Trying to get NVIDIA GEorce256 DVI to work

Hi,

I just installed Linux on my old PC and I would like to share my new
monitor between this old PC and my new one (new monitor has a DVI &
VGA connector). My new PC has only a VGA connector, so that leaves
DVI for the old one. The old PC has an add on NVIDIA GeForce256 card
(with a whopping 32 MB of RAM). It has two connectors (VGA & DVI). I
recall there is also on-board graphics with a VGA connector.

I connected a spare monitor that has both VGA & DVI connectors to the
old PC. VGA works fine. When I disconnect VGA cable & connect DVI I
get no video and can't even tell if Linux booted or not (something is
running 'cause the caps lock button lights the light).

I doubt this is related, but...since the old PC essentially has two
graphics cards, I looked in the bios thinking that could be the
issue. I found a setting that allowed me to select AGP or PCI as my
video card. I assumed this was what the BIOS POST would use.
Interestingly, I tried both settings and POST stuff only came out the
NVIDIA VGA card.

Thanks for your help,


Jim

Jim wrote:
Hi,

I just installed Linux on my old PC and I would like to share my new
monitor between this old PC and my new one (new monitor has a DVI &
VGA connector). My new PC has only a VGA connector, so that leaves
DVI for the old one. The old PC has an add on NVIDIA GeForce256 card
(with a whopping 32 MB of RAM). It has two connectors (VGA & DVI). I
recall there is also on-board graphics with a VGA connector.

I connected a spare monitor that has both VGA & DVI connectors to the
old PC. VGA works fine. When I disconnect VGA cable & connect DVI I
get no video and can't even tell if Linux booted or not (something is
running 'cause the caps lock button lights the light).

I doubt this is related, but...since the old PC essentially has two
graphics cards, I looked in the bios thinking that could be the
issue. I found a setting that allowed me to select AGP or PCI as my
video card. I assumed this was what the BIOS POST would use.
Interestingly, I tried both settings and POST stuff only came out the
NVIDIA VGA card.

Thanks for your help,


Jim

Old_PC ------ Build_In_Graphics ----- VGA ???
(Linux)
------ GeForce256 ------------ VGA (Works)
AGP ??? ------------ DVI (No video)

Linux runs XWindows. There is a configuration file. It is pretty
complicated. You may have to look in there, because I don't recollect
any fancy GUI for looking into this stuff. (Some Linux distro,
may have a GUI for it, but not all of them.)

On some old computers, if you place an AGP video card in the
computer, that automatically disables the Build_In_Graphics at
the hardware level. Then, an OS like Linux, is only going to see
the GeForce256. If the card was plugged into a PCI slot, it
could be a tossup, as to whether the Build_In_Graphics
remains detectable and operational.

The BIOS setting "Primary display" [AGP, PCI, Built_In),
controls which device is used first. Perhaps the BIOS is
displayed on the Primary display for example. But then,
it might be a function of what driver is installed in the
OS, as to what card or cards are used. But if your AGP
slot causes the Built_In to be disabled, then there
aren't very many choices.

From your description, you also have a total of two monitors,
and you can connect one of the monitors to the 256_VGA and
the other monitor to the 256_DVI connector and test. So you have
more equipment to work with, than a lot of people do when
they're debugging a situation like this.

This is probably exactly the wrong article to be giving, but
may give you some ideas. At least, to test.

http://en.wikipedia.org/wiki/Xinerama

Perhaps between the contents of "dmesg" and running the command
"lspci", you can get some idea what hardware is being seen by
Linux.

I have Ubuntu running in a virtual machine, and it has
something called "gnome-display-properties", which looks
like an attempt at a Display control panel. But because
that virtual machine only has one video device, it can't
display what would happen with a more complicated setup.

http://people.gnome.org/~federico/mi...ties-thumb.png

http://people.gnome.org/~federico/ne...properties.png

For the screens to be detected, either the video card has
impedance sensing (unlikely on a Geforce256), or the
video card uses DDC serial interface on each connector,
to probe for a monitor.

And a Linux newsgroup is more likely to have XWindows
experts in it.

Paul

...

The BIOS setting "Primary display" [AGP, PCI, Built_In),
controls which device is used first. Perhaps the BIOS is
displayed on the Primary display for example. But then,
it might be a function of what driver is installed in the
OS, as to what card or cards are used. But if your AGP
slot causes the Built_In to be disabled, then there
aren't very many choices.

*From your description, you also have a total of two monitors,
and you can connect one of the monitors to the 256_VGA and
the other monitor to the 256_DVI connector and test. So you have
more equipment to work with, than a lot of people do when
they're debugging a situation like this.

This is probably exactly the wrong article to be giving, but
may give you some ideas. At least, to test.

http://en.wikipedia.org/wiki/Xinerama

Perhaps between the contents of "dmesg" and running the command
"lspci", you can get some idea what hardware is being seen by
Linux.

I have Ubuntu running in a virtual machine, and it has
something called "gnome-display-properties", which looks
like an attempt at a Display control panel. But because
that virtual machine only has one video device, it can't
display what would happen with a more complicated setup.

http://people.gnome.org/~federico/mi...operties-thumb...

http://people.gnome.org/~federico/ne...7-gnome-displa...

For the screens to be detected, either the video card has
impedance sensing (unlikely on a Geforce256), or the
video card uses DDC serial interface on each connector,
to probe for a monitor.

And a Linux newsgroup is more likely to have XWindows
experts in it.

* * Paul
Paul,

I didn't think you could have VGA and DVI connected at the same time.
I would assume the graphics card would only use one--probably the
first one it felt a monitor was connected to it. But, I tried
connecting both and I got the same symptom--no video and unable to
tell if it booted.

NEW INFO: I realized I have an ethernet card in that PC. I connected
it my hub and tried to ping the PC after waiting long enough for Linux
to boot. I couldn't, so it looks like Linux isn't booting. I just
checked /var/log/messages and sure enough there's about 15 minutes
between shutdown and startup. That deadtime is probably when I tried
to boot via DVI. So, it's hardware, the BIOS or grub. Any ideas? Is
there any way to get grub to log something to the hard drive?

Thanks much,


Jim

Jim wrote:


I didn't think you could have VGA and DVI connected at the same time.
I would assume the graphics card would only use one--probably the
first one it felt a monitor was connected to it. But, I tried
connecting both and I got the same symptom--no video and unable to
tell if it booted.

NEW INFO: I realized I have an ethernet card in that PC. I connected
it my hub and tried to ping the PC after waiting long enough for Linux
to boot. I couldn't, so it looks like Linux isn't booting. I just
checked /var/log/messages and sure enough there's about 15 minutes
between shutdown and startup. That deadtime is probably when I tried
to boot via DVI. So, it's hardware, the BIOS or grub. Any ideas? Is
there any way to get grub to log something to the hard drive?

Thanks much,


Jim


I think you're right. The Geforce256 isn't a dual head GPU. In the multi-monitor
configurations here, it looks like they may only be using one monitor with
the card.

http://www.realtimesoft.com/multimon...ype=SearchCard

In this review article on the LeadTek version of the card,
there is a "selector" box for output options. I gather that
is how you go about selecting the output device (i.e. not dual
head, just 1-of-N connectors). The LeadTek has VGA (the default)
plus S-Video. The S-Video is created by a separate chip, translating
some digital interface on the GPU, to S-Video. Your card with
DVI, is using the "flat panel option", and there would be
an external chip to convert to TMDS (a TMDS transmitter of
some sort). TMDS is the signaling method used on DVI.
The "flat panel option" has a limit of 1600x1200 for resolution.

http://www.xbitlabs.com/articles/vid...rce256ddr.html

Perhaps what you need, to add to your Linux, is a tainted MVidia
driver, assuming one can be found that supports Geforce256. Like
all drivers these days, it's hard to find something that supports
them all. Nvidia and ATI, both drop support for older hardware,
in their latest drivers, so seem to be in no mood to embrace the
Linux notion of supporting old hardware. If Nvidia and ATI don't support
it, then you use the default driver that comes with the OS.
The potential difference might be, that the Nvidia driver would
know about the selection function, whereas the default driver
might just use the "primary" connector.

So somehow, you need the equivalent of that LeadTek custom display
control panel, with its option to select an alternate connector
on the faceplate.

Now, a while back, someone had a video card problem, and I tried to
investigate what it would take, to set up a good test environment for
such a purpose. I eventually found a distro (several years old), where
there was a "single button" in the interface of the distro, to install
the tainted Nvidia or ATI driver. The purpose of doing so, was to get
as much acceleration from the video card as possible. The repository
for that particular distro was shut down, so you couldn't install
any software from the repository. I was able to use "glxgears" and
note it was working faster, but I decided that the method I was pursuing,
was too expensive time-wise, for anyone to care about it. Every
time I've needed to modify anything involving graphics in Linux,
it takes bags and bags of time.

*******

I can barely get grub to behave at the best of times. I wouldn't
be the right guy to ask about logging from grub :-)

I like some of my older Knoppix CDs for testing, because the
boot sequence is written to the screen. No covering graphics to
hide the text output.

I'd either:

1) Test with LiveCD discs. Preferably, something verbose, that leaves
lots of text on the screen. My favorite would be Knoppix 5.3.1 DVD
version (but that won't boot on a CD drive), while there is a remaster
done by some people in Japan, that fits 5.3.1 on a CD. On occasion, I've
used that Knoppix DVD for testing overclocked systems, and you can
watch the boot sequence crash, if the thing is really unstable. Or
watch icons disappear from the desktop, seconds after the system has
booted.

2) Boot and bring up to level 2. On Knoppis, that would be
something like "knoppix 2" and that puts you in text mode.
That would be handy, if you needed to edit the XWindows configuration
file by hand. It would also be a test of whether booting part way,
and not altering the graphics operations, works any better.

I thought your boot log was effectively "dmesg", but I don't know
at what point it is flushed. I just looked at the /var/log/dmesg file
on my Ubuntu virtual machine, and that looks to contain the boot
messages from the last bootup. There is an executable "dmesg" that
also echoes the contents of that file to the screen.

Paul

Am 17.08.2010 00:31, * Paul:

I think you're right. The Geforce256 isn't a dual head GPU.

Of course it is a dual-head capable GPU. I have used several Geforce256
cards in dual monitor config.

Benjamin

Benjamin Gawert wrote:
Am 17.08.2010 00:31, * Paul:

I think you're right. The Geforce256 isn't a dual head GPU.

Of course it is a dual-head capable GPU. I have used several Geforce256
cards in dual monitor config.

Benjamin

So why does that LeadTek software have what looks like
a 1-of-N selector ? Is this a function of the OS being used
or something ?

http://www.xbitlabs.com/images/video.../drivers-2.jpg

*******

I can see a reference here to the term "Twinview".

http://http.download.nvidia.com/XFre...ppendix-g.html

"Nothing gets displayed on my second monitor; what is wrong?

Monitors that do not support monitor detection using Display Data Channel
(DDC) protocols (this includes most older monitors) are not detectable by
your NVIDIA card. You need to explicitly tell the NVIDIA X driver what
you have connected using the "ConnectedMonitor" option; e.g.:

Option "ConnectedMonitor" "CRT, CRT"
"

I have a feeling the only way to get this running, is to be
sitting in front of it.

Paul

Am 17.08.2010 07:57, * Paul:

So why does that LeadTek software have what looks like
a 1-of-N selector ? Is this a function of the OS being used
or something ?

http://www.xbitlabs.com/images/video.../drivers-2.jpg

I don't know, but I never used any of the crap drivers provided by gfx
card manufacturers but the original drivers from Nvidia.

Benjamin

Am 17.08.2010 07:57, * Paul:
So why does that LeadTek software have what looks like
a 1-of-N selector ? Is this a function of the OS being used
or something ?

http://www.xbitlabs.com/images/video.../drivers-2.jpg

BTW: you might want to read this to see why this particular card is a
bit different from other Geforce 256 cards:

http://www.anandtech.com/show/494/1

Benjamin