TwinView: 1600x1024 too large?

[direvus]

Hi all,

I am currently trying to run two 22" widescreen TFTs. Both monitors are the AOC 223VW, with a native mode of 1680x1050 @ 60Hz.

I have a FX 5200 (AGP, with DVI and VGA outputs) as well as two PCI MX 4000s. I've tried various configurations but no success so far.

First I tried just driving both monitors from the FX 5200 ... makes sense right? I was using separate X screens. For some reason, the monitor on the DVI output would not go higher than 1280x1024 (which looks really quite awful on a widescreen monitor).

So, figuring that the card couldn't drive both monitors at that high a resolution, I tried driving the second monitor from one of the MX 4000s. That worked for a few minutes, then crashed (WAIT errors in the Xorg log, "Xid" messages in the kernel log).

So I tried using both MX 4000s, one monitor apiece. Worked for a few minutes then crashed (WAIT and Xid).

I noticed that the two video cards seemed to be sharing an IRQ so I spent some time messing around with kernel options and BIOS settings trying to resolve the conflict. I resolved the conflict (the cards were then on separate IRQs) but the crash still happened, and in fact it seemed to crash harder. Even after sshing into the box and killing X, my display was all screwed up and I had to reboot.

After doing some reading on WAIT/Xid problems, I got the impression that these errors are completely impossible to diagnose. So I gave up on the multiple card approach.

As a last ditch effort I have plugged the AGP card back in and tried to use TwinView (even though this is not my preference). It doesn't work either. I get the following line in my Xorg log:

(WW) NVIDIA(0): Mode "1600x1024" is too large for AOC 223VW (DFP-0);
(WW) NVIDIA(0): discarding.

Can anybody tell me where, in the name of all things sane, the nvidia driver is getting the idea to try 1600x1024? This resolution does not appear anywhere in my configuration. My MetaModes are:

"1680x1050,1680x1050; 1680x1050,NULL"

And my Screen Section has only one Display subsection:

SubSection "Display"
Depth 24
Modes "1680x1050"
EndSubSection

So this "1600x1024" resolution seems to have coalesced from thin air. I certainly didn't specify it, I know my monitor can't handle it, and it doesn't even conform to any of the normal aspect ratios.

Any suggestions?

[direvus]

Originally Posted by direvus

As a last ditch effort I have plugged the AGP card back in and tried to use TwinView (even though this is not my preference). It doesn't work either. ...

By "doesn't work", I mean the primary monitor displays at the desired resolution, but the second monitor, the one on the DVI output, doesn't come up at (no signal).

[AaronP]

Please see the instructions in http://www.nvnews.net/vbulletin/showthread.php?t=46678 for generating a verbose bug report.

Edit: D'oh, didn't read carefully enough.

[direvus]

Originally Posted by AaronP

Please see the instructions in http://www.nvnews.net/vbulletin/showthread.php?t=46678 for generating a verbose bug report.

Edit: D'oh, didn't read carefully enough.

Okay, here goes.

My xorg.conf file is included below, and I have attached the nvidia-bug-report log and output from lspci -v.

Section "ServerLayout"
	Identifier     "X.org Configured"
	Screen      0  "Screen0" 0 0
	InputDevice    "Mouse0" "CorePointer"
	InputDevice    "Keyboard0" "CoreKeyboard"
EndSection

Section "Files"
	RgbPath      "/usr/share/X11/rgb"
	ModulePath   "/usr/lib/xorg/modules"
	FontPath     "/usr/share/fonts/misc/"
	FontPath     "/usr/share/fonts/TTF/"
	FontPath     "/usr/share/fonts/OTF"
	FontPath     "/usr/share/fonts/Type1/"
	FontPath     "/usr/share/fonts/CID/"
	FontPath     "/usr/share/fonts/100dpi/"
	FontPath     "/usr/share/fonts/75dpi/"
EndSection

Section "Module"
	Load  "glx"
	Load  "extmod"
	Load  "xtrap"
	Load  "record"
	Load  "dbe"
	#Load  "dri"
	Load  "freetype"
	Load  "type1"
EndSection

Section "InputDevice"
	Identifier  "Keyboard0"
	Driver      "kbd"
EndSection

Section "InputDevice"
	Identifier  "Mouse0"
	Driver      "mouse"
	Option	    "Protocol" "auto"
	Option	    "Device" "/dev/input/mice"
	Option	    "ZAxisMapping" "4 5 6 7"
EndSection

Section "Device"
	Identifier  "AGP"
	Driver      "nvidia"
	VendorName  "nVidia Corporation"
	BoardName   "NV34 [GeForce FX 5200]"
	BusID       "AGP:1:0:0"
	Option	    "NoLogo" "true"
	Option	    "TwinView"
	Option	    "MetaModes" "1680x1050,1680x1050; 1680x1050,NULL"
	Option	    "HorizSync" "CRT-0: 30-80; DFP-0: 30-80"
	Option	    "VertRefresh" "CRT-0: 60; DFP-0: 60"
	Option	    "TwinViewOrientation" "RightOf"
	Option      "NoTwinViewXineramaInfo" "true"
EndSection

Section "Monitor"
	Identifier   "Monitor0"
	VendorName   "AOC"
	ModelName    "223VW"
	Option	     "dpms"
EndSection

Section "Screen"
	Identifier "Screen0"
	Device     "AGP"
	Monitor    "Monitor0"

	DefaultDepth	24

	SubSection "Display"
		Depth     24
		Modes "1680x1050" 
	EndSubSection
EndSection

[AaronP]

Your bug report is missing the verbose mode validation log. Make sure that you start X with "startx -- -logverbose 6" and not "startx -logverbose 6" (startx will silently eat the -logverbose option -- helpful, that. ). According to the log, your 1680x1050 modes are being validated and that's what you're getting. If this is causing one of your displays to turn off, then it sounds like that display doesn't actually support that mode. The message about 1600x1024 looks unrelated.

[direvus]

Thanks to that handy "logverbose 6" option (thanks AaronP) I found the crucial bit of info.

(II) NVIDIA(0): --- Building ModePool for AOC 223VW (DFP-0) ---
(II) NVIDIA(0):   Validating Mode "1680x1050":
(II) NVIDIA(0):     1680 x 1050 @ 60 Hz
(II) NVIDIA(0):     Mode Source: EDID
(II) NVIDIA(0):       Pixel Clock      : 146.25 MHz
(II) NVIDIA(0):       HRes, HSyncStart : 1680, 1960
(II) NVIDIA(0):       HSyncEnd, HTotal : 2136, 2240
(II) NVIDIA(0):       VRes, VSyncStart : 1050, 1053
(II) NVIDIA(0):       VSyncEnd, VTotal : 1059, 1089
(II) NVIDIA(0):       H/V Polarity     : -/+
(WW) NVIDIA(0):     Mode is rejected: PixelClock (146.2 MHz) too high for
(WW) NVIDIA(0):     Display Device (Max: 135.0 MHz).

I don't really grok what a "pixel clock" is, but my primary monitor reported a pixel clock of 350 MHz, while the secondary apparently can only do 135 MHz.

Here's the funny part: they are the same monitor. Both AOC 223VW.

The only difference between the two is that the primary is plugged into the video card's VGA output, and the secondary into the DVI.

I figured the pixel clock could be a property of the DVI input on the monitor, so I got a DVI-VGA adapter and hooked the secondary up with a VGA cable, to the VGA input on the monitor. And suddenly everything just works.

So, in the end I had to downgrade to an older technology to make it work (?), and apparently the DVI input on the monitor isn't actually capable of driving the monitor at its own native resolution.

I still have no idea why the DVI didn't work. I'm left feeling puzzled and annoyed, but this is mitigated by the fact that I'm now sitting in front of 3360x1050 pixels of combined screen real estate. =)

[Thunderbird]

The pixelclock is mainly a property of the videocard. The higher the resolution and refresh rate you want to use, the higher the pixelclock. For most (older) videocards the maximum DVI pixelclock is relatively low (135-150MHz). Older LCD screens didn't use resolution higher than 1280x1024 and digital transmitters are expensive.

The old analog signal isn't that limited most cards easily support pixelclocks of 300MHz or higher (even upto 400MHz), that's why you aren't seeing issues using analog.

[pe1chl]

Originally Posted by direvus

apparently the DVI input on the monitor isn't actually capable of driving the monitor at its own native resolution.

That is not a problem of the monitor, but of the video card. Most 5200 cards will only do up to 135 MHz but usually a monitor assumes it can do 155 MHz.

135 should be enough for a DFP at 1680x1050. Try a modeline like this:

Modeline "1680x1050_60.00_rb" 119.00 1680 1728 1760 1840 1050 1053 1059 1080 +HSync -Vsync

When the monitor accepts it, it should work fine. It is also possible to turn off the "135 MHz check" but that could cause problems. A newer videocard will also fix it, they usually have the 155 MHz limit.

[direvus]

I see.

I had thought that it must have been a monitor issue, because the only thing I changed was which input I plugged into on the monitor.

I suppose that, when I plug into the VGA input on my monitor, the video card recognises this and switches from digital to analog mode.

Thanks for the info.

I won't worry about trying to get the digital mode working. At least, not immediately. Analog is working fine, and I want to actually get some use out of this setup before I start tinkering again!

[pe1chl]

Originally Posted by direvus

I suppose that, when I plug into the VGA input on my monitor, the video card recognises this and switches from digital to analog mode.

Yes. On a DVI-I output (the white connector with a field of pins and a separate area with 4 pins and a blade) there are actually two sets of outputs: analog and digital.
The card recognizes what is plugged into it (or what you select in the config) and outputs a digital or an analog signal.

The trouble is the digital signal has a much lower pixel rate limit than the analog signal. Well, actually the signalling limit is much higher, but the analog output uses 3 leads to output R, G and B separately and one pixel at a time. The digital output sends the RGB value as a serial dataword of 24 bits over a single lead. So in the same interval the analog output sends 1 multi-level signal per lead, the digital output has to send 24 bits one at a time.