Plugin_CgProgramManager needed?
Moderator: Oberlus
Plugin_CgProgramManager needed?
I just run FreeOrion on a linux box with nvidia gfx without having loaded this OGRE plugin. So, is it needed at all?
Re: Plugin_CgProgramManager needed?
Yes, cg.so and Plugin_CgProgramManager.so are needed. The PagedGeometry code (added a couple of weeks ago) uses Cg.
Re: Plugin_CgProgramManager needed?
Would be interesting to know if modern ATI cards support Cg. Maybe those having "Vertex texture fetch: yes" do so?
Re: Plugin_CgProgramManager needed?
It isn't so simple. I don't know if modern ATI cards support Cg but for a while we presume that answer is yes in binary driver fglrx. ATI released documentation for almost all their cards and for rest of remaining they are planning to do. But after some time, they simply throw away support for this cards and only open source ATI drivers will be available. Therefore answer to this question should be for both drivers - open source and binary.kroddn wrote:Would be interesting to know if modern ATI cards support Cg. Maybe those having "Vertex texture fetch: yes" do so?
Re: Plugin_CgProgramManager needed?
One after another - first it should be clear IF modern ATI cards support Cg - if answer to this is no, then many people will not be able to use FreeOrion at all.
Re: Plugin_CgProgramManager needed?
AFAICT, Cg is hardware-agnostic. Almost all of the Ogre demos use Cg, and it seems hard to imagine that they would use that if only NVidia cards could run the demos.
Re: Plugin_CgProgramManager needed?
I agree. I have an ATI card, get the same segfault when running FO, and have successfully compiled Ogre with Cg and all Ogre demos work fine!tzlaine wrote:AFAICT, Cg is hardware-agnostic. Almost all of the Ogre demos use Cg, and it seems hard to imagine that they would use that if only NVidia cards could run the demos.
Re: Plugin_CgProgramManager needed?
Are you sure that ogre demos even make usage of this plugin?
Re: Plugin_CgProgramManager needed?
Problem may be not in plugin, but in bad usage of it. But it is hard to say without modern ATI graphic card. We need someone with modern ATI graphic card(which supports vertex texture fetch) to test FreeOrion on Windows and Linux.
But problem with segfault was already before revision 2883 after which FO requires this plugin. Therefore discussion about segfault on ATI is going wrong way. We need to test revision 2882(or 2807=FO 0.3.12)
and try to find why it segfaults. But for this we need Ogre programmer with modern ATI graphic card. To test it we need FreeOrion release with revision before revision 2883 to minimize problems with Cg Toolkit.
But problem with segfault was already before revision 2883 after which FO requires this plugin. Therefore discussion about segfault on ATI is going wrong way. We need to test revision 2882(or 2807=FO 0.3.12)
Code: Select all
svn update -r 2882
Re: Plugin_CgProgramManager needed?
I'll compile a debug version of OGRE to debug the problem in detail. For now, I only know that it crashes when executing OgreGUI::Enter2DMode() in HumanClientApp::Enter2DMode() - the debugger was not able to debug into the ogre code.
Re: Plugin_CgProgramManager needed?
Francys in his thread patches Ogre with some specific ATI patch which he linked.