KVM switch + Windows 7 creates monitor switching nightmare

Hi Xenubite,

Thanks for the info.  It's plausible that this is happening. My cables are DVI-I and the non determenistic behaviour certainly has the aroma of a problem of this nature.

Unless there is some other way to verify if my dislay is a "dual EDID digital display" as referred to in the link I can try running on of the displays through VGA temporarily to see what the effect is.

I like your adventurous suggestion. I would probably do this except that I paid a fair bit for these high quality long dvi-i cables and butchering makes me wonder when I'll wish I didnt - albeit unlikely.

I've got options for the vga connector anyway, so if I need to I'll give this a whirl.

Replacing one of these nice high quality and lengthy DVI-I cables may be a bit much of a sacrifice to make it a permenant solution, but at least narrowing down the fault should help with a resolution.

I'll report back my findings.

The dual EDID issue could well be the reason for the problems with a single monitor too. Emre and Nikk0n, are your monitors connected with a DVI-I cable?

Hi Trees,

I wonder if your problem could be related to this:

http://www.ddwg.org/lib/Dual_EDID_guide10.pdf

Apparently the EDIDs for the digital and analog part of a DVI interface may be different, but they are polled through the same DDC pins. If the monitor is connected with a DVI-I cable the computer side may not know which one it is getting. They suggest using a DVI-D or a DVI-A cable as a workaround.

In your case the HDMI-DVI connection should work, since there's no analog part involved. Changing the other one to use the VGA would force that one to use analog and might fix it, or use a DVI-I to DVI-D dongle if there is one, or if you are adventurous just snap off the analog pins from the connector. The pinouts are here:

http://www.interfacebus.com/Design_Connector_Digital_Visual_Interface_DVI_Bus.html

Commentary here and my experience reflects a problem with communication between windows 7 and drivers.

In my case with a HDMI (in use), DVI (in use) and VGA port off a Gigabyte GTX275 Windows 7 gets inconsistant results varying from 1 to 4 identified displays mapped to any combination of (including phantom duplicates)

• Dell 24 (Digital)
• SyncMaster (doesnt specify digital --- what does that mean?)
• Generic non PnP

Right now the combinations are

1 - Dell (screen pictue in Control Panel accurately reflects active and 1920x1200 res.)

2 - Syncmaster (screen pictue in Control Panel accurately reflects active and 1600x1200 res.)

3 - Generic non PnP (screen picture in Control Panel in is grey and small)

4 - Syncmaster (screen picture in Control Panel is grey and small)

the dell 24 lcd is currently plugged to HDMI (DVI Cable, HDMI-DVI adapter)

the samsung 214T lcd is currently plugged to DVI (DVI cable)

the VGA port isnt used (hmm... i might just put this to some use and see what happens next)

Hi 12 Stringer,

If there is no "Default_Monitor" section in the registry, you shouldn't add one yourself. The existence of that section is an indication Windows was not able to get the EDID from the monitor, and if that section is missing the problem is somewhere else.

In your case it sounds like for some reason the internal monitor keeps waking up every now and then by itself, and if the monitors use clone mode that would force Windows to make the resolutions match. Using extended mode as a workaround might help.

For the ones that have a problem with just a single monitor, assuming you don't have Default_Monitor section in the registry either, the problem could be W7 not getting the same EDID from the monitor every time. Assuming W7 always assumes the EDID it gets is valid, and if the EDID polling sometimes stops halfway for some reason, it could then make Windows think the monitor keeps changing, because the capabilities it reports back change.

If that is the case a workaround could be using EDID override to override all EDIDs in the registry with a good copy, but knowing what to replace with what could be tricky. And it may not really fix the whole problem, because if the EDID polling really is erratic, who knows how many different bad copies end up getting added to the registry.