Bjorn, I have seen the same problem with GeForce2 GTS cards running in stereo mode -- nVidia chips seem to have a problem running in mono mode when they are configured with quad-buffer stereo-enabled, and this problem isn't unique or specific to PyMOL. Until this problem is resolved, I can not recommend nVidia-based hardware for PC StereoGraphics -- so far, the Quadro simply doesn't work as well under Win2K as competing stereo-capable cards (such as 3DLabs or ATI FireGL series).
Does anyone have examples of stereo-capable, OpenGL-based programs which don't exhibit this behavior when QBS is enabled in the driver configuration panel? We might then figure out what, if anything, PyMOL and other major scientific 3D programs are doing wrong... FYI: a workaround with PyMOL is to leave hardware stereo enabled, but then eliminate the stereo characteristics. set stereo_angle=0 set stereo_shift=0 Will make the L/R images identical. You'll only be get 50% performance, but that is far better than what is currently happening in monocular mode. Anyone else had experience with stereo-capable nVidia-based hardware? Cheers, Warren On Fri, 28 Jun 2002, [iso-8859-1] Bj?rn Kauppi wrote: > Sorry, not strictly PyMOL. But it relates to my usage of the program. > I have a severe performance problem running the Open-GL based rendering in > mono mode. The graphics are very "jumpy". If I switch to quad buffered stereo > it works smooth and beautifully. > I use windows 2000, PIII 733MHz and a Elsa Gloria II 64Mb graphic card. > It is like the mono mode does not use the hardware acceleration at all! But > when I switch to stereo it kicks in??? > I *think* this was not a problem when I used windows-NT. Does anyone have a > clue what is going on here. And more importantly what can be done about it.
