NVIDIA Accelerated Linux Driver Set README & Installation Guide
Last Updated: $Date: 2002/11/08 $
Most Recent Driver: 1.0-4050
The NVIDIA Accelerated Linux Driver Set brings both accelerated 2D
functionality and high performance OpenGL support to Linux x86 with the
use of NVIDIA graphics processing units (GPUs).
These drivers provide optimized hardware acceleration of OpenGL
applications via a direct-rendering X Server and support nearly all
NVIDIA graphics chips (please see APPENDIX A for a complete list of
supported chips). TwinView, TV-Out and flat panel displays are also
supported.
This README describes how to install, configure, and use the NVIDIA
Accelerated Linux Driver Set. This file is posted on NVIDIA's web site
(www.nvidia.com), and is installed in /usr/share/doc/NVIDIA_GLX-1.0/
when the NVIDIA_GLX package is installed.
__________________________________________________________________________
CONTENTS:
(sec-01) CHOOSING THE NVIDIA PACKAGES APPROPRIATE FOR YOUR SYSTEM
(sec-02) INSTALLING THE NVIDIA_KERNEL AND NVIDIA_GLX PACKAGES
(sec-03) EDITING YOUR XF86CONFIG FILE
(sec-04) FREQUENTLY ASKED QUESTIONS
(sec-05) CONTACTING US
(sec-06) FURTHER RESOURCES
(app-a) APPENDIX A: SUPPORTED NVIDIA GRAPHICS CHIPS
(app-b) APPENDIX B: MINIMUM SOFTWARE REQUIREMENTS
(app-c) APPENDIX C: INSTALLED COMPONENTS
(app-d) APPENDIX D: XF86CONFIG OPTIONS
(app-e) APPENDIX E: OPENGL ENVIRONMENT VARIABLE SETTINGS
(app-f) APPENDIX F: CONFIGURING AGP
(app-g) APPENDIX G: ALI SPECIFIC ISSUES
(app-h) APPENDIX H: TNT SPECIFIC ISSUES
(app-i) APPENDIX I: CONFIGURING TWINVIEW
(app-j) APPENDIX J: CONFIGURING TV-OUT
(app-k) APPENDIX K: CONFIGURING A LAPTOP
(app-l) APPENDIX L: PROGRAMMING MODES
(app-m) APPENDIX M: PAGE FLIPPING, WINDOW FLIPPING, AND UBB
(app-n) APPENDIX N: KNOWN ISSUES
(app-o) APPENDIX O: PROC INTERFACE
(app-p) APPENDIX P: XVMC SUPPORT
(app-q) APPENDIX Q: GLX SUPPORT
Please note that, in order to keep the instructions more concise, most
caveats and frequently encountered problems are not detailed in the
installation instructions, but rather in the FREQUENTLY ASKED QUESTIONS
section. Therefore, it is recommended that you read this entire README
before proceeding to perform any of the steps described.
__________________________________________________________________________
(sec-01) CHOOSING THE NVIDIA PACKAGES APPROPRIATE FOR YOUR SYSTEM
__________________________________________________________________________
NVIDIA has a unified driver architecture model; this means that one driver
set can be used with all supported NVIDIA hardware. Please see Appendix
A for a list of the NVIDIA hardware supported by the current drivers.
The NVIDIA Accelerated Linux Driver Set consists of two packages
which you will need to download and install: the NVIDIA_GLX package
and the NVIDIA_kernel package. The NVIDIA_GLX package contains the
OpenGL libraries and the XFree86 driver; the NVIDIA_kernel package
contains the NVIDIA kernel module which is needed by the libraries in the
NVIDIA_GLX package (for more details on the components of each package,
please see Appendix C). You will need to install both packages, with
matching version numbers (eg NVIDIA_GLX-1.0-2960 should only be used
with NVIDIA_kernel-1.0-2960 and not NVIDIA_kernel-1.0-2880).
The packages are available in several formats: rpm, srpm, and tar file.
Installation of each package type is described below. The package
type is largely a matter of personal preference, though please note
that the binary rpms are for use only with the kernel shipped with a
particular distribution (eg NVIDIA_kernel-1.0-2960.rh73up.i386.rpm should
only be used with the uni-processor kernel shipped with RedHat 7.3).
Where appropriate, NVIDIA has provided separate rpms for the distinct SMP
and uni-processor kernels of each distribution. If you have upgraded
your kernel (either manually, or through a distribution upgrade), or
a specific NVIDIA_kernel rpm is not available for your distribution,
then please use either the NVIDIA_kernel srpm or tar file.
In the case where distributors ship multiple kernels (as is often the
case with uni-processor and SMP machines), there will be multiple
rpms available, eg: NVIDIA_kernel-1.0-2960.rh73up.i686.rpm and
NVIDIA_kernel-1.0-2960.rh73smp.i686.rpm.
The NVIDIA_GLX rpm, however, is not dependent upon the kernel version,
and therefore an srpm is not needed. Install the NVIDIA_GLX package
either by rpm or tar file.
__________________________________________________________________________
(sec-02) INSTALLING THE NVIDIA_KERNEL AND NVIDIA_GLX PACKAGES
__________________________________________________________________________
BEFORE YOU BEGIN DRIVER INSTALLATION
Before beginning the driver installation, you should exit the X server.
In addition you should set your default run level so you will boot to
console and not start up X (please consult the documentation that came
with your Linux distribution if you are unsure how to do this). This will
make it easier to recover if there is a problem during the installation.
Please note that package revision numbers have been omitted in the
following directions to make them as general as possible. While the
directions might say "NVIDIA_kernel.tar.gz" you should replace
that with the name of the driver version you are installing; eg:
"NVIDIA_kernel.1.0-2960.tar.gz".
INSTALLING BY RPM
Instructions for the Impatient:
$ rpm -ivh NVIDIA_kernel.i386.rpm
$ rpm -ivh NVIDIA_GLX.i386.rpm
Instructions:
Before installing from rpm, make sure that you have downloaded the
NVIDIA_kernel rpm appropriate for your kernel. Once you have verified
that you do indeed have the correct rpm, install NVIDIA_kernel by doing:
$ rpm -ivh NVIDIA_kernel.i386.rpm
Next, install the NVIDIA_GLX rpm by doing:
$ rpm -ivh NVIDIA_GLX.i386.rpm
UPGRADING BY RPM
Instructions for the Impatient:
$ rpm -Uvh NVIDIA_kernel.i386.rpm
$ rpm -e NVIDIA_GLX
$ rpm -ivh NVIDIA_GLX.i386.rpm
Instructions:
Before upgrading from rpm, make sure that you have downloaded the
NVIDIA_kernel rpm appropriate for your kernel. Once you have verified
that you do indeed have the correct rpm, upgrade the NVIDIA_kernel
package by doing:
$ rpm -Uvh NVIDIA_kernel.i386.rpm
You should not use the '-U' option to rpm to upgrade the NVIDIA_GLX
rpm because a bug in the uninstall section of older NVIDIA rpms will
cause some files to be removed that shouldn't be. Instead, use '-e'
to remove the old NVIDIA_GLX rpm, and then install the new one:
$ rpm -e NVIDIA_GLX
$ rpm -ivh NVIDIA_GLX.i386.rpm
INSTALLING/UPGRADING BY SRPM
Instructions for the Impatient:
$ rpm --rebuild NVIDIA_kernel.src.rpm
$ rpm -ivh /path/to/rpms/RPMS/i386/NVIDIA_kernel.i386.rpm
$ rpm -ivh NVIDIA_GLX.i386.rpm
Instructions:
To build a custom NVIDIA_kernel rpm for your system, first ensure that
the headers for your kernel are installed (RedHat, for example includes
the kernel headers in an rpm called "kernel-source"). Then, pass rpm the
'--rebuild' flag:
$ rpm --rebuild NVIDIA_kernel.src.rpm
NOTE: more recent versions of rpm no longer support the "--rebuild"
option, in which case you must instead perform the command:
$ rpmbuild --rebuild NVIDIA_kernel.src.rpm
Watch for the line that looks something like (the path may be different):
Wrote: /usr/src/redhat/RPMS/i386/NVIDIA_kernel.i386.rpm
and use that as input to rpm to install:
$ rpm -ivh /usr/src/redhat/RPMS/i386/NVIDIA_kernel.i386.rpm
or upgrade:
$ rpm -Uvh /usr/src/redhat/RPMS/i386/NVIDIA_kernel.i386.rpm
To install the NVIDIA_GLX package, follow the instructions above for
either installing or upgrading NVIDIA_GLX from rpm.
INSTALLING/UPGRADING BY TAR FILE
Instructions for the Impatient:
$ tar xvzf NVIDIA_kernel.tar.gz
$ tar xvzf NVIDIA_GLX.tar.gz
$ cd NVIDIA_kernel
$ make install
$ cd ../NVIDIA_GLX
$ make install
Instructions:
To install from tar file, unpack each file:
$ tar xvzf NVIDIA_kernel.tar.gz
$ tar xvzf NVIDIA_GLX.tar.gz
cd into the NVIDIA_kernel directory. Type 'make install'. This will
compile the kernel interface to the NVIDIA kernel module, link it,
copy it into place, and attempt to load it into the running kernel:
$ cd NVIDIA_kernel
$ make install
Next, move into the NVIDIA_GLX directory. Type 'make install' -- this
will copy the needed OpenGL and XFree86 files into place:
$ cd ../NVIDIA_GLX
$ make install
Note that the "make install" for each package will remove any previously
installed NVIDIA drivers.
__________________________________________________________________________
(sec-03) EDITING YOUR XF86CONFIG FILE
__________________________________________________________________________
When XFree86 4.0 was released, it used a slightly different XF86Config
file syntax than the 3.x series did, and so to allow both 3.x and 4.x
versions of XFree86 to co-exist on the same system, it was decided that
XFree86 4.x was to use the configuration file "/etc/X11/XF86Config-4"
if it existed, and only if that file did not exist would the file
"/etc/X11/XF86Config" be used (actually, that is an over-simplification
of the search criteria; please see the XF86Config man page for a complete
description of the search path). Please make sure you know what
configuration file XFree86 is using. If you are in doubt, look for a
line beginning with "(==) Using config file:" in your XFree86 log file
("/var/log/XFree86.0.log"). This README will use "XF86Config" to refer
to your configuration file, whatever it is named.
If you do not have a working XF86Config file, there are several ways
to start: there is a sample config file that comes with XFree86, and
there is a sample config file included with the NVIDIA_GLX package (it
gets installed in /usr/share/doc/NVIDIA_GLX-1.0/). You could also use
a program like 'xf86config'; some distributions provide their own tool
for generating an XF86Config file. For more on XF86Config file syntax,
please refer to the man page.
If you already have an XF86Config file working with a different driver
(such as the 'nv' or 'vesa' driver), then all you need to do is find
the relevant Device section and replace the line:
Driver "nv"
(or Driver "vesa")
with
Driver "nvidia"
In the Module section, make sure you have:
Load "glx"
You should also remove the following lines:
Load "dri"
Load "GLcore"
if they exist. There are also numerous options that can be added to
the XF86Config file to fine-tune the NVIDIA XFree86 driver. Please see
Appendix D for a complete list of these options.
Once you have configured your XF86Config file, you are ready to restart
X and begin using the accelerated OpenGL libraries. After you restart X,
you should be able to run any OpenGL application and it will automatically
use the new NVIDIA libraries. If you encounter any problems, please
see the FREQUENTLY ASKED QUESTIONS section below.
__________________________________________________________________________
(sec-04) FREQUENTLY ASKED QUESTIONS
__________________________________________________________________________
Q: Where should I start when diagnosing display problems?
A: One of the most useful tools for diagnosing problems is the XFree86
log file in /var/log (the file is named: "/var/log/XFree86.<#>.log",
where "<#>" is the server number -- usually 0). Lines that begin with
"(II)" are information, "(WW)" are warnings, and "(EE)" are errors.
You should make sure that the correct config file (ie the config file
you are editing) is being used; look for the line that begins with:
"(==) Using config file:". Also check that the NVIDIA driver is being
used, rather than the 'nv' or 'vesa' driver; you can look for: "(II)
LoadModule: "nvidia"", and lines from the driver should begin with:
"(II) NVIDIA(0)".
Q: How can I increase the amount of data printed in the XFree86 log file?
A: By default, the NVIDIA X driver prints relatively few messages to
stderr and the XFree86 log file. If you need to troubleshoot, then
it may be helpful to enable more verbose output by using the XFree86
command line options "-verbose" and "-logverbose" which can be used
to set the verbosity level for the stderr and log file messages,
respectively. The NVIDIA X driver will output more messages when the
verbosity level is at or above 5 (XFree86 defaults to verbosity level
1 for stderr and level 3 for the log file). So, to enable verbose
messaging from the NVIDIA X driver to both the log file and stderr,
you could start X by doing the following: 'startx -- -verbose 5
-logverbose 5'.
Q: My X server fails to start, and my XFree86 log file contains the error:
"(EE) NVIDIA(0): Failed to initialize the NVIDIA kernel module!"
A: Nothing will work if the NVIDIA kernel module doesn't function
properly. If you see anything in the X log file like "(EE) NVIDIA(0):
Failed to initialize the NVIDIA kernel module!" then there is
most likely a problem with the NVIDIA kernel module. First, you
should verify that if you installed from rpm that the rpm was built
specifically for the kernel you are using. You should also check
that the module is loaded ('/sbin/lsmod'); if it is not loaded try
loading it explicitly with 'insmod' or 'modprobe' (be sure to exit the
X server before installing a new kernel module). If you receive errors
about unresolved symbols, then the kernel module has most likely been
built using header files for a different kernel revision than what
you are running. You can explicitly control what kernel header files
are used by building the NVIDIA kernel module from the NVIDIA_kernel
tar file with: 'make install SYSINCLUDE=/path/to/kernel/headers'.
Please note that the convention for the location of kernel header
files changed approximately at the time of the 2.4.0 kernel release,
as did the location of kernel modules. If the kernel module fails to
load properly, modprobe/insmod may be trying to load an older kernel
module (assuming you've upgraded). cd'ing into the directory with
the new kernel module and doing 'insmod ./nvidia.o' may help.
Another cause may be that the /dev/nvidia* device files may be missing.
Finally, the NVIDIA kernel module may print error messages indicating
a problem -- to view these messages please check /var/log/messages, or
wherever syslog is directed to place kernel messages. These messages
are prepended with "NVRM".
Q: X starts for me, but OpenGL applications terminate immediately.
A: If X starts, but OpenGL causes problems, you most likely have a
problem with other libraries in the way, or there are stale symlinks.
See Appendix C for details. Sometimes, all it takes is to rerun
'ldconfig'.
You should also check that the correct extensions are present;
'xdpyinfo' should show the "GLX", "NV-GLX" and "NVIDIA-GLX" extensions
present. If these three extensions are not present, then there is
most likely a problem with the glx module getting loaded or it is
unable to implicitly load GLcore. Check your XF86Config file and make
sure that you are loading glx (see "Editing Your XF86Config File"
above). If your XF86Config file is correct, then check the XFree86
log file for warnings/errors pertaining to GLX. Also check that all
of the necessary symlinks are in place (refer to Appendix C).
Q: When installing/upgrading by srpm, the command:
`rpm --rebuild NVIDIA_kernel-1.0-4050.src.rpm`
only prints out a list of rpm command line options.
A: You most likely don't have the rpm development packages installed.
In most situations you can fix this problem by installing the rpm-devel
package for your distribution. Alternatively, you can install/upgrade
by tar file as the tar files don't require rpm.
Q: When installing/upgrading by srpm, the command:
`rpm --rebuild NVIDIA_kernel-1.0-4050.src.rpm`
reports the error:
NVIDIA_kernel-.src.rpm:no such file or directory
A: You need to install the rpm-build package for your distribution.
Alternatively, you can install/upgrade by tar file as the tar files
don't require rpm.
Q: Installing the NVIDIA_kernel module gives an error message like:
#error Modules should never use kernel-headers system headers
#error but headers from an appropriate kernel-source
A: You need to install the source for the Linux kernel. In most
situations you can fix this problem by installing the kernel-source
package for your distribution
Q: OpenGL applications exit with the following error message:
Error: Could not open /dev/nvidiactl because the permissions
are too restrictive. Please see the FREQUENTLY ASKED QUESTIONS
section of /usr/share/doc/NVIDIA_GLX-1.0/README for steps
to correct.
A: It is likely that a security module for the PAM system may be
changing the permissions on the NVIDIA device files. In most cases
this security system works, but it can get confused. To correct this
problem it is recommended that you disable this security feature.
Different Linux distributions have different files to control this;
please consult with your distributor for the correct method of
disabling this security feature. As an example, if your system has
the file
/etc/security/console.perms
then you should edit the file and remove the line that starts with
"<dri>" (we have also received reports that additional references to
<dri> in console.perms must be removed, but this has not been verified
by NVIDIA). If instead your system has the file
/etc/logindevperms
then you should edit the file and remove the line that lists
/dev/nvidiactl. The above steps will prevent the PAM security system
from modifying the permissions on the NVIDIA device files. Next,
you will need to reset the permissions on the device files back
to their original permissions and owner. You can do that with the
following commands:
chmod 0666 /dev/nvidia* chown root /dev/nvidia*
Q: OpenGL applications crash and print out the following warning:
WARNING: Your system is running with a buggy dynamic loader.
This may cause crashes in certain applications. If you
experience crashes you can try setting the environment
variable __GL_SINGLE_THREADED to 1. For more information
please consult the FREQUENTLY ASKED QUESTIONS section in
the file /usr/share/doc/NVIDIA_GLX-1.0/README.
A: The dynamic loader on your system has a bug which will cause
applications linked with pthreads, and that dlopen() libGL multiple
times, to crash. This bug is present in older versions of the dynamic
loader. Distributions that shipped with this loader include but
are not limited to RedHat Linux 6.2 and Mandrake Linux 7.1. Version
2.2 and later of the dynamic loader are known to work properly. If
the crashing application is single threaded then setting the environment
variable __GL_SINGLE_THREADED to 1 will prevent the crash.
In the bash shell you would enter:
export __GL_SINGLE_THREADED=1
and in csh and derivatives use:
setenv __GL_SINGLE_THREADED 1
Previous releases of the NVIDIA Accelerated Linux Driver Set attempted
to work around this problem, however the workaround caused problems with
other applications and was removed after version 1.0-1541.
Q: When I run Quake3, it crashes when changing video modes; what's wrong?
A: You are probably experiencing the problem described above. Please
check the text output for the "WARNING" message describe in the
previous hint. Setting __GL_SINGLE_THREADED to 1 as described
above, before running Quake3 will fix the problem.
Q: When I start X it fails and my XFree86 log file contains:
(II) LoadModule: "nvidia"
(II) Loading /usr/X11R6/lib/modules/drivers/nvidia_drv.o
No symbols found in this module
(EE) Failed to load /usr/X11R6/lib/modules/drivers/nvidia_drv.o
(II) UnloadModule: "nvidia"
(EE) Failed to load module "nvidia" (loader failed, 256)
...
(EE) No drivers available.
A: The nvidia_drv.o X driver has been stripped of needed symbols;
some versions of rpm (wrongly) strip object files while installing.
You should probably upgrade your version of rpm. Or, you can install
the NVIDIA_GLX package from tar file.
Q: My system runs, but seems unstable. What's wrong?
A: Your stability problems may be AGP-related. See Appendix F for
details.
Q: The kernel module doesn't get loaded dynamically when X starts;
I always have to do 'modprobe nvidia' first. What's wrong?
A: Make sure the line "alias char-major-195 nvidia" appears in
your module configuration file, generally one of "/etc/conf.modules",
"/etc/modules.conf" or "/etc/modutils/alias"; consult the documentation
that came with your distribution for details.
Q: I can't build the NVIDIA kernel module, or I can build the NVIDIA
kernel module, but modprobe/insmod fails to load the module into
my kernel. What's wrong?
A: These problems are generally caused by the build using the wrong kernel
header files (ie header files for a different kernel version than
the one you are running). The convention used to be that kernel
header files should be stored in "/usr/include/linux/", but that is
deprecated in favor of "/lib/modules/`uname -r`/build/include".
The NVIDIA_kernel Makefile should be able to determine the
location on your system; however, if you encounter a problem you
can force the build to use certain header files by doing: 'make
SYSINCLUDE=/path/to/kernel/headers'. Obviously, for any of this to
work, you need the appropriate kernel header files installed on your
system. Consult the documentation that came with your distribution;
some distributions don't install the kernel header files by default,
or they install headers that don't coincide properly with the kernel
you are running.
Q: Why do OpenGL applications run so slow?
A: The application is probably using a different library still on your
system, rather than the NVIDIA supplied OpenGL library. Please see
APPENDIX C for details.
Q: There are problems running Quake2.
A: Quake2 requires some minor setup to get it going. First, in the Quake2
directory, the install creates a symlink called libGL.so that points
at libMesaGL.so. This symlink should be removed or renamed. Then,
to run Quake2 in OpenGL mode, you would type: 'quake2 +set vid_ref glx
+set gl_driver libGL.so'. Quake2 does not seem to support any kind of
full-screen mode, but you can run your X server at whatever resolution
Quake2 runs at to emulate full-screen mode.
Q: There are problems running Heretic II.
A: Heretic II also installs, by default, a symlink called libGL.so in
the application directory. You can remove or rename this symlink, since
the system will then find the default libGL.so (which our
drivers install in /usr/lib). From within Heretic II you
can then set your render mode to OpenGL in the video menu.
There is also a patch available to Heretic II from lokigames at:
http://www.lokigames.com/products/heretic2/updates.php3
Q: Where can I get gl.h or glx.h so I can compile OpenGL programs?
A: Most systems come with these headers preinstalled. However, NVIDIA
has provided our own gl.h and glx.h file in case your system did not
come with them or in case you want to develop OpenGL apps that use
the new NVIDIA OpenGL extensions. These files have been installed in
/usr/share/doc/NVIDIA_GLX-1.0/include/GL to avoid conflicting with
the system installed versions. To use these headers copy them
into /usr/include/GL.
Q: Can I receive email notification of new NVIDIA Accelerated Linux
Driver Set releases?
A: Yes. Fill out the form at:
http://www.nvidia.com/view.asp?FO=driver_update
Q: My system hangs when vt-switching if I have rivafb enabled.
A: Using both rivafb and the NVIDIA kernel module at the same time is
currently broken. In general, using two independent software drivers
to drive the same piece of hardware is a bad idea.
Q: Compiling the NVIDIA kernel module gives this error:
You appear to be compiling the NVIDIA kernel module with
a compiler different from the one that was used to compile
the running kernel. This may be perfectly fine, but there
are cases where this can lead to unexpected behaviour and
system crashes.
If you know what you are doing and want to override this
check, you can do so by setting IGNORE_CC_MISMATCH.
In any other case, set the CC environment variable to the
name of the compiler that was used to compile the kernel.
A: You should compile the NVIDIA kernel module with the same compiler
version that was used to compile your kernel. Some Linux kernel data
structures are dependent on the version of gcc used to compile it;
for example, in include/linux/spinlock.h:
...
* Most gcc versions have a nasty bug with empty initializers.
*/
#if (__GNUC__ > 2)
typedef struct { } rwlock_t;
#define RW_LOCK_UNLOCKED (rwlock_t) { }
#else
typedef struct { int gcc_is_buggy; } rwlock_t;
#define RW_LOCK_UNLOCKED (rwlock_t) { 0 }
#endif
If the kernel is compiled with gcc 2.x, but gcc 3.x is used when the
open files in the NVIDIA_kernel package are built (or vice versa),
the size of rwlock_t will vary, and things like ioremap will fail.
To check what version of gcc was used to compile your kernel, you
can examine the output of:
cat /proc/version
To check what version of gcc is currently in your $PATH, you can
examine the output of:
gcc -v
Q: X fails with error "Failed to allocate LUT context DMA"
A: This is one of the possible consequences of compiling the open files
in the NVIDIA_kernel package with a different gcc version than used
to compile the Linux kernel (see above).
Q: There aren't NVIDIA_kernel RPMs available for release N from
<insert your favorite distro here>. I tried installing the RPM for
version N-1, but that didn't work. What should I do?
A: Like it says in (sec-01) CHOOSING THE NVIDIA PACKAGES APPROPRIATE
FOR YOUR SYSTEM, if "a specific NVIDIA_kernel rpm is not available for
your distribution, then use either the NVIDIA_kernel srpm or tar file."
Q: When I install the NVIDIA_GLX package it says:
--- The above file(s) possibly belong to a conflicting MESA rpm.
--- They have been renamed to xxx.<originalFile>.RPMSAVE to
--- avoid conflicting with the files contained within this
--- package.
--- Please see the FREQUENTLY ASKED QUESTIONS section of
--- /usr/share/doc/NVIDIA_GLX-1.0/README for more details.
What's wrong?
A: Like the message says conflicting files have been moved aside to
insure that your applications find the newly installed OpenGL
libraries. There is no need for alarm, the message is purely
informational. If you uninstall the NVIDIA_GLX package then the
original files will be restored automatically.
Q: What is NVIDIA's policy towards development series Linux kernels?
A: NVIDIA does not officially support development series kernels.
However, all the kernel module source code that interfaces with the
Linux kernel is available in the NVIDIA_kernel package, and NVIDIA
encourages members of the Linux community to develop patches to these
source files to support development series kernels. A google search
will most likely yield several community supported patches.
Q: I recently updated various libraries on my system using my Linux
distributor's update utility, and the NVIDIA graphics driver no
longer works. What's wrong?
A: Conflicting libraries may have been installed by your
distribution's update utility; please see APPENDIX C: INSTALLED
COMPONENTS for details on how to diagnose this.
Q: `rpm --rebuild` gives an error "unknown option".
A: Recent versions of rpm no longer support the "--rebuild" option;
if you have such a version of rpm, you should instead use the command
`rpmbuild --rebuild`. The `rpmbuild` executable is provided by the
rpm-build package.
__________________________________________________________________________
(sec-05) CONTACTING US
__________________________________________________________________________
There is an NVIDIA Linux Driver web forum. You can access it by going
to www.nvnews.net and following the "Forum" and "Linux Discussion Area"
links. This is the preferable tool for seeking help; users can post
questions, answer other users' questions, and search the archives of
previous postings.
If, all else fails, you can contact NVIDIA for support at:
linux-bugs@nvidia.com. But please, only send email to this address
after you've followed the FREQUENTLY ASKED QUESTIONS section in this
README and asked for help on the nvnews.net web forum.
__________________________________________________________________________
(sec-06) FURTHER RESOURCES
__________________________________________________________________________
Linux OpenGL ABI
http://oss.sgi.com/projects/ogl-sample/ABI/
NVIDIA Linux HowTo
http://www.tldp.org/HOWTO/XFree86-Video-Timings-HOWTO/index.html
OpenGL
www.opengl.org
The XFree86 Project
www.xfree86.org
#nvidia (irc.openprojects.net)
__________________________________________________________________________
(app-a) APPENDIX A: SUPPORTED NVIDIA GRAPHICS CHIPS
__________________________________________________________________________
NVIDIA CHIP NAME DEVICE PCI ID
o RIVA TNT 0x0020
o RIVA TNT2 0x0028
o RIVA TNT2 Ultra 0x0029
o Vanta 0x002C
o RIVA TNT2 Model 64 0x002D
o Aladdin TNT2 0x00A0
o GeForce 256 0x0100
o GeForce DDR 0x0101
o Quadro 0x0103
o GeForce2 MX/MX 400 0x0110
o GeForce2 MX 100/200 0x0111
o GeForce2 Go 0x0112
o Quadro2 MXR/EX/Go 0x0113
o GeForce2 GTS 0x0150
o GeForce2 Ti 0x0151
o GeForce2 Ultra 0x0152
o Quadro2 Pro 0x0153
o GeForce4 MX 460 0x0170
o GeForce4 MX 440 0x0171
o GeForce4 MX 420 0x0172
o GeForce4 MX 440-SE 0x0173
o GeForce4 440 Go 0x0174
o GeForce4 420 Go 0x0175
o GeForce4 420 Go 32M 0x0176
o GeForce4 460 Go 0x0177
o Quadro4 500/550 XGL 0x0178
o GeForce4 440 Go 64M 0x0179
o Quadro4 200/400 NVS 0x017A
o Quadro4 500 GoGL 0x017C
o GeForce4 410 Go 16M 0x017D
o GeForce4 MX 440 with AGP8X 0x0181
o GeForce4 MX 440SE with AGP8X 0x0182
o GeForce4 MX 420 with AGP8X 0x0183
o Quadro4 580 XGL 0x0188
o Quadro4 280 NVS 0x018A
o Quadro4 380 XGL 0x018B
o NV18M 0x0186
o NV18M Pro 0x0187
o GeForce2 Integrated GPU 0x01A0
o GeForce4 MX Integrated GPU 0x01F0
o GeForce3 0x0200
o GeForce3 Ti 200 0x0201
o GeForce3 Ti 500 0x0202
o Quadro DCC 0x0203
o GeForce4 Ti 4600 0x0250
o GeForce4 Ti 4400 0x0251
o NV25 0x0252
o GeForce4 Ti 4200 0x0253
o Quadro4 900 XGL 0x0258
o Quadro4 750 XGL 0x0259
o Quadro4 600 XGL 0x025A
o Quadro4 700 XGL 0x025B
o GeForce4 Ti 4800 0x0280
o GeForce4 Ti 4200 with AGP8X 0x0281
o GeForce4 Ti 4800 SE 0x0282
o Quadro4 980 XGL 0x0288
o Quadro4 780 XGL 0x0289
o GeForce4 4000 Go 0x0286
o Quadro4 700 GoGL 0x028C
o NV30 0x0300
o NV30 0x0301
o NV30 0x0302
o NV30GL 0x0308
o NV30GL 0x0309
Please note that the RIVA 128/128ZX chips are supported by the open
source 'nv' driver for XFree86, but not by the NVIDIA Accelerated Linux
Driver Set.
If you want to check your Device PCI IDs for comparison with the table
above, you can use either `cat /proc/pci` or `lspci -n`; in the later
case, look for the device with vendor id "10de", eg:
02:00.0 Class 0300:10de:0100 (rev 10)
__________________________________________________________________________
(app-b) APPENDIX B: MINIMUM SOFTWARE REQUIREMENTS
__________________________________________________________________________
o linux kernel 2.2.12 # cat /proc/version
o XFree86 4.0.1 # XFree86 -version
o Kernel modutils 2.1.121 # insmod -V
If you need to build the NVIDIA kernel module:
o binutils 2.9.5 # size --version
o GNU make 3.77 # make --version
o gcc 2.91.66 # gcc --version
If you build from source rpms:
o spec-helper rpm # rpm -qi spec-helper
All official stable kernel releases from 2.2.12 and up are supported;
"prerelease" versions such as "2.4.3-pre2" are not supported, nor are
development series kernels such as 2.3.x or 2.5.x. The linux kernel
can be downloaded from www.kernel.org or one of its mirrors.
binutils and gcc are required only if you install the NVIDIA_kernel
package by srpm or tar file and can be retrieved from www.gnu.org or
one of its mirrors. Note: binutils and gcc are not required by binary
RPM installations.
If you are using XFree86, but do not have a file /var/log/XFree86.0.log,
then you probably have a 3.x version of XFree86 and must upgrade.
If you are setting up XFree86 4.x for the first time, it is often easier
to begin with one of the open source drivers that ships with XFree86
(either 'nv', 'vga' or 'vesa'). Once XFree86 is operating properly with
the open source driver, then it is easier to switch to the nvidia driver.
Note that newer NVIDIA GPUs may not work with older versions of the "nv"
driver shipped with XFree86. For example, the "nv" driver that shipped
with XFree86 version 4.0.1 did not recognize the GeForce2 family and
the Quadro2 MXR GPUs. However, this was fixed in XFree86 version 4.0.2
(XFree86 can be retrieved from www.xfree86.org).
These software packages may also be available through your linux
distributor.
__________________________________________________________________________
(app-c) APPENDIX C: INSTALLED COMPONENTS
__________________________________________________________________________
The NVIDIA Accelerated Linux Driver Set consists of the following
components (the file in parenthesis is the full name of the component
after installation; "x.y.z" denotes the current version -- in these
cases appropriate symlinks are created during installation):
o An XFree86 driver (/usr/X11R6/lib/modules/drivers/nvidia_drv.o);
this driver is needed by XFree86 to use your NVIDIA hardware.
The nvidia_drv.o driver is binary compatible with XFree86 4.0.1
and greater.
o A GLX extension module for XFree86
(/usr/X11R6/lib/modules/extensions/libglx.so.x.y.z); this module is
used by XFree86 to provide server-side glx support.
o An OpenGL library (/usr/lib/libGL.so.x.y.z); this library
provides the API entry points for all OpenGL and GLX function calls.
It is linked to at run-time by OpenGL applications.
o An OpenGL core library (/usr/lib/libGLcore.so.x.y.z); this
library is implicitly used by libGL and by libglx. It contains the
core accelerated 3D functionality. You should not explicitly load
it in your XF86Config file -- that is taken care of by libglx.
o A kernel module (/lib/modules/`uname -r`/video/nvidia.o
or /lib/modules/`uname -r`/kernel/drivers/video/nvidia.o). This
kernel module provides low-level access to your NVIDIA hardware
for all of the above components. It is generally loaded into the
kernel when the X server is started, and is used by the XFree86
driver and OpenGL. nvidia.o consists of two pieces: the binary-only
core, and a kernel interface that must be compiled specifically
for your kernel version. Note that the linux kernel does not have
a consistent binary interface like XFree86, so it is important that
this kernel interface be matched with the version of the kernel that
you are using. This can either be accomplished by compiling yourself,
or using precompiled binaries provided for the kernels shipped with
some of the more common linux distributions.
o OpenGL and GLX header files
(/usr/share/doc/NVIDIA_GLX-1.0/include/GL/gl.h,
/usr/share/doc/NVIDIA_GLX-1.0/include/GL/glx.h). In most
circumstances the system provided headers in /usr/include/GL should
suffice for OpenGL development. But NVIDIA has provided these
headers as they contain the most up to date versions of NVIDIA's
OpenGL extensions. If you wish to make use of these headers it is
recommended that you copy them to /usr/include/GL/.
The first four components listed above (XFree86 driver, GLX module, libGL,
and libGLcore) are included in the NVIDIA_GLX package. The NVIDIA
kernel module is included in the NVIDIA_kernel package.
Documentation and the OpenGL and GLX header files are also part of the
NVIDIA_GLX package and get installed in /usr/share/doc/NVIDIA_GLX-1.0.
Problems will arise if applications use the wrong version of a library.
This can be the case if there are either old libGL libraries or stale
symlinks left lying around. If you think there may be something awry
in your installation, check that the following files are in place
(these are all the files of the NVIDIA Accelerated Linux Driver Set,
plus their symlinks):
/usr/X11R6/lib/modules/drivers/nvidia_drv.o
/usr/X11R6/lib/modules/extensions/libglx.so.x.y.z
/usr/X11R6/lib/modules/extensions/libglx.so -> libglx.so.x.y.z
/usr/lib/libGL.so.x.y.z
/usr/lib/libGL.so.x -> libGL.so.x.y.z
/usr/lib/libGL.so -> libGL.so.x
/usr/lib/libGLcore.so.x.y.z
/usr/lib/libGLcore.so.x -> libGLcore.so.x.y.z
/lib/modules/`uname -r`/video/nvidia.o, or
/lib/modules/`uname -r`/kernel/drivers/video/nvidia.o
Installation of the NVIDIA_kernel package will also create the /dev files:
crw-rw-rw- 1 root root 195, 0 Feb 15 17:21 nvidia0
crw-rw-rw- 1 root root 195, 1 Feb 15 17:21 nvidia1
crw-rw-rw- 1 root root 195, 2 Feb 15 17:21 nvidia2
crw-rw-rw- 1 root root 195, 3 Feb 15 17:21 nvidia3
crw-rw-rw- 1 root root 195, 255 Feb 15 17:21 nvidiactl
If there are other libraries whose "soname" conflicts with that of
the NVIDIA libraries, ldconfig may create the wrong symlinks. It is
recommended that you manually remove or rename conflicting libraries (be
sure to rename clashing libraries to something that ldconfig won't look at
-- we've found that prepending "XXX" to a library name generally does the
trick), rerun 'ldconfig', and check that the correct symlinks were made.
Some libraries that often create conflicts are "/usr/X11R6/lib/libGL.so*"
and "/usr/X11R6/lib/libGLcore.so*".
If the libraries checks out, then verify that the application is using
the correct libraries. For example, to check that the application
/usr/X11R6/bin/gears is using the NVIDIA libraries, you would do:
$ ldd /usr/X11R6/bin/gears
libglut.so.3 => /usr/lib/libglut.so.3 (0x40014000)
libGLU.so.1 => /usr/lib/libGLU.so.1 (0x40046000)
libGL.so.1 => /usr/lib/libGL.so.1 (0x40062000)
libc.so.6 => /lib/libc.so.6 (0x4009f000)
libSM.so.6 => /usr/X11R6/lib/libSM.so.6 (0x4018d000)
libICE.so.6 => /usr/X11R6/lib/libICE.so.6 (0x40196000)
libXmu.so.6 => /usr/X11R6/lib/libXmu.so.6 (0x401ac000)
libXext.so.6 => /usr/X11R6/lib/libXext.so.6 (0x401c0000)
libXi.so.6 => /usr/X11R6/lib/libXi.so.6 (0x401cd000)
libX11.so.6 => /usr/X11R6/lib/libX11.so.6 (0x401d6000)
libGLcore.so.1 => /usr/lib/libGLcore.so.1 (0x402ab000)
libm.so.6 => /lib/libm.so.6 (0x4048d000)
libdl.so.2 => /lib/libdl.so.2 (0x404a9000)
/lib/ld-linux.so.2 => /lib/ld-linux.so.2 (0x40000000)
libXt.so.6 => /usr/X11R6/lib/libXt.so.6 (0x404ac000)
Note the files being used for libGL and libGLcore -- if they are something
other than the NVIDIA libraries, then you will need to either remove the
libraries that are getting in the way, or adjust your ld search path.
If any of this seems foreign to you, then you may want to read the man
pages for "ldconfig" and "ldd" for pointers.
__________________________________________________________________________
(app-d) APPENDIX D: XF86CONFIG OPTIONS
__________________________________________________________________________
The following driver options are supported by the NVIDIA XFree86 driver:
Option "NvAGP" "integer"
Configure AGP support. Integer argument can be one of:
0 : disable agp
1 : use NVIDIA's internal AGP support, if possible
2 : use AGPGART, if possible
3 : use any agp support (try AGPGART, then NVIDIA's AGP)
Please note that NVIDIA's internal AGP support cannot
work if AGPGART is either statically compiled into your
kernel or is built as a module, but loaded into your
kernel (some distributions load AGPGART into the kernel
at boot up). Default: 3 (the default was 1 until after
1.0-1251).
Option "NoLogo" "boolean"
Disable drawing of the NVIDIA logo splash screen at
X startup. Default: the logo is drawn.
Option "RenderAccel" "boolean"
Enable or disable hardware acceleration of the RENDER
extension. Default: hardware acceleration of the RENDER
extension is disabled.
Option "NoRenderExtension" "boolean"
Disable the RENDER extension. Other than recompiling
the X-server, XFree86 doesn't seem to have another way
of disabling this. Fortunatly, we can control this from
the driver so we export this option. This is useful in
depth 8 where RENDER would normally steal most of the
default colormap. Default: RENDER is offered when possible.
Option "UBB" "boolean"
Enable or disable Unified Back Buffer on any Quadro
based GPUs (Quadro4 200/400NVS excluded);
please see Appendix M for a description of UBB.
This option has no affect on non-Quadro chipsets.
Default: UBB is on for Quadro chipsets.
Option "WindowFlip" "boolean"
Enable or disable window flipping when UBB is enabled;
please see Appendix M for a description. This has no
affect when UBB is off. This may improve performance
for 3D applications. Default: Window flipping is off
by default even when UBB is enabled.
Option "PageFlip" "boolean"
Enable or disable page flipping; please see Appendix M for
a description. Default: page flipping is enabled.
Option "DigitalVibrance" "integer"
Enables Digital Vibrance Control. The range of valid
values are 0 through 255. This feature is not available
on products older than GeForce2. Default: 0.
Option "Overlay" "boolean"
Enables RGB workstation overlay visuals. This is only
supported on Quadro4 chips (Quadro4 200/400NVS excluded)
in depth 24. This option causes the server to advertise
the SERVER_OVERLAY_VISUALS root window property and GLX will
report single and double buffered, Z-buffered 16 bit overlay
visuals. The transparency key is pixel 0x0000 (hex). There
is no gamma correction support in the overlay plane. This
feature requires XFree86 version 4.1.0 or newer. NV17/18
based Quadros (ie. 500/550 XGL) have additional restrictions,
namely, overlays are not supported in TwinView mode or with
virtual desktops larger than 2046x2047 in any dimension (eg.
it will not work in 2048x1536 modes). Quadro 7xx/9xx do not
have this restriction.
Default: off.
Option "CIOverlay" "boolean"
Enables Color Index workstation overlay visuals with identical
restrictions to Option "Overlay" above. The server will
offer visuals both with and without a transparency key.
Default: off.
Option "TransparentIndex" "integer"
When color index overlays are enabled, this option allows
the user to choose which pixel is used for the transparent
pixel in visuals featuring transparent pixels. This value
is clamped between 0 and 255 (Note: some applications such
as Alias/Wavefront's Maya require this to be zero in order to
work correctly). Default: 0.
Option "OverlayDefaultVisual" "boolean"
When overlays are used, this option sets the default visual
to an overlay visual thereby putting the root window in the
overlay. This option is not recommended for RGB overlays.
Default: off.
Option "SWCursor" "boolean"
Enable or disable software rendering of the X cursor.
Default: off.
Option "HWCursor" "boolean"
Enable or disable hardware rendering of the X cursor.
Default: on.
Option "CursorShadow" "boolean" Enable or disable use of a
shadow with the hardware accelerated cursor; this is a
black translucent replica of your cursor shape at a
given offset from the real cursor. This option is
only available on GeForce2 or better hardware (ie
everything but TNT/TNT2, GeForce 256, GeForce DDR and
Quadro). Default: no cursor shadow.
Option "CursorShadowAlpha" "integer"
The alpha value to use for the cursor shadow; only
applicable if CursorShadow is enabled. This value must
be in the range [0, 255] -- 0 is completely transparent;
255 is completely opaque. Default: 64.
Option "CursorShadowXOffset" "integer"
The offset, in pixels, that the shadow image will be
shifted to the right from the real cursor image; only
applicable if CursorShadow is enabled. This value must
be in the range [0, 32]. Default: 4.
Option "CursorShadowYOffset" "integer"
The offset, in pixels, that the shadow image will be
shifted down from the real cursor image; only applicable
if CursorShadow is enabled. This value must be in the
range [0, 32]. Default: 2.
Option "ConnectedMonitor" "string"
Allows you to override what the NVIDIA kernel module
detects is connected to your video card. This may
be useful, for example, if you use a KVM (keyboard,
video, mouse) switch and you are switched away when
X is started. In such a situation, the NVIDIA kernel
module can't detect what display devices are connected,
and the NVIDIA X driver assumes you have a single CRT
connected. If, however, you use a digital flat panel
instead of a CRT, use this option to explicitly tell the
NVIDIA X driver what is connected. Valid values for this
option are "CRT" (cathode ray tube), "DFP" (digital flat
panel), or "TV" (television); if using TwinView, this
option may be a comma-separated list of display devices;
e.g.: "CRT, CRT" or "CRT, DFP". Default: string is NULL.
Option "UseEdidFreqs" "boolean"
This option causes the X server to use the HorizSync
and VertRefresh ranges given in a display device's EDID,
if any. EDID provided range information will override
the HorizSync and VertRefresh ranges specified in the
Monitor section. If a display device does not provide an
EDID, or the EDID doesn't specify an hsync or vrefresh
range, then the X server will default to the HorizSync
and VertRefresh ranges specified in the Monitor section.
Option "IgnoreEDID" "boolean"
Disable probing of EDID (Extended Display Identification
Data) from your monitor. Requested modes are compared
against values gotten from your monitor EDIDs (if any)
during mode validation. Some monitors are known to lie
about their own capabilities. Ignoring the values that
the monitor gives may help get a certain mode validated.
On the other hand, this may be dangerous if you don't
know what you are doing. Default: Use EDIDs.
Option "NoDDC" "boolean"
Synonym for "IgnoreEDID"
Option "FlatPanelProperties" "string"
Requests particular properties of any connected flat
panels as a comma-separated list of property=value pairs.
Currently, the only two available properties are 'Scaling'
and 'Dithering'. The possible values for 'Scaling' are:
'default' (the driver will use whatever scaling state
is current), 'native' (the driver will use the flat
panel's scaler, if it has one), 'scaled' (the driver
will use the NVIDIA scaler, if possible), 'centered'
(the driver will center the image, if possible),
and 'aspect-scaled' (the driver will scale with the
NVIDIA scaler, but keep the aspect ratio correct).
The possible values for 'Dithering' are: 'default'
(the driver will decide when to dither), 'enabled' (the
driver will always dither when possible), and 'disabled'
(the driver will never dither). If any property is not
specified, it's value shall be 'default'. An example
properties string might look like:
"Scaling = centered, Dithering = enabled"
Option "UseInt10Module" "boolean"
Enable use of the XFree86 Int10 module to soft-boot all
secondary cards, rather than POSTing the cards through
the NVIDIA kernel module. Default: off (POSTing is
done through the NVIDIA kernel module).
Option "TwinView" "boolean"
Enable or disable TwinView. Please see APPENDIX I for
details. Default: TwinView is disabled.
Option "TwinViewOrientation" "string"
Controls the relationship between the two display devices
when using TwinView. Takes one of the following values:
"RightOf" "LeftOf" "Above" "Below" "Clone". Please see
APPENDIX I for details. Default: string is NULL.
Option "SecondMonitorHorizSync" "range(s)"
This option is like the HorizSync entry in the Monitor
section, but is for the second monitor when using
TwinView. Please see APPENDIX I for details. Default:
none.
Option "SecondMonitorVertRefresh" "range(s)"
This option is like the VertRefresh entry in the Monitor
section, but is for the second monitor when using
TwinView. Please see APPENDIX I for details. Default:
none.
Option "MetaModes" "string"
This option describes the combination of modes to use
on each monitor when using TwinView. Please see APPENDIX
I for details. Default: string is NULL.
Option "NoTwinViewXineramaInfo" "boolean"
When in TwinView, the NVIDIA X driver normally provides a
Xinerama extension that allows X clients (such as window
managers) to call XineramaQueryScreens() to discover
the current TwinView configuration. This confuses some
window mangers, so this option is provided to disable
this behavior. Default: TwinView Xinerama information
is provided.
Option "UseClipIDs" "boolean"
This allows usage of hardware clip id buffers to improve
rendering performance to drawables that are clipped in a
complex way. This is only supported on Quadro4 chips when
UBB is enabled. Enabling this sets aside a small amount
of video ram for the clip id surfaces, typically less than
two megabytes. Default: Clip id surfaces are not used.
Option "Stereo" "integer"
Enable offering of quad-buffered stereo visuals on Quadro.
Integer indicates the type of stereo glasses being used:
1 - DDC glasses. The sync signal is sent to the glasses
via the DDC signal to the monitor. These usually
involve a passthrough cable between the monitor and
video card.
2 - "Blueline" glasses. These usually involve
a passthrough cable between the monitor and video
card. The glasses know which eye to display based
on the length of a blue line visible at the bottom
of the screen. When in this mode, the root window
dimensions are one pixel shorter in the Y dimension
than requested. This mode does not work with virtual
root window sizes larger than the visible root window
size (desktop panning).
3 - Onboard stereo support. This is usually only found
on professional cards. The glasses connect via a
DIN connector on the back of the video card.
4 - TwinView clone mode stereo. On video cards that
support TwinView, the left eye is displayed on the
first display, and the right eye is displayed on the
second display. This is normally used in conjuction
with special projectors to produce 2 polarized
images which are then viewed with polarized glasses.
To use this stereo mode, you must also configure
TwinView in clone mode with the same resolution,
panning offset, and panning domains on each display.
Stereo is only available on Quadro cards, and is not
supported in TwinView (with the exception of TwinView
clone mode stereo, option #4 above). Currently, stereo
operation may be "quirky" on the original Quadro (NV10)
chip and left-right flipping may be erratic. We are
trying to resolve this issue for a future release.
Default: Stereo is not enabled.
__________________________________________________________________________
(app-e) APPENDIX E: OPENGL ENVIRONMENT VARIABLE SETTINGS
__________________________________________________________________________
FULL SCENE ANTI-ALIASING
Anti-aliasing is a technique used to smooth the edges of objects in a
scene to reduce the jagged "stairstep" effect that sometimes appears.
Full scene anti-aliasing is supported on GeForce or newer hardware.
By setting the appropriate environment variable, you can enable full
scene anti-aliasing in any OpenGL application on these GPUs.
Several anti-aliasing methods are available and you can select between
them by setting the __GL_FSAA_MODE environment variable appropriately.
Note that increasing the number of samples taken during FSAA rendering
may decrease performance.
The following tables describe the possible values for __GL_FSAA_MODE
and their effect on various NVIDIA GPUs.
__GL_FSAA_MODE GeForce, GeForce2, Quadro, and Quadro2 Pro
-----------------------------------------------------------------------
0 FSAA disabled
1 FSAA disabled
2 FSAA disabled
3 1.5 x 1.5 Supersampling
4 2 x 2 Supersampling
5 FSAA disabled
__GL_FSAA_MODE GeForce4 MX, Quadro4 500/550 XGL, and
Quadro4 200/400 NVS
-----------------------------------------------------------------------
0 FSAA disabled
1 2x Bilinear Multisampling
2 2x Quincunx Multisampling
3 FSAA disabled
4 2 x 2 Supersampling
5 FSAA disabled
__GL_FSAA_MODE GeForce3, Quadro DCC, GeForce4 Ti, Quadro4 700 XGL,
Quadro4 750 XGL, and Quadro4 900 XGL
-----------------------------------------------------------------------
0 FSAA disabled
1 2x Bilinear Multisampling
2 2x Quincunx Multisampling
3 FSAA disabled
4 4x Bilinear Multisampling
5 4x Gaussian Multisampling
Please note that to use FSAA on Quadro-family products, you must disable
UBB (please see APPENDIX M: PAGE FLIPPING, WINDOW FLIPPING, AND UBB
for details).
ANISOTROPIC TEXTURE FILTERING
Automatic anisotropic texture filtering can be enabled by setting
the environment variable __GL_DEFAULT_LOG_ANISO, The useful values
are:
__GL_DEFAULT_LOG_ANISO GeForce/GeForce2/GeForce4 MX Description
-----------------------------------------------------------------------
0 No anisotropic filtering
1 Enable automatic anisotropic filtering
__GL_DEFAULT_LOG_ANISO GeForce3/GeForce4 Ti Description
-----------------------------------------------------------------------
0 No anisotropic filtering
1 Low anisotropic filtering
2 Medium anisotropic filtering
3 Maximum anisotropic filtering
VBLANK SYNCING
Setting the environment variable __GL_SYNC_TO_VBLANK to a non-zero value
will force glXSwapBuffers to sync to your monitor's vertical refresh rate
(perform a swap only during the vertical blanking period) on GeForce or
newer hardware (ie: everything but TNT/TNT2 products).
__________________________________________________________________________
(app-f) APPENDIX F: CONFIGURING AGP
__________________________________________________________________________
There are several choices for configuring the NVIDIA kernel module's
use of AGP: you can choose to either use NVIDIA's AGP module (NVAGP),
or the AGP module that comes with the linux kernel (AGPGART). This is
controlled through the "NvAGP" option in your XF86Config file:
Option "NvAgp" "0" ... disables AGP support
Option "NvAgp" "1" ... use NVAGP, if possible
Option "NvAgp" "2" ... use AGPGART, if possible
Option "NvAGP" "3" ... try AGPGART; if that fails, try NVAGP
The default is 3 (the default was 1 until after 1.0-1251).
You should use the AGP module that works best with your AGP chip set.
If you are experiencing problems with stability, you may want to start
by disabling AGP and observing if that solves the problems. Then you
can experiment with either of the other AGP modules.
You can query the current AGP status at any time via the /proc filesystem
interface (see APPENDIX O: PROC INTERFACE).
To use the Linux AGPGART module, it will need to be compiled with
your kernel, either statically linked in, or built as a module.
NVIDIA AGP support cannot be used if AGPGART is loaded in the kernel.
It's recommended that you compile AGPGART as a module and make sure that
it is not loaded when trying to use NVIDIA AGP.
Please also note that changing AGP drivers generally requires a reboot
before the changes actually take effect.
The following AGP chipsets are supported by NVIDIA's AGP; for all other
chipsets it's recommended that you use the AGPGART module.
o Intel 440LX
o Intel 440BX
o Intel 440GX
o Intel 815 ("Solano")
o Intel 820 ("Camino")
o Intel 830
o Intel 840 ("Carmel")
o Intel 845 ("Brookdale")
o Intel 845G
o Intel 850 ("Tehama")
o Intel 860 ("Colusa")
o AMD 751 ("Irongate")
o AMD 761 ("IGD4")
o AMD 762 ("IGD4 MP")
o VIA 8371
o VIA 82C694X
o VIA KT133
o VIA KT266
o RCC 6585HE
o Micron SAMDDR ("Samurai")
o Micron SCIDDR ("Scimitar")
o nForce AGP
o ALi 1621
o ALi 1631
o ALi 1647
o ALi 1651
o ALi 1671
o SiS 630
o SiS 633
o SiS 635
o SiS 645
o SiS 730
o SiS 733
o SiS 735
o SiS 745
If you are experiencing AGP stability problems, you should be aware of
the following:
o Support for the processor's Page Size Extension on Athlon Processors
Some linux kernels have a conflicting cache attribute bug that is
exposed by advanced speculative caching in newer AMD Athlon family
processors (AMD Athlon XP, AMD Athlong 4, AMD Athlon MP, and Models 6
and above AMD Duron). This kernel bug usually shows up under heavy use
of accelerated 3D graphics with an AGP graphics card.
Linux distributions based on kernel 2.4.19 and later *should*
incorporate the bug fix. But, older kernels require help from the user
in ensuring that a small portion of advanced speculative caching is
disabled (normally done through a kernel patch) and a boot option is
specified in order to apply the whole fix.
Nvidia's driver automatically disables the small portion of advanced
speculative caching for the affected AMD processors without the need
to patch the kernel; it can be used even on kernels which do already
incorporate the kernel bug fix. Additionally, for older kernels the
user performs the boot option portion of the fix by explicitly disabling
4MB pages. This can be done from the boot command line by specifying:
mem=nopentium
Or by adding the following line to etc/lilo.conf:
append = "mem=nopentium"
o AGP drive strength BIOS setting (Via based mainboards)
Many Via based mainboards allow adjusting the AGP drive strength in
the system BIOS. The setting of this option largely affects system
stability, the range between 0xEA and 0xEE seems to work best for
NVIDIA hardware. Setting either nibble to 0xF generally restults in
severe stability problems.
If you decide to experiment with this, you need to be aware of
the fact that you are doing so at your own risk and that you may
render your system unbootable with improper settings until you
reset the setting to a working value (w/ a PCI graphics card or
by resetting the BIOS to its default values).
o System BIOS version
Make sure to have the latest system BIOS provided by the board
manufacturer.
o AGP Rate
You may want to decrease the AGP rate setting if you are seeing
lockups with the value you are currently using. You can do so with
the NVreg_ReqAGPRate NVIDIA kernel module parameter.
If you are inserting the module manually:
insmod nvidia NVreg_ReqAGPRate=2 # force AGP Rate to 2x
insmod nvidia NVreg_ReqAGPRate=1 # force AGP Rate to 1x
If you are using modprobe (/etc/modules.conf):
alias char-major-195 nvidia
options nvidia NVreg_ReqAGPRate=2 # force AGP Rate to 2x
options nvidia NVreg_ReqAGPRate=1 # force AGP Rate to 1x
On Athlon motherboards with the VIA KX133 or 694X chip set, such as the
ASUS K7V motherboard, NVIDIA drivers default to AGP 2x mode to work around
insufficient drive strength on one of the signals. You can force AGP 4x
by setting NVreg_EnableVia4x to 1. Note that this may cause the system
to become unstable.
On ALi1541 and ALi1647 chipsets, NVIDIA drivers disable AGP to work
around timing issues and signal integrity issues. You can force AGP
to be enabled on these chipsets by setting NVreg_EnableALiAGP to 1.
Note that this may cause the system to become unstable.
__________________________________________________________________________
(app-g) APPENDIX G: ALI SPECIFIC ISSUES
__________________________________________________________________________
The following tips may help stabilize problematic ALI systems:
o Disable TURBO AGP MODE in the BIOS.
o When using a P5A upgrade to BIOS Revision 1002 BETA 2.
o When using 1007, 1007A or 1009 adjust the IO Recovery Time to
4 cycles.
o AGP is disabled by default on some ALi chipsets (ALi1541, ALi1647)
to work around severe system stability problems with these chipsets.
See the comments for NVreg_EnableALiAGP in os-registry.c to force
AGP on anyway.
__________________________________________________________________________
(app-h) APPENDIX H: TNT SPECIFIC ISSUES
__________________________________________________________________________
Most issues pertaining to SGRAM/SDRAM TNT cards should be resolved.
There is the rare chance, however, that your video card has the wrong
BIOS installed, and that this driver will continue to fail for you.
If this driver fails for you, do the following:
o watch your monitor as the system boots. The very first, brief screen
will identify the type of video memory your card has. This will be
either SGRAM or SDRAM.
o get the most recent NVIDIA_kernel tar file
o edit the file "os-registry.c" from the kernel module sources. Look
for the variable "NVreg_VideoMemoryTypeOverride". Set the value of
the variable to the type of memory you have (numerically, see the
line just above it).
o since we don't normally use this variable, change the "#if 0" that is
about 10 lines above the variable to "#if 1".
o rebuild and reinstall the new driver ("make")
__________________________________________________________________________
(app-i) APPENDIX I: CONFIGURING TWINVIEW
__________________________________________________________________________
The TwinView feature is only supported on NVIDIA GPUs that support
dual-display functionality, such as the GeForce2 MX, GeForce2 Go,
Quadro2 MXR, Quadro2 Go, and any of the GeForce4 GPUs. Please consult
with your video card vendor to confirm that TwinView is supported on
your card.
TwinView is a mode of operation where two display devices (digital
flat panels, CRTs, and TVs) can display the contents of a single X screen
in any arbitrary configuration. This method of multiple monitor use
has several distinct advantages over other techniques (such as Xinerama):
o A single X screen is used. The NVIDIA driver conceals all
information about multiple display devices from the X server; as
far as X is concerned, there is only one screen.
o Both display devices share one frame buffer. Thus, all the
the functionality present on a single display (e.g. accelerated
OpenGL) is available on TwinView.
o No additional overhead is needed to emulate having a single
desktop.
XF86CONFIG TWINVIEW OPTIONS
To enable TwinView, you must specify the following options in the Screen
section of your XF86Config file:
Option "TwinView"
Option "SecondMonitorHorizSync" "<hsync range(s)>"
Option "SecondMonitorVertRefresh" "<vrefresh range(s)>"
Option "MetaModes" "<list of metamodes>"
You may also use any of the following options, though they are not
required:
Option "TwinViewOrientation" "<relationship of head 1 to head 0>"
Option "ConnectedMonitor" "<list of connected display devices>"
Please see the detailed descriptions of each option below:
o TwinView
This option is required to enable TwinView; without it, all
other TwinView related options are ignored.
o SecondMonitorHorizSync, SecondMonitorVertRefresh
You specify the constraints of the second monitor through these
options. The values given should follow the same convention as
the "HorizSync" and "VertRefresh" entries in the Monitor section.
As the XF86Config man page explains it: the ranges may be a
comma separated list of distinct values and/or ranges of values,
where a range is given by two distinct values separated by
a dash. The HorizSync is given in kHz, and the VertRefresh
is given in Hz. You may, if you trust your display devices'
EDIDs, use the "UseEdidFreqs" option instead of these options
(see APPENDIX D for a description of the "UseEdidFreqs" option).
o MetaModes
A single MetaMode describes what mode should be used on each
display device at a given time. Multiple MetaModes list the
combinations of modes and the sequence in which they should be
used. When the NVIDIA driver tells X what modes are available,
it is really the minimal bounding box of the MetaMode that is
communicated to X, while the "per display device" mode is kept
internal to the NVIDIA driver. In MetaMode syntax, modes within
a MetaMode are comma separated, and multiple MetaModes are
separated by semicolons. For example:
"<mode name 0>, <mode name 1>; <mode name 2>, <mode name 3>"
Where <mode name 0> is the name of the mode to be used on display
device 0 concurrently with <mode name 1> used on display device 1.
A mode switch will then cause <mode name 2> to be used on display
device 0 and <mode name 3> to be used on display device 1. Here
is a real MetaMode entry from the XF86Config sample config file:
Option "MetaModes" "1280x1024,1280x1024; 1024x768,1024x768"
If you want a display device to not be active for a certain
MetaMode, you can use the mode name "NULL", or simply omit the
mode name entirely:
"1600x1200, NULL; NULL, 1024x768"
or
"1600x1200; , 1024x768"
Optionally, mode names can be followed by offset information
to control the positioning of the display devices within the
virtual screen space; e.g.:
"1600x1200 +0+0, 1024x768 +1600+0; ..."
Offset descriptions follow the conventions used in the X
"-geometry" command line option; i.e. both positive and negative
offsets are valid, though negative offsets are only allowed when
a virtual screen size is explicitly given in the XF86Config file.
When no offsets are given for a MetaMode, the offsets will be
computed following the value of the TwinViewOrientation option
(see below). Note that if offsets are given for any one of the
modes in a single MetaMode, then offsets will be expected for
all modes within that single MetaMode; in such a case offsets
will be assumed to be +0+0 when not given.
When not explicitly given, the virtual screen size will be
computed as the the bounding box of all MetaMode bounding boxes.
MetaModes with a bounding box larger than an explicitly given
virtual screen size will be discarded.
A MetaMode string can be further modified with a "Panning Domain"
specification; eg:
"1024x768 @1600x1200, 800x600 @1600x1200"
A panning domain is the area in which a display device's viewport
will be panned to follow the mouse. Panning actually happens on
two levels with TwinView: first, an individual display device's
viewport will be panned within its panning domain, as long as
the viewport is contained by the bounding box of the MetaMode.
Once the mouse leaves the bounding box of the MetaMode, the entire
MetaMode (ie all display devices) will be panned to follow the
mouse within the virtual screen. Note that individual display
devices' panning domains default to being clamped to the position
of the display devices' viewports, thus the default behavior is
just that viewports remain "locked" together and only perform
the second type of panning.
The most beneficial use of panning domains is probably to
eliminate dead areas -- regions of the virtual screen that are
inaccessible due to display devices with different resolutions.
For example:
"1600x1200, 1024x768"
produces an inaccessible region below the 1024x768
display. Specifying a panning domain for the second display
device:
"1600x1200, 1024x768 @1024x1200"
provides access to that dead area by allowing you to pan the
1024x768 viewport up and down in the 1024x1200 panning domain.
Offsets can be used in conjunction with panning domains to
position the panning domains in the virtual screen space (note
that the offset describes the panning domain, and only affects
the viewport in that the viewport must be contained within the
panning domain). For example, the following describes two modes,
each with a panning domain width of 1900 pixels, and the second
display is positioned below the first:
"1600x1200 @1900x1200 +0+0, 1024x768 @1900x768 +0+1200"
If no MetaMode string is specified, then the X driver uses the
modes listed in the relevant "Display" subsection, attempting
to place matching modes on each display device.
o TwinViewOrientation
This option controls the positioning of the second display
device relative to the first within the virtual X screen, when
offsets are not explicitly given in the MetaModes. The possible
values are:
"RightOf" (the default)
"LeftOf"
"Above"
"Below"
"Clone"
When "Clone" is specified, both display devices will be assigned
an offset of 0,0.
o ConnectedMonitor
This option allows you to override what the NVIDIA kernel
module detects is connected to your video card. This may be
useful, for example, if any of your display devices do not
support detection using Display Data Channel (DDC) protocols.
Valid values for this option are "CRT" (cathode ray tube), "DFP"
(digital flat panel), or "TV" (television); when using TwinView,
this option may be a comma-separated list of display devices;
e.g.: "CRT, CRT" or "CRT, DFP".
Just as in all XF86Config entries, spaces are ignored and all entries
are case insensitive.
FREQUENTLY ASKED TWINVIEW QUESTIONS:
Q: Nothing gets displayed on my second monitor; what's wrong?
A: Monitors that do not support monitor detection using Display Data
Channel (DDC) protocols (this includes most older monitors) aren't
detectable by your NVIDIA card. You need to explicitly tell the NVIDIA
XFree86 driver what you have connected using the "ConnectedMonitor"
option; e.g.:
Option "ConnectedMonitor" "CRT, CRT"
Q: Will window managers be able to appropriately place windows
(e.g. avoiding placing windows across both display devices, or in
inaccessible regions of the virtual desktop)?
A: Yes. The NVIDIA X driver provides a Xinerama extension that allows
X clients (such as window managers) to call XineramaQueryScreens() to
discover the current TwinView configuration. Note that the Xinerama
protocol provides no way to inform clients of when a configuration
change occurs. So, if you modeswitch to a different MetaMode, your
window manager will still think you have the previous configuration.
Using the Xinerama extension, in conjunction with the XF86VidMode
extension to get modeswitch events, window managers should be
able to determine the TwinView configuration at any given time.
Unfortunately, the data provided by XineramaQueryScreens() appears to
confuse some window managers; to workaround such broken window mangers,
you can disable communication of the TwinView screen layout with the
"NoTwinViewXineramaInfo" XF86Config Option (please see Appendix D
for details).
Another solution is to use panning domains to eliminate inaccessible
regions of the virtual screen (see the MetaMode description above).
Be aware that the NVIDIA driver cannot provide the Xinerama extension if
XFree86's own Xinerama extension is being used. Explicitly specifying
Xinerama in the XF86Config file or on the XFree86 commandline will
prohibit NVIDIA's Xinerama extension from installing, so make sure
that XFree86's /var/log/XFree86.0.log is not reporting:
(++) Xinerama: enabled
if you wish the NVIDIA driver to be able to provide the Xinerama
extension while in TwinView.
Q: Why can I not get a resolution of 1600x1200 on the second display
device when using a GeForce2 MX?
A: Because the second display device on the GeForce2 MX was designed to
be a digital flat panel, the Pixel Clock for the second display device
is only 150 MHz. This effectively limits the resolution on the second
display device to somewhere around 1280x1024 (for a description of
how Pixel Clock frequencies limit the programmable modes, see the
XFree86 Video Timings HOWTO). This constraint is not present on
GeForce4 chips -- the maximum pixel clock is the same on both heads.
Q: Do video overlays work across both display devices?
A: Hardware video overlays only work on the first display device.
The current solution is that blitted video is used instead on TwinView.
Q: How are virtual screen dimensions determined in TwinView?
A: After all requested modes have been validated, and the offsets
for each MetaMode's viewports have been computed, the NVIDIA driver
computes the bounding box of the panning domains for each MetaMode.
The maximum bounding box width and height is then found.
Note that one side effect of this is that the virtual width and
virtual height may come from different MetaModes. Given the following
MetaMode string:
"1600x1200,NULL; 1024x768+0+0, 1024x768+0+768"
the resulting virtual screen size will be 1600 x 1536.
Q: Can I play full screen games across both display devices?
A: Yes. While the details of configuration will vary from game to game,
the basic idea is that a MetaMode presents X with a mode whose
resolution is the bounding box of the viewports for that MetaMode.
For example, the following:
Option "MetaModes" "1024x768,1024x768; 800x600,800x600"
Option "TwinViewOrientation" "RightOf"
produce two modes: one whose resolution is 2048x768, and another whose
resolution is 1600x600. Games such as Quake 3 Arena use the VidMode
extension to discover the resolutions of the modes currently available.
To configure Quake 3 Arena to use the above MetaMode string, add the
following to your q3config.cfg file:
seta r_customaspect "1"
seta r_customheight "600"
seta r_customwidth "1600"
seta r_fullscreen "1"
seta r_mode "-1"
Note that, given the above configuration, there is no mode with a
resolution of 800x600 (remember that the MetaMode "800x600, 800x600"
has a resolution of 1600x600"), so if you change Quake 3 Arena to use
a resolution of 800x600, it will display in the lower left corner of
your screen, with the rest of the screen grayed out. To have single
head modes available as well, an appropriate MetaMode string might
be something like:
"800x600,800x600; 1024x768,NULL; 800x600,NULL; 640x480,NULL"
More precise configuration information for specific games is beyond the
scope of this document, but the above examples coupled with numerous
online sources should be enough to point you in the right direction.
__________________________________________________________________________
(app-j) APPENDIX J: CONFIGURING TV-OUT
__________________________________________________________________________
NVIDIA GPU-based video cards with a TV-Out (S-Video) connector can be
employed to use a television as another display device, just like a CRT
or digital flat panel. The TV can be used by itself, or (on appropriate
video cards) in conjunction with another display device in a TwinView
configuration.
If a TV is the only display device connected to your video card, it will
be used as the primary display when you boot your system (ie the console
will come up on the TV just as if it were a CRT). To use your TV with X,
there are a few parameters that you should pay special attention to in
your XF86Config file:
o The VertRefresh and HorizSync values in your monitor section;
please make sure these are appropriate for your television.
Values are generally:
HorizSync 30-50
VertRefresh 60
o The Modes in your screen section; the only valid modes for TV are
640x480 and 800x600, and possibly 1024x768 if the TV encoder on
your video card is a BrookTree 871 -- your XFree86 log file should
tell you what encoder you have (look for the line: "(--) NVIDIA(0):
TV Encoder detected as").
o The "TVStandard" option should be added to your screen section; valid
values are:
"PAL-B" : used in Belgium, Denmark, Finland, Germany, Guinea,
Hong Kong, India, Indonesia, Italy, Malaysia, The
Netherlands, Norway, Portugal, Singapore, Spain,
Sweden, and Switzerland
"PAL-D" : used in China and North Korea
"PAL-G" : used in Denmark, Finland, Germany, Italy, Malaysia,
The Netherlands, Norway, Portugal, Spain, Sweden,
and Switzerland
"PAL-H" : used in Belgium
"PAL-I" : used in Hong Kong and The United Kingdom
"PAL-K1" : used in Guinea
"PAL-M" : used in Brazil
"PAL-N" : used in France, Paraguay, and Uruguay
"PAL-NC" : used in Argentina
"NTSC-J" : used in Japan
"NTSC-M" : used in Canada, Chile, Colombia, Costa Rica, Ecuador,
Haiti, Honduras, Mexico, Panama, Puerto Rico, South
Korea, Taiwan, United States of America, and Venezuela
The line in your XF86Config file should be something like:
Option "TVStandard" "NTSC-M"
If you don't specify a TVStandard, or you specify an invalid value,
the default "NTSC-M" will be used. Note: if your country is not in
the above list, select the country closest to your location.
o The "ConnectedMonitor" option can be used to tell X to use the TV for
display. This should only be needed if your TV is not detected by
the video card, or you use a CRT (or digital flat panel) as your
boot display, but want to redirect X to use the TV. The line in
your config file should be:
Option "ConnectedMonitor" "TV"
o The "TVOutFormat" option can be used to force SVIDEO or COMPOSITE
output. Without this option the driver autodetects the output format.
Unfortunately, it doesn't always do this correctly. The output format
can be forced with the options:
Option "TVOutFormat" "SVIDEO"
or
Option "TVOutFormat" "COMPOSITE"
__________________________________________________________________________
(app-k) APPENDIX K: CONFIGURING A LAPTOP
__________________________________________________________________________
INSTALLATION AND CONFIGURATION
Installation and configuration of the NVIDIA Accelerated Linux Driver
Set on a laptop is the same as for any desktop environment, with a few
minor exceptions, listed below.
Starting in the 1.0-2802 release, information about the internal flatpanel
for use in initializing the display is by default generated on the fly
from data stored in the video BIOS. This can be disabled by setting
the "SoftEDIDs" kernel option to 0. If "SoftEDIDs" is turned off, then
hardcoded data will be chosen from a table, based on the value of the
"Mobile" kernel option.
The "Mobile" kernel option can be set to any of the following values:
0xFFFFFFFF : let the kernel module auto detect the correct value
1 : Dell laptops
2 : non-Compal Toshiba laptops
3 : all other laptops
4 : Compal Toshiba laptops
5 : Gateway laptops
Again, the "Mobile" kernel option is only needed if SoftEDIDs is
disabled; when it is used, it's usually safest to let the kernel
module auto detect the correct value (this is the default behavior).
Should you need to alter either of these options, this can be done by
doing any of the following:
o editing os-registry.c in the NVIDIA_kernel package
o setting the value on the modprobe command line (eg: `modprobe
nvidia NVreg_SoftEDIDs=0 NVreg_Mobile=3`)
o adding an "options" line to your module configuration file,
usually /etc/modules.conf (eg: "options nvidia
NVreg_Mobile=5")
ADDITIONAL FUNCTIONALITY
TWINVIEW
All mobile NVIDIA chips support TwinView. TwinView on a laptop can
be configured in the same way as on a desktop machine (please refer
to APPENDIX I above); note that in a TwinView configuration using
the laptop's internal flat panel and an external CRT, the CRT is the
primary display device (specify it's HorizSync and VertRefresh in
the Monitor section of your XF86Config file) and the flat panel is
the secondary display device (specify it's HorizSync and VertRefresh
through the SecondMonitorHorizSync and SecondMonitorVertRefresh options).
You can also employ the UseEdidFreqs option to acquire the HorizSync and
VertRefresh from the EDID of each display devices, and not worry about
setting them in your XF86Config file (this should only be done if you
trust your display device's reported EDIDs -- please see the description
of the UseEdidFreqs option in APPENDIX D for details).
HOTKEY SWITCHING OF DISPLAY DEVICES
Besides TwinView, mobile NVIDIA chips also have the capacity to react to
an LCD/CRT hotkey event, toggling between each of the connected display
devices and each possible combination of the connected display devices
(note that only 2 display devices may be active at a time). TwinView as
configured in your XF86Config file and hotkey functionality are mutually
exclusive -- if you enable TwinView in your XF86Config file, then the
NVIDIA X driver will ignore LCD/CRT hotkey events.
Another important aspect of hotkey functionality is that you can
dynamically connect and remove display devices to/from your laptop and
hotkey to them without restarting X.
A concern with all of this is how to validate and determine what modes
should be programmed on each display device. First, it is immensely
helpful to use the UseEdidFreqs so that the hsync and vrefresh for
each display device can be retrieved from the display devices' EDID --
otherwise, the semantics of what the contents of the monitor section
mean constantly changes with each hotkey event.
When X is started, or when a change is detected in the list of connected
display devices, a new hotkey sequence list is constructed -- this lists
what display devices will be used with each hotkey event. When a hotkey
event occurs, then the next hotkey state in the sequence is chosen.
Each mode requested in the XF86Config file is validated against each
display device's constraints, and the resulting modes are made available
for that display device. If multiple display devices are to be active
at once, then the modes from each display device are paired together;
if an exact match (same resolution) can't be found, then the closest fit
is found, and the display device with the smaller resolution is panned
within the resolution of the other display device.
When vt-switching away from X, the vga console will always be restored on
the display device on which it was present when X was started. Similarly,
when vt-switching back into X, the same display device configuration
will be used as when you vt-switched away from X, regardless of what
LCD/CRT hotkey activity occurred while vt-switched away.
NON-STANDARD MODES ON LCD DISPLAYS
Some users have had difficulty programming a 1400x1050 mode (the native
resolution of some laptop LCDs). In version 4.0.3, XFree86 added several
1400x1050 modes to its database of default modes, but if you're using
an older version of XFree86, here is a modeline that you can use:
# -- 1400x1050 --
# 1400x1050 @ 60Hz, 65.8 kHz hsync
Modeline "1400x1050" 129 1400 1464 1656 1960
1050 1051 1054 1100 +HSync +VSync
KNOWN LAPTOP ISSUES
o Power Management is not currently supported.
o LCD/CRT hotkey switching is not currently functioning on any
Toshiba laptop, with the exception of the Toshiba Satellite
3000 series.
o TwinView on Satellite 2800 series Toshbia laptops is not currently
functioning.
o The video overlay only works on the first display device on which
you started X. For example, if you start X on the internal LCD,
run a video application that uses the video overlay (uses the
"Video Overlay" adaptor advertised through the XV extension), and
then hotkey switch to add a second display device, the video will
not appear on the second display device. To work around this,
you can either configure the video application to use the "Video
Blitter" adaptor advertised through the XV extension (this is always
available), or hotkey switch to the display device on which you want
to use the video overlay *before* starting X.
__________________________________________________________________________
(app-l) APPENDIX L: PROGRAMMING MODES
__________________________________________________________________________
The NVIDIA Accelerated Linux Driver Set supports all standard VGA and VESA
modes, as well as most user-written custom mode lines; double-scan modes
are supported on all hardware, and interlaced modes are supported on:
GeForce 256, GeForce DDR, Quadro, GeForce2 GTS/GeForce2 Pro, GeForce2 Ti,
GeForce2 Ultra, Quadro2 Pro, and all TNT products.
In general, your display device (monitor/flat panel/television) will be
a greater constraint on what modes you can use than either your NVIDIA
GPU-based video board or the NVIDIA Accelerated Linux Driver Set.
To request one or more standard modes for use in X, you can simply add a
"Modes" line such as:
Modes "1600x1200" "1024x768" "640x480"
in the appropriate Display subsection of your XF86Config file (please see
the XF86Config(4/5) man page for details). The following documentation
is primarily of interest if you compose your own custom mode lines,
experiment with xvidtune(1), or are just interested in learning more.
Please note that this is neither an explanation nor a guide to the fine
art of crafting custom mode lines for XFree86. We leave that, rather,
to documents such as the XFree86 Video Timings HOWTO (which can be found
at www.tldp.org).
DEPTH, BITS PER PIXEL, AND PITCH
While not directly a concern when programming modes, the bits used per
pixel is an issue when considering the maximum programmable resolution;
for this reason, it is worthwhile to address the confusion surrounding
the terms "depth" and "bits per pixel". Depth is how many bits of
data are stored per pixel. Supported depths are 8, 15, 16, and 24.
Most video hardware, however, stores pixel data in sizes of 8, 16, or
32 bits; this is the amount of memory allocated per pixel. When you
specify your depth, X selects the bits per pixel (bpp) size in which to
store the data. Below is a table of what bpp is used for each possible
depth:
depth bpp
===== =====
8 8
15 16
16 16
24 32
Lastly, the "pitch" is how many bytes in the linear frame buffer there are
between one pixel's data, and the data of the pixel immediately below.
You can think of this as the horizontal resolution multiplied by the
bytes per pixel (bits per pixel divided by 8). In practice, the pitch may
be more than this product because video hardware often has requirements
that the pitch be a multiple of some value.
MAXIMUM RESOLUTIONS
The NVIDIA Accelerated Linux Driver Set and NVIDIA GPU-based video boards
support resolutions up to 2048x1536, though the maximum resolution
your system can support is also limited by the amount of video memory
(see USEFUL FORMULAS for details) and the maximum supported resolution
of your display device (monitor/flat panel/television). Also note that
while use of a video overlay does not limit the maximum resolution or
refresh rate, video memory bandwidth used by a programmed mode does
effect the overlay quality.
USEFUL FORMULAS
The maximum resolution is a function both of the amount of video memory
and the bits per pixel you elect to use:
HR * VR * (bpp/8) = Video Memory Used
In other words, the amount of video memory used is equal to the horizontal
resolution (HR) multiplied by the vertical resolution (VR) multiplied by
the bytes per pixel (bits per pixel divided by eight). Technically, the
video memory used is actually the pitch times the vertical resolution,
and the pitch may be slightly greater than (HR * (bpp/8)) to accommodate
hardware requirements that the pitch be a multiple of some value.
Please note that this is just memory usage for the frame buffer; video
memory is also used by other things such as OpenGL or pixmap caching.
Another important relationship is that between the resolution, the pixel
clock (aka dot clock) and the vertical refresh rate:
RR = PCLK / (HFL * VFL)
In other words, the refresh rate (RR) is equal to the pixel clock (PCLK)
divided by the total number of pixels: the horizontal frame length (HFL)
multiplied by the vertical frame length (VFL) (note that these are the
frame lengths, and not just the visible resolutions). As described in
the XFree86 Video Timings HOWTO, the above formula can be rewritten as:
PCLK = RR * HFL * VFL
Given a maximum pixel clock, you can adjust the RR, HFL and VFL as
desired, as long as the product of the three is consistent. The pixel
clock is reported in the log file when you run X with verbose logging:
`startx -- -logverbose 5`. Your XFree86.0.log should contain several
lines like:
(--) NVIDIA(0): Display Device 0: maximum pixel clock at 8 bpp: 350 MHz
(--) NVIDIA(0): Display Device 0: maximum pixel clock at 16 bpp: 350 MHz
(--) NVIDIA(0): Display Device 0: maximum pixel clock at 32 bpp: 300 MHz
which indicate the maximum pixel clock at each bit per pixel size.
HOW MODES ARE VALIDATED
During the PreInit phase of the X server, the NVIDIA X driver validates
all requested modes by doing the following:
o Take the intersection of the HorizSync and VertRefresh ranges given
by the user in the XF86Config with the ranges reported by the monitor
in the EDID (Extended Display Identification Data); this behavior
can be disabled by using the "IgnoreEDID" option in which case the
X driver will blindly accept the HorizSync and VertRefresh ranges
given by the user.
o Call the xf86ValidateModes() helper function, which finds modes with
the names the user specified in the XF86Config file, pruning
out modes with invalid horizontal sync frequencies or vertical
refresh rates, pixel clocks larger than the maximum pixel clock
for the video card, or resolutions larger than the virtual
screen size (if a virtual screen size was specified in the
XF86Config file). Several other constraints are applied; see
xc/programs/Xserver/hw/xfree86/common/xf86Mode.c:xf86ValidateModes().
o All modes returned from xf86ValidateModes() are then examined to make
sure their resolutions are not larger than the largest mode reported
by the monitor's EDID (this can be disabled with the "IgnoreEDID"
option. If the display is a TV, each mode is checked to make sure
it has a resolution that is supported by the TV encoder (usually
only 800x600 and 640x480 are supported by the encoder).
o All remaining modes are then checked to make sure they pass the
constraints described below in ADDITIONAL MODE CONSTRAINTS.
The last two steps are also done when each mode is programmed, to
catch potentially invalid modes submitted by the XF86VidModeExtension
(eg xvidtune(1)). For TwinView, the above validation is done for the
modes requested for each display device.
ADDITIONAL MODE CONSTRAINTS
Below is a list of additional constraints on a mode's parameters that
must be met. In some cases these are specific to particular chips.
o The horizontal resolution (HR) must be a multiple of 8 and be less
than or equal to the value in the table below.
o The horizontal blanking width (the maximum of the horizontal frame
length and the horizontal sync end minus the minimum of the horizontal
resolution and the horizontal sync start (max(HFL,HSE) - min(HR,HSS))
must be a multiple of 8 and be less than or equal to the value in
the table below.
o The horizontal sync start (HSS) must be a multiple of 8 and be less
than or equal to the value in the table below.
o The horizontal sync width (the horizontal sync end minus the
horizontal sync start (HSE - HSS)) must be a multiple of 8 and be
less than or equal to the value in the table below.
o The horizontal frame length (HFL) must be a multiple of 8, must be
greater than or equal to 40, and must be less than or equal to the
value in the table below.
o The vertical resolution (VR) must be less than or equal to the
value in the table below.
o The vertical blanking width (the maximum of the vertical frame length
and the vertical sync end minus the minimum of the vertical resolution
and the vertical sync start (max(VFL,VSE) - min(VR,VSS)) must be
less than or equal to the value in the table below.
o The vertical sync start (VSS) must be less than or equal to the
value in the table below.
o The vertical sync width (the vertical sync end minus the vertical sync
start (VSE - VSS)) must be less than or equal to the value in the
table below.
o The vertical frame length (VFL) must be greater than or equal to 2 and
less than or equal to the value in the table below.
Maximum DAC Values
------------------
| GeForce/TNT Geforce2 & 3 Geforce4 or newer
______|_______________________________________________
|
HR | 4096 4096 8192
HBW | 1016 1016 2040
HSS | 4088 4088 8224
HSW | 256 256 512
HFL | 4128 4128 8224
VR | 2048 4096 8192
VBW | 128 128 256
VSS | 2047 4095 8192
VSW | 16 16 16
VFL | 2049 4097 8192
Here is an example mode line demonstrating the use of each abbreviation
used above:
# Custom Mode line for the SGI 1600SW Flatpanel
# name PCLK HR HSS HSE HFL VR VSS VSE VFL
Modeline "sgi1600x1024" 106.9 1600 1632 1656 1672 1024 1027 1030 1067
SEE ALSO:
An XFree86 modeline generator, conforming to the GTF Standard has
been posted to the XFree86 Xpert mailing list:
http://www.xfree86.org/pipermail/xpert/2001-October/012070.html
For additional modeline generators, try searching for "modeline"
on freshmeat.net.
__________________________________________________________________________
(app-m) APPENDIX M: PAGE FLIPPING, WINDOW FLIPPING, AND UBB
__________________________________________________________________________
Starting with the 1.0-2313 driver release, the NVIDIA Accelerated
Linux Driver Set supports Unified Back Buffer (UBB), Page Flipping,
and Window Flipping. These features can provide performance gains in
certain situtations. Here is a discription of each:
o Page Flipping: This feature is available on all GeForce or newer
hardware (ie: not TNT/TNT2 products), and is enabled in the
case of a single full screen unobscured OpenGL application when
syncing to vblank. Buffer swapping is done by changing which
buffer the DAC scans out rather than copying the back buffer
contents to the front buffer; this is generally a much higher
performance mechanism and allows tearless swapping during the
retrace (when __GL_SYNC_TO_VBLANK is set). This feature can be
disabled with the PageFlip XF86Config option.
o Unified Back Buffer (UBB): UBB is available only on the Quadro family
of GPUs (Quadro4 200/400NVS excluded) and is enabled by default
when there is sufficient video memory available. This can be
disabled with the UBB XF86Config option described in Appendix D.
When UBB is enabled, all windows share the same back, stencil
and depth buffer. When there are many windows, the back, stencil
and depth usage will never exceed the size of that used by a
full screen window. However, even for a single small window
the back, stencil and depth usage are that of a full screen
window so in that case video ram may be used less efficiently
than in the non-UBB case.
o Window Flipping: This feature requires UBB, and thus is only available
on Quadro parts. When there is a single OpenGL window this
window's buffers can be swapped by changing which buffer the DAC
scans out rather than blitting the back buffer contents to the
front buffer. This is similar to Page Flipping but removes the
restriction that the window be unobscured and be full screen.
This only works when there is a single OpenGL window. Window
Flipping is disabled by default and can be enabled with the
"WindowFlip" XF86Config option described in Appendix D.
__________________________________________________________________________
(app-n) APPENDIX N: KNOWN ISSUES
__________________________________________________________________________
The following problems still exist in this release and are in the process
of being resolved.
o OpenGL + Xinerama
Currently, OpenGL will not display to anything other than the
first head in a Xinerama environment.
o OpenGL and dlopen()
There are some issues with older versions of the glibc dynamic
loader (e.g., the version that shipped with RedHat 7.2) and
applications such as Quake3 and Radiant, that use dlopen().
See the FREQUENTLY ASKED QUESTIONS section for more details.
o DPMS and TwinView
DPMS Modes "suspend" and "standby" do not work correctly on
a second CRT when using TwinView. The screen becomes blank
instead of the monitor being set to the requested DPMS state.
o DPMS and Flat Panel
DPMS modes "suspend" and "standby" do not work correctly on a
flat panel display. The screen becomes blank instead of the
flat panel being set to the requested DPMS state.
o Multicard, Multimonitor
In some cases, the secondary card is not initialized correctly
by the NVIDIA kernel module. You can work around this by enabling
the XFree86 Int10 module to soft-boot all secondary cards. See
"APPENDIX D: XF86CONFIG OPTIONS" for details.
o Laptop
If you are using a laptop please see the "Known Laptop Issues" in
APPENDIX D.
o FSAA
When FSAA is enabled (the __GL_FSAA_MODE environment variable
is set to a value that enables FSAA and a multisample visual is
chosen), the rendering may be corrupted when resizing the window.
HARDWARE ISSUES
This section describes problems that will not be fixed. Usually, the
source of the problem is beyond the control of NVIDIA. Following is
the list of problems:
o Gigabyte GA-6BX Motherboard
This motherboard uses a LinFinity regulator on the 3.3-V rail
that is rated to only 5 A -- less than the AGP specification,
which requires 6 A. When diagnostics or applications are
running, the temperature of the regulator rises, causing the
voltage to the NVIDIA chip to drop as low as 2.2 V. Under these
circumstances, the regulator cannot supply the current on the
3.3-V rail that the NVIDIA chip requires.
This problem does not occur when the graphics board has a
switching regulator or when an external power supply is connected
to the 3.3-V rail.
o VIA KX133 and 694X Chip sets with AGP 2x
On Athlon motherboards with the VIA KX133 or 694X chip set, such
as the ASUS K7V motherboard, NVIDIA drivers default to AGP 2x mode
to work around insufficient drive strength on one of the signals.
o Irongate Chip sets with AGP 1x
AGP 1x transfers are used on Athlon motherboards with the Irongate
chip set to work around a problem with the signal integrity of
the chip set.
o ALi chipsets, ALi1541 and ALi1647
On ALi1541 and ALi1647 chipsets, NVIDIA drivers disable AGP to work
around timing issues and signal integrity issues. See "APPENDIX G:
ALI SPECIFIC ISSUES" for more information on ALi chipsets.
__________________________________________________________________________
(app-o) APPENDIX O: PROC INTERFACE
__________________________________________________________________________
The /proc filesystem interface allows you to obtain run-time information
about the driver, any installed NVIDIA graphics cards and the AGP status.
This information is held by several files in /proc/driver/nvidia. This is
a brief description for each one of these files:
o version
Lists the installed driver revision and the version of the GNU C
compiler used to build the Linux kernel module.
o cards/0...3
Provides information about each of the installed NVIDIA graphics
adapters (model name, IRQ, BIOS version, Bus Type). Please note
that the BIOS version is only available while X is running.
o agp/card
Information about the installed AGP card's AGP capabilities.
o agp/host-bridge
Information about the host bridge (model and AGP capabilities).
o agp/status
The current AGP status. If AGP support has been enabled on your
system, the AGP driver being used, the AGP rate and information
about the status of AGP Fast Writes and Side Band Addressing is
shown.
The AGP driver is either one of NVIDIA (NVIDIA's built-in AGP
driver) or AGPGART (the Linux kernel's agpgart.o driver). If
you see "inactive" next to AGPGART, then this means that the
AGP chipset was programmed by AGPGART, but is not currently in
use.
SBA and Fast Writes indicate whether either one of the features
is currently in use. Please note that several factors decide if
support for either will be enabled. First of all, both the AGP
card and the host bridge must support the feature. Even if both
do support it, the driver may decide not to use it in favor of
system stability. This is particularly true of AGP Fast Writes.
__________________________________________________________________________
(app-p) APPENDIX P: XVMC SUPPORT
__________________________________________________________________________
This release includes support for the X-Video Motion Compensation
(XvMC) version 1.0 API on GeForce4 and only GeForce4 products.
There is a static library "libXvMCNVIDIA.a" and a dynamic one
"libXvMCNVIDIA_dynamic.so" which is suitable for dlopening.
GeForce4 MX products support both XvMC's "IDCT" and "motion-compensation"
levels of acceleration. GeForce4 Ti products only support
the motion-compensation level. AI44 and IA44 subpictures
are supported. 4:2:0 Surfaces up to 2032x2032 are supported.
libXvMCNVIDIA observes the XVMC_DEBUG environment variable and will
provide some debug output to stderr when set to an appropriate integer
value. '0' disables debug output. '1' enables debug output for failure
conditions. '2' or higher enables output of warning messages.
__________________________________________________________________________
(app-q) APPENDIX Q: GLX SUPPORT
__________________________________________________________________________
This release supports GLX 1.2 with the following extensions
GLX_EXT_visual_info
GLX_EXT_visual_rating
GLX_SGIX_fbconfig
GLX_SGIX_pbuffer
GLX_ARB_get_proc_address
For a description of these extensions, please see the OpenGL extension
registry at http://oss.sgi.com/projects/ogl-sample/registry/index.html
GLX 1.3 is not yet supported.
Download Driver Pack
After your driver has been downloaded, follow these simple steps to install it.
Expand the archive file (if the download file is in zip or rar format).
If the expanded file has an .exe extension, double click it and follow the installation instructions.
Otherwise, open Device Manager by right-clicking the Start menu and selecting Device Manager.
Find the device and model you want to update in the device list.
Double-click on it to open the Properties dialog box.
From the Properties dialog box, select the Driver tab.
Click the Update Driver button, then follow the instructions.
Very important: You must reboot your system to ensure that any driver updates have taken effect.
For more help, visit our Driver Support section for step-by-step videos on how to install drivers for every file type.