DisplayPort hot-plug woes

[djsb]

I did not want to derail the 1440p IPS and Displayport Only thread, so I'll ask this here: have there been any advances toward finding way to disable hot-plug detection of DisplayPort monitors in Windows 7?

I'll summarize for people who aren't familiar with this bug: when a DisplayPort connected monitor is powered down, the entire device gets removed and its portion of the extended desktop goes with it. The expected behavior is that the desktop layout remains untouched while the monitor goes dark, but instead any open windows or desktop icons get haphazardly resized and dumped onto another monitor. As you can imagine, repositioning your icons and open windows after turning on your monitor gets old, quickly.

I recently started digging into this problem again, but all I could find were the typical half-rumor-half-witchcraft remedies that tend to pop up on question/answer sites. Seriously, after reading a lot of these threads, I started expecting people to suggest pouring a circle of flour around their computer chairs before opening regedit.

Notably, it was claimed that Nvidia cards could prevent the unwanted behavior by disabling a certain service; but it was then claimed that only Quadro drivers could do it. The same was claimed about disabling the "AMD External Events Utility" service (tried it, no dice). Anyone here have any concrete info on the subject?

 

[kevinsbane]

Unfortunately it's not a bug... it's intended behaviour. It's a feature! (a terrible one)

 

[taq8ojh]

It's not just DP. HDMI does the same thing.

 

[djsb]

It's not just DP. HDMI does the same thing.

Strange, one of my monitors is connected via HDMI and it doesn't have that problem. Maybe because there's a HDMI->DVI adapter attached to the monitor.

Oh well, guess I'll be keeping my DVI ports till they can pry them from my cold, dead hands. (Or until I get tired of grappling for those damned screws.)

 

[iCyborg]

With HDMI it's up to the monitor whether it reports the on/off signal, but with DP, it's mandatory by the spec.

I also hate this "feature", they should at least make it configurable. I have a TV connected to my PC, and if I want to watch a movie with lights off, I'd like to turn the monitor off too, without all the stuff from my monitor being sent to the TV. And when you reenable it, the monitor returns as a secondary display Freaking annoying, so I just keep it connected with DVI, which renders my second monitor literally useless, it hasn't been turned on for 7-8 months. It's a crappy 22" TN panel, but still...

 

[blackened23]

It's not a bug, there are a lot of crap DP cables. You need a DP 1.2 cable that is actually compliant with the VESA spec. A lot of cheap cables aren't.

 

[iCyborg]

It's nothing to do with DP cables. If you unplug the monitor, it will be detected and the monitor will be disabled with all the accompanying side effects. I'm fine with this, and no cable will help you since it's not even connected to the graphics card. The problem, for me at least, is that with DP, this unplug behaviour also kicks in when you power off the monitor without unplugging.

For HDMI and DVI, there's a signal that indicates presence, and most monitors will keep it even if they are powered off. The reason there's no "bug" in this case is that the driver simply doesn't know if the monitor is powered off or not, so it can't do anything about it.

In short, it is the very VESA DP spec that requires the monitor to report this "power off" event to the driver for DP, and this gives the chance to the driver to do something about it. This is a good feature in theory, but the OP and myself just wish the driver would do nothing, or at least give us the option to choose what to do.

 

[KingFatty]

The approach of powering-off the monitor seems like a bottom-up approach to get the monitor to shut off, resulting in the side-effects.

Have you tried the top-down approach, would that avoid the side-effects? What I mean is, you use windows+P key combo to reconfigure the displays to not use that display, and then the display's own power savings will kick-in and effectively shut off the monitor. Maybe then, because windows itself initiated the shut off via this top-down approach, maybe there won't be any side effects?

If you have 2 monitors only, it will be trivial to use windows+p. But if you have 3 monitors, you will need to adjust display settings under each of the toggle options that windows+p calls up. For example, you could use "extend" to keep all 3 monitors active, but then you can use win+P to toggle to duplicate or projector only, and then adjust display settings to keep 2 of 3 monitors active, with that 3rd monitor inactive.

 

[Juraj.M.]

Hello from the future!
5 years passed - it's year 2018 and the bug is still here!

It could be fixed by Microsoft, AMD or nVidia, but nobody cares.
How is this even possible? Are there no Windows users with more than one screen in these companies? If I was working there, I would take an unpaid vacation just to fix this or provide one checkbox "Disable hot-plug detection".
If anyone has any idea how to get this fix, how to escalate this to the right people, please let me know.

 

[Mr Evil]

I've had to deal with this both at home and at work for years. I've resigned myself to having to flip all the windows around every time the computer wakes up, and never manually turning off the monitors.

There have been many threads here and across the wider internet expressing discontent with the situation, but apparently someone somewhere thinks this is the way it should be.

 

[tommo123]

would something like a HD Fury Linker work? obv a DP version (if it exists). people use them so their equipment doesn't detect a device being disconnected?

are there DP switches you could get? that way the monitor being off is irrelevant as the PC could only see the switch?

 

[Juraj.M.]

I can confirm that DP to DVI adapters solves this problem - but many new monitors doesn't have DVI anymore. Also this is only up to 1440p@60 (QHD), if you want 4K, you can get only 30Hz. And you will loose special features like Adaptive sync or HDR.

I haven't tested this with DP to HDMI adapters but it could work. If somebody tested it, please let me know.

Anyway buying new hardware to fix a software issue - missing checkbox - is crazy!

 

[iCyborg]

You should read my posts before - it's not a SW issue, there's nothing MS, AMD or NV can do about it. They will get HPD (hot-plug-detect) signal off - basically SW cannot distinguish between monitor unplugged and turned off. AFAIK, you'd need to get VESA to change DP spec, because right now it's a feature...

 

[Feld]

You should read my posts before - it's not a SW issue, there's nothing MS, AMD or NV can do about it. They will get HPD (hot-plug-detect) signal off - basically SW cannot distinguish between monitor unplugged and turned off. AFAIK, you'd need to get VESA to change DP spec, because right now it's a feature...

MS could certainly add a checkbox to turn off desktop resetting until reboot regardless of whether the screen is turned off or unplugged. They don't need to differentiate to do that.

 

[Alpha One Seven]

I did not want to derail the 1440p IPS and Displayport Only thread, so I'll ask this here: have there been any advances toward finding way to disable hot-plug detection of DisplayPort monitors in Windows 7?

I'll summarize for people who aren't familiar with this bug: when a DisplayPort connected monitor is powered down, the entire device gets removed and its portion of the extended desktop goes with it. The expected behavior is that the desktop layout remains untouched while the monitor goes dark, but instead any open windows or desktop icons get haphazardly resized and dumped onto another monitor. As you can imagine, repositioning your icons and open windows after turning on your monitor gets old, quickly.

I recently started digging into this problem again, but all I could find were the typical half-rumor-half-witchcraft remedies that tend to pop up on question/answer sites. Seriously, after reading a lot of these threads, I started expecting people to suggest pouring a circle of flour around their computer chairs before opening regedit.

Notably, it was claimed that Nvidia cards could prevent the unwanted behavior by disabling a certain service; but it was then claimed that only Quadro drivers could do it. The same was claimed about disabling the "AMD External Events Utility" service (tried it, no dice). Anyone here have any concrete info on the subject?

I wouldn't expect the desktop layout to remain the same, but I have been doing it for a while. If you keep your icons on the taskbar, they stay where you leave them. If you close the open windows on the display you are turning off, it makes it simpler as well.

 

[iCyborg]

MS could certainly add a checkbox to turn off desktop resetting until reboot regardless of whether the screen is turned off or unplugged. They don't need to differentiate to do that.

Fair enough, although you might be left with an unusable desktop if your primary monitor actually fails.

And I was wrong - AMD FirePro has an 'EDID simulation' feature where you can force a monitor over some connector even if it's not there (obviously, you force the one that's usually connected to that port). Normally it's used for those display walls with 10+ monitors where it's pretty annoying when a monitor is disconnected or goes down, and you need to redo all the setup LOL.
So, there is a solution, but they limit it to with workstation GPUs.

 

[Juraj.M.]

You should read my posts before - it's not a SW issue, there's nothing MS, AMD or NV can do about it. They will get HPD (hot-plug-detect) signal off - basically SW cannot distinguish between monitor unplugged and turned off. AFAIK, you'd need to get VESA to change DP spec, because right now it's a feature...

Of course it can be solved by software. Since this bugs was introduced to my after upgrading my GPU driver from 16.11.5 to 16.12.2 and I was then running the old version for half a year just to avoid it. See my thread here: https://community.amd.com/thread/210595

 

[iCyborg]

I've worked with DisplayPort and this is a fact: OS/driver cannot distinguish between unplug and power off, unless perhaps the monitor has a different policy for power off, and this is up to the monitor.
Any SW solution is basically "hacking" around this: spoofing / emulating a fake monitor when the real one is gone. Another SW solution is to ignore disconnect which is not good either, even if it may be the lesser of two evils here.
I've had this problem since 2011 with AMD. It sounds more like 16.11.5 was broken, as far as I know, this behavior is by design, and I hate it too...

Here's the same problem with GTX1080 with pretty much the same answer:
https://forums.geforce.com/default/topic/984146/displayport-how-to-disable-power-off-detection/

 

[Juraj.M.]

seems like you have some "alternative facts".
Because I can tell you as a FACT that I've been using my 7870XT GPU for many years with 3 monitors, two of them connected with DP and never had this issue. AND It could correctly distinguish "power off" / "sleep" and "cable unplug" so I'm pretty sure it's possible.
And if I make a driver upgrade and this problem appears, then I would say it's pretty software related, especially when I'm using same monitors. And if nVidia is saying they cannot fix it's because they don't want to, not because it's impossible.

 

[iCyborg]

Seems like Microsoft also has the "alternative facts":
Some DisplayPort monitors, when powered off, may de-assert their Hotplug Detect signal, indicating to the computer that the monitor may have been unplugged
https://support.microsoft.com/en-us...ion-when-turning-off-displayport-monitor-on-w

 

[Alpha One Seven]

Seems like Microsoft also has the "alternative facts":
Some DisplayPort monitors, when powered off, may de-assert their Hotplug Detect signal, indicating to the computer that the monitor may have been unplugged
https://support.microsoft.com/en-us...ion-when-turning-off-displayport-monitor-on-w

"Some" lol there's your answer.