-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Robert Osfield wrote: > Hi Jan, > > On 6/26/06, Jan Ciger <[EMAIL PROTECTED]> wrote: >> Just tested - current CVS checkout still hangs in Viewer::Realize() >> (shows black window) if I enable SLI (both AFR and SFR mode). Ltrace >> shows the application hanging on some ioctl call, most likely somewhere >> in Nvidia driver. > > Does the stack trace go up to a Producer call, if so which one? I > wonder if NVidia have broken some of the GLX calls related to vsync.
The last calls (as determined by ltrace -C -n 4 osgviewer dna.osg) are these: osgUtil::Optimizer::optimize(osg::Node*)(0xbfbb8aa0, 0x8080ad0, 0x3fbe72c9, 118939, 65515) = 0xbfbb872c osgProducer::OsgCameraGroup::setSceneData(osg::Node*)(0xbfbb88b4, 0x8080ad0, 0x3fbe72c9, 118939, 65515) = 0x806ffe4 osgProducer::Viewer::realize()(0xbfbb88b4, 0x8080ad0, 0x3fbe72c9, 118939, 65515 Then it hangs on some ioctl, probably in the driver and doesn't seem to reach Producer. If I run and break it in gdb, the stack is messed up for some reason (just lots of question marks and nothing sensible), pointing into the libGL library. Without SLI this works fine. > I would very much like to drive SLI from the scene graph/Producer > level, with use specifying the composting of the two graphics outputs. > It wouldn't a very difficult extension to the OSG/Producer and it > would allow much better scaling in performance as we could cull > geometry and state at a much higher level. This could be rather cool. Right now we could drive the cards separately, but compositing would be very expensive due to necessary copying of the data back-and-forth between the cards :( Jan - -- Jan Ciger GPG public key: http://www.keyserver.net/ -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.3 (GNU/Linux) Comment: Using GnuPG with Mandriva - http://enigmail.mozdev.org iD8DBQFEoE44n11XseNj94gRAjSDAKClF8bUPoih5x5Er1NH178EYAG0fACfSVrs pagdvV3mrCvWjN/2H+tm1sU= =Aj39 -----END PGP SIGNATURE----- _______________________________________________ osg-users mailing list [email protected] http://openscenegraph.net/mailman/listinfo/osg-users http://www.openscenegraph.org/
