Horde3D

Next-Generation Graphics Engine
It is currently 25.11.2024, 08:30

All times are UTC + 1 hour




Post new topic Reply to topic  [ 14 posts ] 
Author Message
 Post subject: Horde3D and MacOsx10.5.3
PostPosted: 12.06.2008, 16:25 
Offline

Joined: 12.06.2008, 16:07
Posts: 1
Hello everyone, and thanks for all the good work going on :)

I'm currently doing some tests on Horde3d under macosx (with a
macbook pro with Intel proc and NVidia gpu). I came across the
fact that with the new release of Leopard, a strange bug occurs on
initialization. This bug was not present with the 10.5.2 version.

I re-compiled the SVN version of Horde but nothing changed (I first
suspected some new Carbon side effects)

With gdb, here is the backtrace of the bug (which makes me think of
a driver bug, but I'm not sure of that) :

Program received signal EXC_BAD_ACCESS, Could not access memory.
Reason: KERN_INVALID_ADDRESS at address: 0x17b1f040
0x93ae155e in glgConvertTo_32<GLGConverter_RGB8_ARGB8, (GLGMemory)2> ()
(gdb) bt
#0 0x93ae155e in glgConvertTo_32<GLGConverter_RGB8_ARGB8, (GLGMemory)2> ()
#1 0x93adeeb3 in glgProcessPixelsWithProcessor ()
#2 0x1425882a in gldAllocVertexBuffer ()
#3 0x142595ce in gldGetTextureLevel ()
#4 0x0053bd14 in gleGenMipmaps ()
#5 0x004147c1 in glTexImage2D_Exec ()
#6 0x95ed59cb in glTexImage2D ()
#7 0x0016ce31 in RendererBase::uploadTextureCube ()
#8 0x00193330 in TextureCubeResource::initializationFunc ()
#9 0x0018153e in ResourceManager::registerType ()
#10 0x0014ae1e in init ()
#11 0x0000591b in Application::init ()
#12 0x00008008 in main ()


Any hints or workarounds with this ?

Croquette


Top
 Profile  
Reply with quote  
PostPosted: 12.06.2008, 23:00 
Offline

Joined: 22.11.2007, 17:05
Posts: 707
Location: Boston, MA
I am on a lousy connection, so I haven't been able to get the 400MB 10.5.3 update installed yet. I will be state-side next week, and will be able to try it then. Chances are though, that it is an NVidia specific problem, so hopefully Vurlix will be able to give it a spin as well (I have an ATI card).

_________________
Tristam MacDonald - [swiftcoding]


Top
 Profile  
Reply with quote  
PostPosted: 28.06.2008, 18:56 
Offline

Joined: 19.03.2008, 01:22
Posts: 79
I'm having the exact same issue. I work around this by disabling cubemap mipmapping or cubemaps altogether, but it's not a good solution. This is definately a driver bug -- I've seen similar reports on other projects.


Top
 Profile  
Reply with quote  
PostPosted: 30.06.2008, 04:20 
Offline

Joined: 19.03.2008, 01:22
Posts: 79
Time for me to figure out why Horde3D 0.14.0 doesn't trigger any issues and 0.15.0 craps out.


Top
 Profile  
Reply with quote  
PostPosted: 30.06.2008, 13:20 
Offline

Joined: 22.11.2007, 17:05
Posts: 707
Location: Boston, MA
Vurlix wrote:
Time for me to figure out why Horde3D 0.14.0 doesn't trigger any issues and 0.15.0 craps out.
So this only affects the 1.0b and 0.15 releases? That could be because the samples that came with 0.14 don't use cubemaps for anything.

_________________
Tristam MacDonald - [swiftcoding]


Top
 Profile  
Reply with quote  
PostPosted: 30.06.2008, 14:55 
Offline

Joined: 19.03.2008, 01:22
Posts: 79
Even 0.14.0 seems to work only randomly, after closer inspection. Nothing I can do about it, apparently. It runs fine ONCE from Xcode. After the sample is closed, it doesn't work anymore if I try to relaunch it. But it does seem to work fine when launched from Finder multiple times. Sigh.


Top
 Profile  
Reply with quote  
PostPosted: 30.06.2008, 16:24 
Offline

Joined: 22.11.2007, 17:05
Posts: 707
Location: Boston, MA
Vurlix wrote:
Even 0.14.0 seems to work only randomly, after closer inspection. Nothing I can do about it, apparently. It runs fine ONCE from Xcode. After the sample is closed, it doesn't work anymore if I try to relaunch it. But it does seem to work fine when launched from Finder multiple times. Sigh.
Nasty. Do they all run correctly if you disable mipmaps?

_________________
Tristam MacDonald - [swiftcoding]


Top
 Profile  
Reply with quote  
PostPosted: 30.06.2008, 19:16 
Offline

Joined: 19.03.2008, 01:22
Posts: 79
It appears so. It's very difficult to test -- it crashes completely at random, so it's hard to tell if it works because of disabled mipmaps. But it seems to work everytime they are disabled. So far.

Specifically, disabling mipmaps in the TextureCube resource registration seems to do the trick.

egTextures.cpp (Horde 0.14.0, but similar in 1.0*)
Code:
void TextureCubeResource::initResType()
{
   unsigned char defaultTexture[] =
      {   255,192,128, 255,192,128, 255,192,128, 255,192,128,
         255,192,128, 255,192,128, 255,192,128, 255,192,128,
         255,192,128, 255,192,128, 255,192,128, 255,192,128,
         255,192,128, 255,192,128, 255,192,128, 255,192,128   };

   // Upload default texture
   defTexObject = Modules::renderer().uploadTextureCube(
      defaultTexture, 4, 4, 3, false, 0, false, false, true );
   for( uint32 j = 1; j < 6; j++ )
   {
      Modules::renderer().uploadTextureCube(
         defaultTexture, 4, 4, 3, false, j, false, false, true, defTexObject );
   }
}
The boolean mipmap parameter is set to false here, avoiding the crash. I don't know if it's worth making this the standard behavior since the driver still crashes for other reasons... I expect all of these issues to clear up when they release an updated driver.


Top
 Profile  
Reply with quote  
PostPosted: 30.06.2008, 20:55 
Offline

Joined: 18.05.2008, 17:47
Posts: 96
is it a known issue(known issues list in the ati driver's release notes)


Top
 Profile  
Reply with quote  
PostPosted: 30.06.2008, 22:13 
Offline

Joined: 22.11.2007, 17:05
Posts: 707
Location: Boston, MA
kal wrote:
is it a known issue(known issues list in the ati driver's release notes)
This is an NVidia issue, ATI cards appear to be unaffected (my X1600 works fine, at least). As far as I know NVidia doesn't officially recognise the problem, but it is reported elsewhere.

_________________
Tristam MacDonald - [swiftcoding]


Top
 Profile  
Reply with quote  
PostPosted: 30.06.2008, 23:06 
Offline

Joined: 18.05.2008, 17:47
Posts: 96
nvidia... good point :oops:
for x1600 viewtopic.php?p=1907#p1907


Top
 Profile  
Reply with quote  
PostPosted: 01.07.2008, 14:25 
Offline

Joined: 22.11.2007, 17:05
Posts: 707
Location: Boston, MA
kal wrote:
nvidia... good point :oops:
for x1600 viewtopic.php?p=1907#p1907
Doesn't really apply - neither ATI nor NVidia seem to use the same code for Mac and Windows drivers, partly because there is no DirectX on the Mac, but also because Apple writes a large part of the driver. So far any bugs seem to be for one platform or the other, not both.

_________________
Tristam MacDonald - [swiftcoding]


Top
 Profile  
Reply with quote  
PostPosted: 01.07.2008, 15:47 
Offline

Joined: 19.03.2008, 01:22
Posts: 79
10.5.4 doesn't solve anything -- no graphics driver update included.


Top
 Profile  
Reply with quote  
PostPosted: 01.07.2008, 17:22 
Offline

Joined: 22.11.2007, 17:05
Posts: 707
Location: Boston, MA
Vurlix wrote:
10.5.4 doesn't solve anything -- no graphics driver update included.

Same here, still no wireframes on ATI.

_________________
Tristam MacDonald - [swiftcoding]


Top
 Profile  
Reply with quote  
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 14 posts ] 

All times are UTC + 1 hour


Who is online

Users browsing this forum: No registered users and 27 guests


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
cron
Powered by phpBB® Forum Software © phpBB Group