login
Search
NVIDIA have announced a new release of their graphic card drivers.
This release has support for Linux 2.6 kernels, fixes AGP failures on some VIA motherboards, and fixes a problem that prevented X from running on Samsung X10 laptops.
For those who've sold their GPL soul to use the binary drivers from NVIDIA (like me) you can get them now.
From the x86 driver README
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.
Works a charm!
Downloaded, installed and am trying it...
Works a charm!
Also it doesn't seem to suffer from the slow framerate the the previous release did (5326) which forced me to drop back a version. Very nice.
I'm sure this release will get a slashdotting...
Hmmm
Tried it.. Had big framerate hit compared to beta 4620 so reverting back to old faithful (patched for 2.6 of course!).
Mike.
Hmm
Darn. I had that problem with 5323 and I kept using 4620. Strangely didn't happen with this one.
minion.de patches?
I wonder if NVIDIA used the minion.de patches?
They did. I found zanders ema
They did. I found zanders email in the Makefile.
No OpenGL 1.5
What?! Still no OpenGL 1.5 support?! C'mon nVidia, the OpenGL 1.5 specifications are almost 6 months old now!! Hell, even Mesa is OpenGL 1.5 compliant now.
Debian-User
There is the possibility to install a Patch directly into your kernel for NVidia-cards. Is that better than using this new driver???
Either Or
Makes no difference. It works identically.
Kernel?
You don't patch the kernel, ever. The patch is for the driver itself. The kernel driver is provided as source, since that's the only way to get the damn thing to install on everyone's individually tweaked kernels.
x86_64
I tried downloading the x86_64 driver (5332) and I cannot get this to compile with my 2.6 kernel.
I will have to try the IA32 version (5336).
mbest@harddata.com
re: x86_64?
You probably want the IA32 = Intel Architecture 32bit unless you have a fancy new 64bit processor (probably not). I should think that 99% of people will need the IA32 driver.
Remember that this driver is not GPL and so will NOT (ever) be part of the kernel. Nvidia produce a fairly decent driver but will not release enough info (in the form of datasheets / very technical info) to let the driver guys write a truly open-source driver, why not? Well that's a whole big debate...
It does seem to work pretty well for me tho.
ifor
re: x86_64
I do have an AMD64 3200 running Fedora Core 0.96
The IA32 one refused to run btw, wrong architecture, I was hoping it might run but it didn't.
mbest@harddata.com
x86_64
5332 still requires patching (see www.minion.de for patches and details). If you're using Gentoo AMD64 then don't worry about it; they'll automatically patch the driver when you emerge it. I have no experience with other distributions in 64-bit mode.
The IA32 won't run for you if you're running in x86_64 mode.
Reinstall every time I reboot??
Every time I reboot my system I have to reinstall the NVIDIA driver to get it to load. X fails otherwise. This is completely strange. Anyone else seeing this issue?
Slackware 9.1/kernel 2.6.1
Re: Reinstall every time I reboot?
The makefile modprobes the driver upon completion of the installation. It's likely that your reinstalling of the drivers is a not-so-quick fix to the driver not being inserted into the kernel yet at boot time. For lack of knowledge of your exact situation, I advise adding a line with "nvidia" to your /etc/modules or a similar file (it was "NVdriver" for 3xxx and prior versions).
Same prob with Slack 9.0 Kernel 2.6.2
Everytime I reboot its the same thing...failure.
Reinstall the patched NVIDIA binary and works like a charm.
I actually had this same problem with vmware 4. Every time I rebooted
I had to run vmware-config to re-install and reload the modules...odd.
I'm thinking this is just into relation to Slack.
Same prob with Debian Kernel 2.6.8
When I reboot, x crash and i have to reinstall the drivers to run x. It's %'$ to install drivers all time... Somoeone can help me ? I have add nvidia to /etc/modules but it does nothing.
reboot and UDEV fails fedore core 3
Hi there I have the latest nvidia driver for a geforce FX5500 on a amd barton 400fsb. Works fabulique once I exchange the xorg.conf for the xorg.conf.nvidia. basically rename xorg.conf to xorg.conf.sto and xorg.conf.nvidia to xorg.conf or simply remove the dri line and change the "nv" to "nvidia"(I think those are the only changes. Ok then it boots up great and nvidia flash screen pops up and I can play bzflag perfecto. However when I go to REBOOT the sucker UDEV fails just after giving an ok for the network, sound etc. It fails I think at the point of video.
So what I do is I load the previous non-nvidia kernal and it boots but complains about the "nvidia" version of xorg.conf so I just let it go to xconfig and set it as a vesa card resetting the screen to 1280 so I can see things well. I DO NOT REINSTALL nvidia all I do is go take out the xorg.conf vesa version and replace it again with the nvidia version boot again and it works.... UNTIL the next boot when i have to do it all again. WHY IS THIS SUCH A PROBLEM Linux is on the BRINK of perfection and there's still BS like this! WHAT IS CAUSING A CHANGE IN THE BOOT!!! IS IT REWRITING THE XORG.CONF OR WHAT!!? Help mois!!!!
I've fallen and I can't get up!
--Buck
Seemingly Fixed
ok with kernel 2.6.10-1.770 and the 7167 invidia driver already rpm'd for fedora core 3(and that kernel version) I had already installed and booted with that kernel and then ran all the installs for the rpm'd nvidia drivers...... Also I had reinstalled everything and did updates of all the software by hand so the auto-update didn't add any of the at updates(the ones that end in at) for the most part they seem to cause mega problems.
I have rebooted several times... sometimes I get the nvidia flash screen and others I don't but each time I get the full opengl nvidia drivers running great on my fx5500(which is a not bad card especially if you have a fast cpu). Anyhow don't forget to rename the xorg.conf.nvidia to xorg.conf.
btw I had compiled my ati computer drivers into the 9 kernal and when I upgraded they like didn't work so i recompiled the same driver into the kernal and it's like not as good fps or something.... that's on an ati8500 (old but still good). I will never spend like 300 for a video card sorry that's just a rip off.
Anyway these video card companies need to start FULLY supporting linux and that INCLUDES making it SMOOTH and EASY to install their freak drivers!
Anyway let me know if you all already figured it all out and I was the only one having problems....
Disappearing drivers: had the same problem with Debian Sarge
One of the nvidia startup scripts (can't remember which) were unlinking the modules, instead of trying to figure out why, I set the immutable flag on them and called it a day.
Outside of the warning that it couldn't unlink them, worked like a charm.
Solution
You have to choose the nvidia version 1.0-5336 (the new versions are bugged). Uninstall the over version (sh NV* --uninstall), install 1.0-5366, and add nvidia to /etc/modules (to load it at start up). Then load nvidia (modprobe nvidia) and run x.
reinstall every reboot
I found out the old nVidia boot-start config files in my system was the problem. I deinstalled my old nVidia drivers and removed /etc/init.d/nvidia-glx and /etc/init.d/nvidia-kernel and reinstalled the new nVidia-file. (sh NVIDIA-Linux-x86-1.0-8178-pkg1.run) Since then it works fine. I hope it helps someone. Good luck! :)
Debian, 2.6.15-1-k7
/Herde
reinstall every reboot *fixed*
Agreed. Uninstalling the driver and removing /etc/init.d/nvidia-glx and /etc/init.d/nvidia-kernel before reinstalling (allow nvidia-xconfig to modify your XF86Config-4/xorg.conf or manually disable loading of dri and GLCore while including load glx) worked out well both before and after restart, assuminging you remember to add nvidia to /etc/modules of course -- good call =)
Debian (testing), 2.6.8-2-i686
xserver-xorg 6.9.0.dfsg.1-4
IA32 nVidia driver 1.0-8178
-xix
Ok, but..
Works quite well, however I get piles of this every time I try to use anything 3d (including glxgears).
0: nvidia: trying to map 0xb7515000 to kernel space, but we're in an interrupt or holding a spinlock
I get this too with my brandnew GeforceFX5200
and I have absolutely no idea why. Didn't get it before with my Geforce2. Does anyone have an idea? Is this AcPI related? I can provide more info if needed...
installer doesn't install
new driver works good for me. But installer does not work, I have to go to usr/src/nv and do
make SYSSRC=/usr/src/linux-2.6.2-rc3 module
mv nvidia.ko /lib/modules/linux-2.6.2-rc3/kernel
modprobe nvidia (in startup)
is this just me? It's not a problem anyhoo...
linux
It only recognises /usr/src/linux so that has to be a symbolic link or the actual source to your running kernel unless you specify it manually.
Actually, that's incorrect; t
Actually, that's incorrect; the installer attempts to locate the kernel source tree through /lib/modules/$(uname -r)/build and falls back to /usr/src/linux if that fails. The installer also respects the --kernel-name and --kernel-source-path command line parameters, which allow manual specification of the kernel source path. SYSSRC is also respected.
Corrupt framebuffer
Heya
I installed these drivers on a new slack9.1 install (with kernel 2.4.22). I load the console framebuffer thing at startup, but when i try to switch back to a tty from X, the display is VERY broken. starting linuck with vga=normal works fine, as it should.
Any ideas how to fix fb? i hate 80x25. :(
fb + nvidia
i have exactly the same problem, fb works only if i use the same resolution and the same color depth in fb as X uses (my X server runs with 1280x1024 - 24bit color, so fb works only with vga=795)...
i think this is a problem of the nvidia-driver - i didn't find a solution till now, but i'd be happy if i found...
greetings
Corrupt framebuffer
I solved this, compiling the framebuffer suport inside de kernel, not as modules and compiled with only the vga16... so i could put vga=791 in my lilo and it's working fine now....
I am having this problen
Hi
So to solve the problem you remove the Vesa VGA, leaving only vga 16 selected, but did you use the option nvidia framebuffer or no?
thank's in advance
Install help
I see that ya'll are alot more experienced with Linux and nVidia than I am, but I am also new to Linux. I was wondering if ya'll could help me out!? I am trying to install the nVidia drivers on RedHat 2.4.20-8. The driver install keeps asking me for the proper kernel. I am pretty sure I have the source installed! I had to install from the gui and I made sure to install the package.
I have given the nVidia installer these options to try:
--kernel-source-path=/usr/src/linux-2.4
--kernel-source-path=/usr/src/linux-2.4.20-8 --kernel-source-path=/usr/src/linux-2.4/kernel --kernel-source-path=/usr/src/linux-2.4.20-8/kernel
They all return the same message:
ERROR: Unable to load the kernel module 'nvidia.o'. This is most likely because the kernel module was built using the wrong kernel source files. Please make sure you have installed the kernel source files for your kernel; on Red Hat Linux systems, for example, be sure you have the 'kernel-source' rpm installed. If you know the correct kernel source files are installed, you may specify the kernel source path with the '--kernel-source-path' commandline option.
Can ya'll help me out on this??
it's possible caused by module conflict
I seems you build nvidia kernel module OK with the kernel source, but the problem appear when loading it. The first thing usefull should be loading the module manually to see what error reported. use modprobe or something like. I met the same problem before, and then the reason is the compiled nvidia module conflict with the preinstalled module nvidiafb.ko, it seems they take the same pci address, after i recompile kernel without nvidiafb. it's OK.
HTH
it's possible caused by module conflict
It seems you build nvidia kernel module OK with the kernel source, but the problem appear when loading it. The first thing usefull should be loading the module manually to see what error reported. use modprobe or something like. I met the same problem before, and then the reason is the compiled nvidia module conflict with the preinstalled module nvidiafb.ko, it seems they take the same pci address, after i recompile kernel without nvidiafb. it's OK.
HTH
Help with Mandrake 10
I'm using a v2.6 kernel version (dunno which, I'm just a noob, it's the one that came with the Mandrake 10.0 Comunity Download version)
I've tried to install the drivers, but it always asks me to recompile something (probably the kernel, since it asks for it's source to work). The thing is that after I compile the 'drivers', everything blows... my sound card stops working (Realtek onboard, using ALSA 1.2.9), the internet stops working (trough LAN), most of the applications that are intenseley using the video cards (e.g. games) won't work, others work slooooooooooooooooooooooooooooooow
Can someone help me?
My email is spike@mymail.ro
reply
did you edit your /etc/X11/XF86Config or /etc/whatever it is/xorg.conf?
If not, then do it as root and under beside the driver in the video section it should say 'nv' or 'vesa' change it to 'nvidia'. Also the X server cannot be running while istalling the driver. (so switch to run level 3 by typing '/sbin/init 3' or '/sbin/telinit 3' as root.
Hope this helps...
mandrake 10 problems
i just downloaded the new drivers and when i switched to run mode 3 in the inittab file and ran the run file as root without x, setup started ok but then it said that if i had the kernel source installed then i had to specify it, because it couldn't find it. I need help w/ this people I'm definately a noob.
Mandrake 10 Problems
You do not have your kernel source files loaded.............go to add /update software, then add Kernel Source found under Kernel Development and install............then try again.......Don't worry, I'm also a newbee, and had this problem before.......
Problems with kernel 2.6.10
Hi. I have tryed the NVidia driver under Linux 2.6.7 and it worked fine.
The I installed Linux 2.6.10, built the driver again (on 2.6.10). When attempting to start X (Xorg) the system (probably the kernel) freezes.
Anyone here with the same problem? Can somebody give me some hints about what should be and what shouldn't be enabled in the kernel.
When I exit X my console is a chaos...
I've installed nvidia 1.0-6629 driver on my Mandrake, everything is OK (after some tuning of course) but I still have the following problem: when I switch to the console (Alt-F1 etc) screen is completelely a mess; (big bliniking characters etc); the same when I exit X.
Do you know how to correct this problem?
I've got the same problem
I've got the same problem (blinking characters etc...) with my debian testing. I've such problem with two custom kernel, built from the source from the debian unstable distribution (2.6.9, 2.6.10). With 2.6.7 and 2.6.8 I've no problem. The configuration file is the same, just a few changes in acpi (pnp, video). The nvidia driver is the same: 1.0-6629. Anyone know a solution?
solved blinking problem with kernel 2.6.9/2.6.10
The solution was posted by zander on http://www.nvnews.net/ forum:
Operating System Forum->NVIDIA Linux forum->1.0-6629 x86/x86-64 and Linux 2.6.10->message from zander(12-25-04, 08:56 AM).
i don't use nvidia driver
using nv.
--
Linux debian 2.6.11-rc2-bk9, gcc (GCC) 4.0.0 20050130 (experimental), Garnome 2.9.4.1
nvidia
iim wanting to patch the 2.6.11.8 kernel with the nvidia driver im using fedora core 2 after taking a break from it and i've forgotten alot so would anyone know where i could get a patch and be willing to tell me how to patch the kernel thanx
Nvidia 1.0-8774 driver reinstall all the time - SOLUTION
I noticed lot of people having the same problem of having to re-install the latest drivers each time, after a re-boot.
I set about finding the steps needed so I started from scratch installation of Ubuntu dapper 6.0.6. Heres the accurate sequence of steps, that should get your Ubuntu blazing with direct rendering support.
Dowloaded the nvidia 1.0-8774 IA32 driver from www.nvidia.com.
'NVIDIA-Linux-x86-1.0-8774-pkg1.run'
I enabled all the repositories from synaptic manager.
Then did re-load and got a whole bunch of updates.
Used synaptic manager to get the following packages:
linux-headers-2.6.15-27-386
linux-headers-2.6.15-27
linux-image-2.6.15-27-386
linux-image-386
linux-kernel-headers
linux-restricted-modules-common
linux-source-2.6.15
Uninstalled the nvidia-kernel-common package
I re-ran the following from the terminal:
sudo apt-get update
sudo apt-get dist-upgrade
Pressed CTRL+ALT+BACKSPACE to logout of the GUI.
Pressed CTRL+ALT+F1 to get to a terminal.
Logged in and stopped the GUI by typing:
sudo /etc/init.d/gdm stop
Then in my directory where I downloaded the nvidia driver I typed:
sudo sh -q
When installation was complete, I said "NO" to not modify the xorg file.
Then in the terminal typed:
sudo modprobe nvidia
sudo dpkg-reconfigure xserver-xorg
Accepted the default settings. Made the selection for "nvidia" driver.
In the modules part, I selected the defaults. Made sure the "dri" was not selected.
I checked the xorg file and made sure the modules part looked like this:
Section "Module"
Load "bitmap"
Load "dbe"
Load "ddc"
Load "extmod"
Load "freetype"
Load "glx"
Load "int10"
Load "record"
Load "type1"
Load "vbe"
EndSection
Re-started the GUI by typing this:
sudo /etc/gdm/gdm start
Got the NVIDIA logo after this ( HOOOrrrraaayyy!!) and checked that direct rendering was enabled by typing:
glxinfo
Tested the re-start and re-boot would not happen by
logging back into a terminal, and typing:
sudo /etc/init.d/gdm stop
sudo init 1
sudo init 5
This brought me back the nice NVIDIA logo and the Ubuntu welcome.
Ready to rock and roll!!!
Hope this helps everyone.
Probablly not all the steps are necessary, and I did a fair bit of re-installing attempts as I didnt have the linux-headers in the first place. There was also the problem of linux-source for my Ubuntu.
This told me what sources, headers I needed:
uname -r.
Now to clone the HD, so I can restore it eaisly :-).