a driver problem then I should get consistently problems? Cannot wait for 1.73 with the fix...any idea how long?

Solas · Post by **Solas** » Sun Mar 01, 2009 10:05 pm

that's strange, an inconsistent version of the bug, maybe that could help ATI.
did you have the same problem with your HD2600XT ?
are you using ATI or manufacturer driver and which version is it ?
are you using Vista64 with a 3850X2 currently ?

Post by **Kaks** » Sun Mar 01, 2009 10:19 pm

Baza wrote:
any idea how long?

We'll still have to do a few more bugfixes before releasing 1.73, which might take a while...

In the meantime, you can try and build your own version of Oolite from trunk. That way you can get to play with the latest developments!
The most up to date instructions to do so on a PC are in

https://bb.oolite.space/viewtopic.php?p=69755#69755

A word of warning: sometimes things that work ok in 1.72.2 may be temporarily broken in trunk...

Screet · Post by **Screet** » Sun Mar 01, 2009 10:25 pm

Baza wrote:
Theres one thing I cannot understand..why is it that I get the nebula and point stars etc occassionally. Surely if this is a driver problem then I should get consistently problems? Cannot wait for 1.73 with the fix...any idea how long?

I also have other failures which appear sometimes (including bugs appearing after some play time)...and that's definitely only happening with my Vista ATI configuration.

When I tested the modified version, I didn't experience the nebulae problems anymore with Vista...

Other issues had been solved with other driver versions, but those introduced other problems.

It's easy to implement bugs that don't happen everytime. For example, reading memory from an uninitialized variable. Good development tools can show such things, but often are not used because they slow things down while testing.

Screet

Baza · Post by **Baza** » Tue Mar 03, 2009 5:56 pm

Thanks Screet that explains things

Baza · Post by **Baza** » Tue Mar 03, 2009 7:23 pm

Enabled triple buffering in CCC and pink effect has gone!

Solas · Post by **Solas** » Wed Mar 11, 2009 2:20 am

after reporting this bug on AMD Game » AMD Technology » ATI Radeon Graphics
http://forums.amd.com/game/messageview. ... erthread=y

I have received a response indicating that the problem has been located and a solution found ..
"This has been implemented ready for a future driver and will probably take a few months to appear in the drivers that the public get."

Screet · Post by **Screet** » Wed Mar 11, 2009 7:42 am

Solas wrote:
I have received a response indicating that the problem has been located and a solution found ..
"This has been implemented ready for a future driver and will probably take a few months to appear in the drivers that the public get."

MONTHS

!?!

If they found and fixed it as they say, why not supply that with the next driver release? It sounds like they may have found something but are afraid that they messed things up?!?

I did write to the vendor and asked them to send me a bill for the price difference of a GTX 295 compared to the 4870x2...as my card definitely suffers from other problems, too. Same OpenGL speed as an X1550 which uses the mainboards RAM! There must be something seriously wrong. No wonder that I have been missing a performance kick when I tried to play other OpenGL games in the past. Asus support completely stopped replying and ATI tech support simply closed the case as "solved" without ever writing any suggestions on what might be wrong - after asking me for a system report which includes my Vista keys (which I did blank out). Oh...and the driver feedback form has simply been disabled!

Screet

JensAyton · Post by **JensAyton** » Wed Mar 11, 2009 1:38 pm

Screet wrote:
If they found and fixed it as they say, why not supply that with the next driver release? It sounds like they may have found something but are afraid that they messed things up?!?

Yes.

This is an entirely reasonable attitude when writing drivers.

Website · Post by **DaddyHoggy** » Wed Mar 11, 2009 1:42 pm

So my plan to stop using ATi cards ten years ago was the right one...

From my friends who still have ATi cards they say Customer support has gotten worse since they were taken over by AMD - is this true? Shocking if it is!

Post by **Kaks** » Wed Mar 11, 2009 4:48 pm

Ahruman wrote:
This is an entirely reasonable attitude when writing drivers.

Or any complex software!

Screet, I'm sorry you got burnt by ATI's broken drivers: however, as someone older & wiser told me many a time: It ain't worth it! walk away, man, walk away & don't look back, life is too short!

Hopefully the nvidia card will go some way to make you forget all this hassle...