Go Back   nV News Forums > Linux Support Forums > NVIDIA Linux

Newegg Daily Deals

Reply
 
Thread Tools
Old 11-08-03, 11:02 AM   #13
utidjian
Registered User
 
Join Date: Nov 2003
Location: Wayne, NJ
Posts: 4
Send a message via AIM to utidjian
Default

I do not agree with your method for getting Nvidia drivers to work with FC1.

I installed FC1 last night and fiddled around with various options before I got the Nvidia drivers working to my satisfaction.

A couple of points/disagreements:

* You do NOT need to recompile your kernel to get Nvidia drivers working. If you need the NTFS module then, yes, you do need to recompile the kernel... otherwise no.

* You should NOT recompile the kernel with gcc-3.3.2. There are sound reasons why the FC1 devel team did not use gcc-3.3.2 for the kernel build compiler. For a discussion on this see:
http://www.redhat.com/archives/fedor.../msg00676.html
I am sure you will see plenty more on this topic as the days go by.

* You do NOT need to edit /etc/inittab (see below).

* You method, while I haven't tested it, seems overly complicated.

This is what I did (all root commands are prepended with a # sign):

0. Download the file NVIDIA-Linux-x86-1.0-4496-pkg2.run from
http://download.nvidia.com/XFree86/L...-4496-pkg2.run
and put it somewhere useful (/tmp or /home/downloads or whatever).

1. Logout of the system then Ctrl-Alt-F1 to get to a virtual console and log in as root.

2. Switch runlevels from 5 to 3 this will stop X (as root):

# telinit 3

It will take a few seconds as things shutdown/restart.
Hit [Enter] if you don't get a prompt back right away.
This will stop X "gracefully". X will restart automatically on reboot OR the command startx OR the command telinit 5.

3. Set the C compiler environment variable to gcc32 (it is currently gcc by default).

# export CC=gcc32

4. Remove the XFree86-Mesa-libGL package:

# rpm -e --nodeps XFree86-Mesa-libGL

5. Now run the Nvidia driver installer as usual:

# cd /path/to/where/you/put/it
# sh NVIDIA-Linux-x86-1.0-4496-pkg2.run

Answer the usual questions and it should build and install correctly.

6. Make a backup copy of your "known good" /etc/X11/XF86Config file:

# cd /etc/X11
# cp XF86Config XF86Config.bak

7. Edit your XF86Config file so that it will use the nvidia driver rather than the stock nv driver. This will be in 'Section "Device"' and there will be a line like this:

Driver "nv"

Change that line to:

Driver "nvidia"

You may also need to add and/or comment out some options in the 'Section "Module"' part of that file. For details on this see the README file which should be in /usr/share/doc/NVIDIA_GLX-1.0/README. Once you are finished editing the XF86Config file save it.

8. Test the new driver:

# startx

Which should start X. If X starts up OK try opening a terminal and type in the command:

# glxgears

If that works fine then you are done installing the driver.

9. Logout of X. You should be abck at a command prompt. Then restart the GUI login for X:

# telinit 5

The GUI login for X should start up.

10. One last thing... you left a virtual console logged in as root. You propably don't want to forget and leave that open so....

Ctrl-Alt-F1
# exit
Alt-F7

Will log you out as root and return you to the GUI login.


Note: The above method is really pretty simple. There are only two added steps from the method one would use in previous versions of Red Hat 9 or earlier... they are steps 3 and 4. If you have installed Nvidia drivers in Red Hat (or other distros) before then you will already be familiar with most of this. Even if you are familiar with the Nvidia Linux drivers you should read the README and other docs because there are a lot of very cool features you may want to try out. This feature set changes with each release of the drivers.
__________________
-DU-...etc...
utidjian is offline   Reply With Quote
Old 11-08-03, 03:44 PM   #14
smthmlk
Registered User
 
Join Date: Nov 2003
Posts: 38
Default

heh

well, that howto is the only way I could get it to install for me. In severn i did not have to recompile the kernel, but for yarrow the driver would not compile or install with any method unless the kernel was compiled with the latest gcc.

this howto wasn't written out of "this sounds like it should work", its from actual trial and error. If you don't like it, don't use it I had tried the export CC=gcc32 (which is essentially what Im doing by adding the line to the 1 Makefile) .. no luck. Same with IGNORE_CC_MISMATCH

Last edited by smthmlk; 11-08-03 at 03:50 PM.
smthmlk is offline   Reply With Quote
Old 11-08-03, 05:25 PM   #15
utidjian
Registered User
 
Join Date: Nov 2003
Location: Wayne, NJ
Posts: 4
Send a message via AIM to utidjian
Default

Well no worries here I certainly will not be using your method... already got it working.

It is interesting that you couldn't get it working with the export CC=gcc32. Did you type that in the exact same shell as you ran the installer? Also did you recompile your kernel *before* trying to use the export/install commands the first time? That is to say, did you already have a custom built kernel installed that was compiled against gcc (not gcc32)? If so, that can sometimes foul up subsequent module builds. I ran the installer against a "virgin" FC1 install.

In any case... good you got it working.
__________________
-DU-...etc...
utidjian is offline   Reply With Quote
Old 11-08-03, 11:13 PM   #16
smthmlk
Registered User
 
Join Date: Nov 2003
Posts: 38
Default

i had the stock kernel. export CC=gcc32 yielded the exact same behavior from the nvidia installer when compiling the kernel interface: errors i went ahead and tried to do the manual make install, errors compiling nvidia.o.

i gave up on it for a bit, expecting it to work since it had worked with severn w/ the same driver. I went and recompiled the kernel to include ntfs support and removed a couple of uneeded things (bluetooth, irda stuff), came back later and tried the nvidia installer again, and out of 'desperation' went ahead and followed the severn procedure in my that howto, and it worked. games work great, and ive tested it with quake3 and ut2k3.

smthmlk
smthmlk is offline   Reply With Quote
Old 11-09-03, 07:31 AM   #17
gemi
Registered User
 
Join Date: May 2003
Location: Zurich
Posts: 2
Default

I have the NVIDIA driver 4496 up and running on Fedora 1. It works mostly, but I have the following issues:
On RH9 I used Xinerama instead of TwinView. Now on Fedora this results in these problems. I get b/w and unanimated cursors. GNOME/GTK fonts are right, but Qt/KDE fonts are non-antialiased. I switched to TwinView and these problems disappear.
gemi is offline   Reply With Quote
Old 11-09-03, 03:24 PM   #18
dpw2atox
Registered User
 
Join Date: May 2003
Posts: 170
Default

the stability issues i noticed was X would randomly lockup and freeze my whole system. I currently use winex 3.1 to run some windows games on my pc and with fedora's build of X i get funny colors on guys in warcraft3 and the game crashes within 10min. I installed RH9's version of X and all my problems are gone.
Im using 44.96 on my Gainward 5600.
dpw2atox is offline   Reply With Quote
Old 11-09-03, 06:18 PM   #19
LordMorgul
Electrical Engineer
 
LordMorgul's Avatar
 
Join Date: Dec 2002
Location: San Luis Obispo, CA
Posts: 872
Default

Following the how-to above would probably work fine even if the gcc 3.2 package wasn't installed, due to recompiling the kernel. Check that you actually have this package installed before trying to use the CC=gcc32 trick.

rpm -q gcc32

if you don't have it this will of course not work. It is still highly advised to compile your kernel with gcc32 if you're building a 2.4 version kernel (so you want this package installed).
__________________
"..the triumph of evil is for good men to do nothing." (Edmond Burke)
nVIDIA video driver RPMs for Fedora :: see yum repo at livna.org.
LordMorgul is offline   Reply With Quote
Old 11-09-03, 10:36 PM   #20
utidjian
Registered User
 
Join Date: Nov 2003
Location: Wayne, NJ
Posts: 4
Send a message via AIM to utidjian
Default

What gets me is... in smthmlks method:

He compiles the kernel with gcc-3.2.3
Runs the installer... short circuits the installer...
Builds the nvidia module with gcc-3.2.3
And it works fine for him.

I don't see why he would need to rebuild the kernel if it was already built with gcc-3.2.3. If you look in the kernel top-level Makefile it specifies gcc32 (aka gcc-3.2.3)

One can check all this with:

gcc -v
gcc32 -v
cat /proc/version

You will see that the kernel was built on gcc32.
The method I outlined worked fine for me. I installed the nvidia drivers on a "virgin" FC1 install. I get a glxgears of approx 2600 FPS on a nVidia GeForce 2 GTS 32M card in a Tyan Tiger 133 with dual PIII 800MHz CPUs and 512M of RAM. I should also add that Quake3 seems smoother and quicker on FC1 than it was on RH9.

Unfortunately... people seem to be getting variable results. Some work with one method and some with others... and some not at all. There doesn't seem to be a tried and true method that will work in all cases. I guess the best thing is to keep trying the different methods until you find one that works for you.

For those of us that have been using Linux for a while and want to try the latest distro... these sorts of problems come with the territory. The problems are far fewer and easier to fix than they used to be.
I am sure that FC2 will be more polished than FC1. If you must run certain apps and games that rely on accelerated 3D stick with RH9 (FC2 will be out by the time that support for RH9 runs out). Or just keep plugging away at the problem.
__________________
-DU-...etc...
utidjian is offline   Reply With Quote

Old 11-10-03, 01:20 AM   #21
Zubir
Registered User
 
Join Date: Oct 2003
Location: Athens, GA, USA
Posts: 17
Unhappy loaded, but...

Quote:
Originally posted by utidjian
I do not agree with your method for getting Nvidia drivers to work with FC1.

I installed FC1 last night and fiddled around with various options before I got the Nvidia drivers working to my satisfaction.

A couple of points/disagreements:

* You do NOT need to recompile your kernel to get Nvidia drivers working. If you need the NTFS module then, yes, you do need to recompile the kernel... otherwise no.

* You should NOT recompile the kernel with gcc-3.3.2. There are sound reasons why the FC1 devel team did not use gcc-3.3.2 for the kernel build compiler. For a discussion on this see:
http://www.redhat.com/archives/fedor.../msg00676.html
I am sure you will see plenty more on this topic as the days go by.

* You do NOT need to edit /etc/inittab (see below).

* You method, while I haven't tested it, seems overly complicated.

This is what I did (all root commands are prepended with a # sign):

0. Download the file NVIDIA-Linux-x86-1.0-4496-pkg2.run from
http://download.nvidia.com/XFree86/L...-4496-pkg2.run
and put it somewhere useful (/tmp or /home/downloads or whatever).

1. Logout of the system then Ctrl-Alt-F1 to get to a virtual console and log in as root.

2. Switch runlevels from 5 to 3 this will stop X (as root):

# telinit 3

It will take a few seconds as things shutdown/restart.
Hit [Enter] if you don't get a prompt back right away.
This will stop X "gracefully". X will restart automatically on reboot OR the command startx OR the command telinit 5.

3. Set the C compiler environment variable to gcc32 (it is currently gcc by default).

# export CC=gcc32

4. Remove the XFree86-Mesa-libGL package:

# rpm -e --nodeps XFree86-Mesa-libGL

5. Now run the Nvidia driver installer as usual:

# cd /path/to/where/you/put/it
# sh NVIDIA-Linux-x86-1.0-4496-pkg2.run

Answer the usual questions and it should build and install correctly.

6. Make a backup copy of your "known good" /etc/X11/XF86Config file:

# cd /etc/X11
# cp XF86Config XF86Config.bak

7. Edit your XF86Config file so that it will use the nvidia driver rather than the stock nv driver. This will be in 'Section "Device"' and there will be a line like this:

Driver "nv"

Change that line to:

Driver "nvidia"

You may also need to add and/or comment out some options in the 'Section "Module"' part of that file. For details on this see the README file which should be in /usr/share/doc/NVIDIA_GLX-1.0/README. Once you are finished editing the XF86Config file save it.

8. Test the new driver:

# startx

Which should start X. If X starts up OK try opening a terminal and type in the command:

# glxgears

If that works fine then you are done installing the driver.

9. Logout of X. You should be abck at a command prompt. Then restart the GUI login for X:

# telinit 5

The GUI login for X should start up.

10. One last thing... you left a virtual console logged in as root. You propably don't want to forget and leave that open so....

Ctrl-Alt-F1
# exit
Alt-F7

Will log you out as root and return you to the GUI login.


Note: The above method is really pretty simple. There are only two added steps from the method one would use in previous versions of Red Hat 9 or earlier... they are steps 3 and 4. If you have installed Nvidia drivers in Red Hat (or other distros) before then you will already be familiar with most of this. Even if you are familiar with the Nvidia Linux drivers you should read the README and other docs because there are a lot of very cool features you may want to try out. This feature set changes with each release of the drivers.
this worked for me. make sure you DON'T update the compatgcc lib before you install the driver, or you'll get the load error.

the problem i'm having is that while chromium and tux racer run well, wolfenstein enemy territory only shows a black screen. the game is running, i can hear the sound, but the screen is black. this is the first time i've installed an nvidia card under linux, so any help would be appreciated.
__________________
I'd rather have a bottle in front of me than a frontal lobotomy.

AMD Athlon 64 3000+ | MSI K8T-NEO FSR | 2 X 512MB Kingston Hyper-X PC4000 | BFG 6800/OC 128 DDR | Samsung SP1213C 120GB SATA | Lite-on LDW-411S 4X DVD +/- RW | Aspire 550W PSU
Zubir is offline   Reply With Quote
Old 11-10-03, 07:08 PM   #22
LordMorgul
Electrical Engineer
 
LordMorgul's Avatar
 
Join Date: Dec 2002
Location: San Luis Obispo, CA
Posts: 872
Default

Probably Mesa conflicts, posted on the 4496 release thread following your post.

http://www.nvnews.net/vbulletin/show...6&pagenumber=9

do try to keep things together, gets very confusing
__________________
"..the triumph of evil is for good men to do nothing." (Edmond Burke)
nVIDIA video driver RPMs for Fedora :: see yum repo at livna.org.
LordMorgul is offline   Reply With Quote
Old 11-14-03, 10:07 AM   #23
skipperczt
Registered User
 
Join Date: Nov 2003
Posts: 1
Default

I typed in export IGNORE_CC_MISMATCH=y before I did sh NVIDIA... didn't have any problems.
skipperczt is offline   Reply With Quote
Old 11-14-03, 01:08 PM   #24
ByteWarrior
Registered User
 
Join Date: Nov 2003
Location: Belgium
Posts: 21
Default SOS!

Well, I just did a fresh install of Fedora Core 1 and tried to install the nVidia drivers with a couple of methods mentionned in this thread.

First, I tried smthmlk's method. That didn't work... The installer says that it didn't found a precompiled header.

Then I tried utidjian's method. Again, no success. The installer keeps saying that there is no precompiled header found...

What should I do? Do I have to recompile the kernel or not? Are there any other methods that I should try?

(For the record: the computer has no internetconnection available.)

You can find all my logs on: http://bytewarrior.madoka.be/linux/

I hope we will find a solution for the installer in the (very) near future...
__________________
AMD Athlon 2400+ 256 MB RAM
ASUS A7N8X with GeForce4 400 MX AGP 8X
HD 60 GB
RedHat 9 (Shrike) - kernel 2.4.20
My system logs
My qualification: Linux beginner
ByteWarrior is offline   Reply With Quote
Reply


Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Forum Jump


Similar Threads
Thread Thread Starter Forum Replies Last Post
Canonical partners with Microsoft to support Ubuntu on Azure News Archived News Items 0 06-08-12 06:20 AM
Remastered Doom 3 BFG Edition Game Coming with Stereo 3D Support News Archived News Items 0 06-01-12 06:30 AM
Intel's Ivy Bridge Core i7 3770K Overheating Issue Detailed News Archived News Items 0 05-16-12 11:40 AM
CPUMark99 - how do you compare fuelrod Benchmarking And Overclocking 66 07-19-11 09:32 AM
Official Detonator 40.xx Feedback Thread pelly NVIDIA Windows Graphics Drivers 299 09-22-02 07:54 AM

All times are GMT -5. The time now is 04:42 AM.


Powered by vBulletin® Version 3.7.1
Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Copyright 1998 - 2014, nV News.