GreenBandit

"Search and Ye Shall Find" By entering the command dmesg I can see that my card is running in PCI mode This happens with the Option "NvAgp" set to either "0" or "1". With it set to "2" or unset (i.e. defaults to "3", I get the black screen freeze. Running glxgears with the graphics window maximised (1024x768x24bpp) I get 146 FPS. Not very good for such a simple bit of OpenGL. I can get better with Tirtanium1.9 (with 1000s of polygons & Mb of textures) in W2K. I'm not yet ready to ditch Windows & use Linux all the time :-( Question - What are the commands to get the Quake3 demo to do a framerate test

Got MDK10 OE re-installed, used the new nVidia 61.06 driver & got Quake3 test demo running. I still get the same effect regarding AGP, namely Option "NVagp" "1" works (don't yet know if this is AGP or PCI mode), but without this line it black screen freezes. One of the threads from the nVidia Linux D/L page points to their forum http://www.nvnews.net/vbulletin/forumdispl...p?s=&forumid=14 One of the nVidia guys (Andy Mecham) lists a load of info that they want in order to help us, so I'm going to post my findings there (& here) once I've done testing. XBOB, you say your card was running in PCI mode at one point. Where did you get this info from ? Also, what are you using to benchmark the system framerate ?

I've just noticed that nVidia have released new IA32 Linux drivers, namely V61.06 rel 30 June 2004. Since I've just trashed my Linux installation, I'm going to reinstall everything & try the new drivers. I want to work out why the nVidia AGP driver works for me when I have a SiS741GX chipset that is not listed as being supported. I will post what I find back here, probably in the next 1 or 3 days.

Yes that was badly worded. What I want to do is run an OpenGL benchmark on my Linux PC, ideally Tirtanium V1.9 ( www.tirtanium.de ) as I have results obtained from Windows systems that go back about 4 years. I have results from Windows PCs taken with and without AGP support, and I want to get the same for my Linux PC. When I tried to run TirtaniumLINUX, it said that it was unable to find "libMesaGLU.so.3". I downloaded libMesaGLU.so.3, but when I tried to install it, Linux said that first it would have to remove the existing libMesaGLU.so.5 I don't know enough about Linux to know if a program (in this case Tirtanium) should be expected to use a later version of the library file that it was looking for. Question 1 - is the fact that Tirtanium would not run with libMesaGLU.so.5 when it wanted libMesaGLU.so.3 normal for a Linux program or have I got a system error ? Question 2 - Is it safe to remove libMesaGLU.so.5 and install libMesaGLU.so.3 or could this stop other OpenGL programs that I intend to install (e.g. Quake3) from running ?

Illogic-Al I've just trashed my MDK 10 installation while trying to read some OS9 (6809 not Mac) floppies. From a clean installation I will try again with the TirtaniumLINUX that says it wants the libMesaGLU.s0.3 file. I did find a copy of libMesaGLU.s0.3, but when I started to install it, I was informes that it would remove the later version already on my system. Question - should I need to remove the MDK version 5 of libMeasGLU that MDK 10 installed or if (there were no system glitches) should TirtaniumLINUX pick up and use a later version of the file that it it asking for (namely libMesaGLU.s0.3) ? xbob Your quote about NVagp numbers is spot on, "NVagp" "1" uses the nVidia AGP module. What I have not yet got my mind around is why with no NVagp line in my /etc/X11/XF86Config-4 things totally freeze/fail, but specifying "NVagp" "1" (i.e. saying use the nVidia module when nVidia say my 741GX chipset is not supported) does work ?

EVEN MORE SORTED. Adding the line to XF86Config-4 (see above post for details) Option "NvAGP" "0" totally disables AGP support. I have just managed to get some sort of AGP support for my SiS741GX chipset by instead adding the line Option "NvAGP" "1" which apparently uses "NVAGP". My TirtaniumLINUX file won't run cos it can't find a "libMesaGLU.so.3", so I can't use that benchmarking s/w (that is still beyond me unless anyone can help). Is it possible to find out if the video drivers and AGP support are working by using glxgears as a benchmark ?

:D SORTED. Just in case this helps anyone else, the problem was due to AGPGART incompatabilities. The solution was to edit the /etc/X11/XF86Config-4 file & add the line Option "NvAgp" "0" in the same section as the line Driver "nvidia" (this line read Driver "nv" prior to editing it as per the nVidia instructions). On my system, the relevant part of the XF86Config-4 file is now Driver "nvidia" Option "DPMS Option "NvAgp" "0" Endsection The rational behind this is:- Look at Appendix F of the nVidia README file (on the same page as where their driver is downloaded, currently http://www.nvidia.com/object/linux_display...2_1.0-5336.html If your chipset is not listed, try what I described above. Can anyone tell me how to use glxgears to benchmark the nVidia driver ? Also I have a file "TirtaniumLINUX". (the Linux version of Tirtanium 1.90). Can anyone tell me how to run it ?

bvc, I do see what you mean about the nv entries in my file & the nvidia entries in yours. I did a "modprobe nvidia" & it did not give an error or generate any other text, just went back to a healthy console prompt (sorry I'm not sure about correct terminology). The result of the last test, which was to delete the XFree96.0.log file before doing a startx with the nvidia drivers called up is that no XFree86.0.log file was there when rebooted. So I would agree with Ixthusdan's comment that X is not the problem. Here's a bit of info, during the boot up there was a line something like no config file generating one. Use Kcontrol as root to customise. I did run Kcontrol & got the error "KUniqueapplication: Can't determine DISPLAY. Aborting." I changed the mouse to a PS2 connected wheel mouse & went into the Control Centre to configure it. With the nvidia driver selected, startx caused a crash, but again, no XFree86.0.log file was created. So, when nvidia is the driver, no XFree86.0.log file is created ! Does tha fact that my linux PC does not have internet access affect me being able to " install X again from a server. In other words, get the x files seperate from your media.. " as suggested by Ixthusdan ? Could this be anything to do with the fact that Mandrake do not list the Elitegroup 741GX-m mobo as being supported ? Could this be a 741GX AGP driver problem ? Could it be worth retrying my nForce2 mobo ? Did I cover all the suggestions made ? Any other ideas ? (Thanks for the help so far)

In answer to am I loading the kernel module, the answer is not knowingly. If it doesn't happen automatically when I start Linux, then its not happening. The first time I installed MDK10, and the nVidia drivers including the Load "glx" and Driver "nvidia" mods to XF86Config-4, "startx" gave a total freeze. That time re-editing XF86Config-4 back to its original (###Load "glx" & Driver "nv") would not then let me start KDE. The other funny was the garbage in the middle of the XFree86.0.log file. MDK 10 was re-installed, and the nVidia drivers were also re-installed after checking their MD5SUM. MKD10 was set up to not auto login & not start the GUI, so that I could login as "root" & use bash commands. The XFree86.0.log file that was last posted was copied to floppy after a reboot, before the XF86Config-4 was re-edited to allow me to start KDE. I'm going to go back, use a standard serial port mouse, delete the existing XFree86.0.log file, edit XF86Config-4 to use the nVidia driver, and after the crash, use a TOMSRTBT linux floppy to copy the new XFree86.0.log. If I have taken onboard what people have said, this should make sure it is the required log file from when . I'm afraid I don't understand the comment " get the x files seperate from your media.. ". I am fairly sure the installation CDs are good because of earlier attempts with an nForce2 mobo.

I've made some progress, but not there yet. The PC is not overclocked, & the W98SE system is able to run OpenGL & D3d demos like Tirtanium. I set the XF86Config-4 file back to original, but the "startx" command still gave a total PC freeze, just as it did when using the nVidia driver. I downloaded the nVidia driver again. It had the same MD5SUM as the one used for the last install, but I had to check. I reinstalled MDK 10 OE onto its own 40Gb drive (hdb) with partitioning based on MDKs /usr option but with /usr set to about 10Gb (so I could install everything to ensure that the kernel would recompile). I entered "urpmi binutils", then "urpmi kernel-source", but in both cases I was told everything already installed (I understand why now). I then installed the nVidia driver, it recompiled the kernel, editing XF8Config-4, and typing in "startx". It crashed & froze. At least this time editing the XF86Config file back to standard allows KDE to run again AND the XFree86.0.log file is not corrupted. Does this attached log file give any clues ? XFree86.0.log

I can't boot in to the GUI, so I don't know how to get to any GUI editor The XFree86.0.log file is attached to this post - it was formatted better when viewed on the linux machine using "cat", but on this W98 machine it is not very clear. Does it give any clues as to the problem ? I have attached the file, but can't see it on the preview, so here is the end of it (posting it all exceed the no of emoticons allowed ) (II) Loading extension GLX (II) LoadModule: "nvidia" (II) Loading /usr/X11R6/lib/modules/drivers/nvidia_drv.o (II) Module nvidia: vendor="NVIDIA Corporation" compiled for 4.0.2, module version = 1.0.5336 Module class: XFree86 Video Driver (II) LoadModule: "mouse" (II) Loading /usr/X11R6/lib/modules/input/mouse_drv.o (II) Module mouse: vendor="The XFree86 Project" compiled for 4.3.0.1, module version = 1.0.0 Module class: XFree86 XInput Driver ABI class: XFree86 XInput driver, version 0.4 (II) v4l driver for Video4Linux (II) NVIDIA XFree86 Driver 1.0-5336 Wed Jan 14 18:30:46 PST 2004 (II) NVIDIA Unified Driver for all NVIDIA GPUs (II) Primary Device is: PCI 01:00:0 (--) Assigning device section with no busID to primary device (--) Chipset NVIDIA GPU found (II) resource ranges after xf86ClaimFixedResources() call: [0] -1 0 0xffe00000 - 0xffffffff (0x200000) MX(B) [1] -1 0 0x00100000 - 0x3fffffff (0x3ff00000) MXE(B) [2] -1 0 0x000f0000 - 0x000fffff (0x10000) MX [3] -1 0 0x000c0000 - 0x000effff (0x30000) MX [4] -1 0 0x00000000 - 0x0009ffff (0xa0000) MX [5] -1 0 0xe3003000 - 0xe3003fff (0x1000) MX [6] -1 0 0xe3002000 - 0xe3002fff (0x1000) MX [7] -1 0 0xe3001000 - 0xe3001fff (0x1000) MX [8] -1 0 0xe3000000 - 0xe3000fff (0x1000) MX [9] -1 0 0xe3004000 - 0xe3004fff (0x1000) MX [10] -1 0 0xd0000000 - 0xcfffffff (0x0) MXO [11] -1 0 0xdc000000 - 0xdc07ffff (0x80000) MX(B) [12] -1 0 0xd8000000 - 0xdbffffff (0x4000000) MX(B) [13] -1 0 0xe0000000 - 0xe0ffffff (0x1000000) MX(B) [14] -1 0 0x0000ffff - 0x0000ffff (0x1) IX [15] -1 0 0x00000000 - 0x000000ff (0x100) IX [16] -1 0 0x0000e400 - 0x0000e43f (0x40) IX [17] -1 0 0x0000e000 - 0x0000e0ff (0x100) IX [18] -1 0 0x00004000 - 0x0000400f (0x10) IX (II) resource ranges after probing: [0] -1 0 0xffe00000 - 0xffffffff (0x200000) MX(B) [1] -1 0 0x00100000 - 0x3fffffff (0x3ff00000) MXE(B) [2] -1 0 0x000f0000 - 0x000fffff (0x10000) MX [3] -1 0 0x000c0000 - 0x000effff (0x30000) MX [4] XFree86.0.log

Unfortunately there are many lines that briefly appear befor the screen goes black, but they are too fast to see. I tries slowing down the FSB but it still did not let me read what appeared before the screen went black, it was that fast. Is there a log file that will give the required info, or another way forward ?

Knoppix 3.4 rel 17/May/2004 has failed me, even with 2.4 & 2.6 kernels on freshly formatted FAT32 partitions& so has MDK move. I can't get this system to boot reliably to any Linux that I have tried so far so the RPM kernel2.6.7 is not yet an option. I now have information overload & I am really gratefull to all that have been helping me, one day I hope I am able to help others. I'm not one of the lucky ones where X fails & Y works, so I have a lot of forum information to wade through before I hassle people for more information. I want to find a MDK patch that gets the nForce2 system going, if not I will use my 550MHz Jetway J542B system for Linux (already proved OK with MKD10CE) & use my 2000+ nForce2 system for my W98SE based Office system until MDK10.1 is released. I do look on the bright side - I will end up not using Windows - somehow ! I'll probably go quite for a while now while I digest all this new info, but I'll be watching this space in case new suggestions arrive - my email notification is enabled. B)

Just used W98 FDISK & FORMAT to clean up buth HDDs. Then tried Knoppix 3.4 rel 17/may/04. I got about 5 seconds of working desktop before it froze. The command knoppix26 just lead to a blank black screen. Reformatted the drive, & tried MDKMove2, it also gave a blank black screen. MDKmove(dated 03/Dec/03) loaded up OK, but when I try to run Control Center or HardDrake, I get a few seconds of the message "detection in progress", then it just dissapears. On the 4th attempt, HardDrake let me create several partitions that (I stayed with its choices), but it doesn't report that a partiton is formated after doing so. I tried the nForce2 link given, but couldn't get through. Any more ideas from anyone

I have now also tried disabling loads of stuff in the Bios including UDMA access to HDD & CD/DVD. I haven't got RAID or SATA & there is no CPU disconnect setting in the BIOS (I've also read something about that causing a problem). I've d/l & tried MDK10Move2, but after showing the Welcome screen and the desktop background, it does nothing else, no toolbar or any buttons. MDK10Move lets me get a bit further. I am able to get to a working desktop and launch HardDrake2 v9.1.1, but when select disk hda then click on the "run config tool" button, the config tool does not run. If I boot the system, then go to Control Centre > Hardware > DiskDrake, but that locks up with "detection in process" on the screen. Occasionally I get a bit further. I did manage to get 3 partitions yesterday (as per my post June 01 0900), but then MDK10 OE locked up during the installation of packages, but at least it got past the selection of which packages to install. I know my partition selection was not ideal, but surely it would at least allow an installation ? I can't get MDK10 Move to work for long enough to get the 4 partitions suggested by Ixthusdan. How big should / /home /usr & /swap be ? (hda=40Gb hdb=40Gb) ? I don't think it is formatting itself that is the problem, it seems to be any type of sustained activity :-(