These instructions for installing an Nvidia card to use with IMOD may be helpful. They were correct as of November 2006. The list of steps assume that you have already installed Linux with xorg-X11 and its associated drivers. 1. You need to have an appropriate source package corresponding to the kernel that you are running. On Fedora systems this is a kernel-devel package or, if you are running a multiprocessor kernel, a kernel-smp-devel package. You can install this with: yum install kernel-devel OR yum install kernel-smp-devel or by installing the corresponding rpm from your distribution disk, e.g. rpm -ivh kernel-devel*.rpm 2. On Fedora Core 5 or 6, the installation procedure will complain, and things may not work, unless you install xorg-x11-server-sdk. You can do this with yum install xorg-x11-server-sdk or by installing the appropriate rpm from your distribution disk. 3. Go to the Nvidia web site (http://www.nvidia.com) and find the page offering Linux drivers (e.g. select Download Drivers - Linux, FreeBSD and Solaris Drivers - Linux IA32 or Linux AMD64/EM64T, Latest Version or Archive for older versions). You must use version 9629 or higher for FC6. Download the driver installer file, e.g. NVIDIA-Linux-x86-1.0-8776-pkg1.run. Notice links to the README file, which you may need to consult. After you have installed the drivers, this file is in /usr/share/doc/NVIDIA_GLX-1.0. 4. Switch the screen to console mode (Ctrl-Alt-F1) and log in, or ssh from another machine. Bring the machine to runlevel 3 with: init 3 5. Go to the location where you have placed your Nvidia install file and run it as recommended in the instructions, e.g. sh NVIDIA-Linux-x86-1.0-8776-pkg1.run If you are on a 32-bit system under FC6 and the procedure fails, claiming that it cannot find the kernel source, then you probably need to install the i586 kernel-devel package instead of the i686 package. To correct this, you will probably need to install an rpm from the distribution disk. First check which kernel source is installed with: ls /usr/src/kernels The directory name will end in either i586 or i686. If it is the wrong one, find out the proper name to use for removing with: rpm -q kernel-devel Then use: rpm -e kernel-devel-... (i.e., use the name just reported) rpm -ivh kernel-devel*i586.rpm 6. Edit your /etc/X11/xorg.conf. Make sure the modules section loads "glx". Make your driver line says "Driver nvidia". A sample xorg.conf file is available for your reference. 7. Type startx and see what happens. Success is indicated by a Gnome or KDE desktop. Failure is a message saying the X server cannot come up. A frequent message is a complaint about no/missing screens. This really means that the nvidia driver could not start for one reason or another. You can examine the start up process in detail by looking in the file /var/log/Xorg.0.log. Make sure that the number of bits specified in the Display line matches up with one of the screens that is declared. 8. Log out of the session started with startx. Return to runlevel 5 with init 5 9. Log in as a normal user. Assuming IMOD is installed normally, run 3dmodv on the file yeastpore.mod (available in http://bio3d.colorado.edu/imod/files/imod_data.tar.gz): 3dmodv yeastpore.mod Start the model spinning by typing Enter and 8 on the numeric keypad. After 5 seconds or more, type o to get an output of frames/second. You should probably get at least 200 FPS (assuming you have at least a GeForce4 with at least a 1.7 GHz processor). If it is a lot slower than this, the Nvidia drivers and GL libraries may not be installed correctly. If the rate is somewhat slower (e.g. below 100), you may have problems with the AGP support. Do cat /proc/driver/nvidia/agp/status and see if AGP is Disabled. 10. Note that you need to reinstall the Nvidia drivers whenever either the kernel or X is updated to a new version. Just follow steps 4, 5 and 8. Or, with recent packages, after the machine is rebooted and fails to start X, you can enter, e.g., sh NVIDIA-Linux-x86-100.14.19-pkg1.run -a -s init 3 init 5