DFP image corruption with DVI

[mbidewel]

I have a Geforce 6200 video card. When I use the DVI input on my monitor with the linux 9xxx drivers, the screen will show only 1280x1024 not the native resolution of 1600x1200. using "NoDFPNativeResolutionCheck" fixes this problem however, I periodically experience severe image corruption and color palette problems. The issues will survice a reboot and continue for some time (generally until I have switched to the analog input). When using the analog input, the correct resolution is detected and there are no image issues. Stable drivers prior to 9xxx did not have this problem and detected the correct resolution through the DVI input. The windows drivers work correctly through the DVI input

Any thoughts on what could cause this?

[mbidewel]

Here are my logs, one with NoDFPNativeResolutionCheck" enabled and one without.

[netllama]

In the bug report, I see:
(--) NVIDIA(0): Envision EN2028 (DFP-0): 155.0 MHz maximum pixel clock
(--) NVIDIA(0): Envision EN2028 (DFP-0): Internal Single Link TMDS

Your graphics card would need to have dual link TMDS support to drive 1600x1200.

[pe1chl]

Originally Posted by netllama

Your graphics card would need to have dual link TMDS support to drive 1600x1200.

That is incorrect, a single-link TMDS at 155 MHz has ample bandwidth for 1600x1200 at 60Hz. It requires about 130 MHz.
My display is 1920x1200 and even that works OK on a single link at 60Hz.
(it remains just below the 155 MHz limit)

[mbidewel]

If this helps, the color problems I experience are similar to the ones shown in this thread

http://www.nvnews.net/vbulletin/show...t=77816&page=3

[netllama]

pe1chl,
I'm afraid that you're misinterpreting the use of reduced blanking as proof that a single link TMDS card can always drive 1600x1200. Running 1600x1200@60 requires at least a 161Hz pixelclock.

I've not come across any GeForce 6200 cards that were capable of driving 1600x1200 without reduced blanking.

That said, I suppose its possible that mbidewel has hardware that is capable of doing 1600x1200 at a refresh rate less than 60Hz, however since a verbose bug report wasn't provided, its not possible to say for certain.

Thanks,
Lonni

[mbidewel]

I am not totally clear on how to produce a more verbose bug report. Is that the Xorg.0.log file generated from running startx -- -logverbose 6 ?

[mbidewel]

Also, I was unaware of some of these issues when buying the 6200. Can anyone recommend resources so that I can make better informed future purchases

[netllama]

Yes, you need to run:
startx -- -logverbose 6

and then generate the bug report. The bug reports that you've attached were not generated after starting X with that command.

As for the hardware, you'd need to purchase a card with dual link TMDS support. Most (but not all) cards with two DVI connectors are usually dual link TMDS capable. Its best to read the fine print, or contact the card manufacturer to confirm.

Thanks,
Lonni

[pe1chl]

Originally Posted by netllama

pe1chl,
I'm afraid that you're misinterpreting the use of reduced blanking as proof that a single link TMDS card can always drive 1600x1200. Running 1600x1200@60 requires at least a 161Hz pixelclock.

I suppose the LCD screens that cannot use reduced blanking are to be found in the same store where they still sell monitors that can be destroyed by a too high resolution???

[mbidewel]

Thank you for your help I will post the verbose log ASAP. However, I am not clear on one thing. I am beginning to understand the hardware capability issue however hardware capabilities do not seem to explain why the problem has never occured under Windows (unless I have booted into Linux and experienced the problem there first) or with the 8xxx driver series.

[mbidewel]

Originally Posted by pe1chl

I suppose the LCD screens that cannot use reduced blanking are to be found in the same store where they still sell monitors that can be destroyed by a too high resolution???

I wouldn't be shocked. NEVER buy an envision monitor. They have low prices but are short on features.