Nvidia GeForce GTX 1060 graphics card failing after hibernate: Solution Found

[Bud]

I thought I could pass this information along so it may help others who is still experiencing the same problem.

It seems to be a widespread problem with Nvidia graphics cards to not re-initialize when the computer returns from sleep or hibernate state . You end up with a black screen, though it can be seen by observing the HDD LED that the computer is running. I had this problem after installing a GTX 1060 3Gb card in my computer with an ASUS motherboard. I before used hibernate very often, hibernation allowed me to return to the applications where I left , i.e. straight to Altium PCB editing, there was no need to reload Altium and the project. With the new GTX 1060 it became hit and miss and eventually I had to stop using hibernate and had to fully shut down the computer for the nights. This caused big inconveniences to me so I tried to investigate what can be done and tried every advice under the Sun that I could find on the Net, including removal of Nvidia audio drivers, changing settings in Windows , etc. Sure enough none worked for me.

This continued until I accidentally noticed that this may have something to do with the monitors I used , a dual setup of HDMI Dell U2515H. When I hibernate the computer for the night I usually also turn off the extension power bars which power the computer and the monitors. I noticed that if next morning I power up the computer before I power up the monitors, I get the dreaded black screen. If I power up the monitors first and then the computer, I was getting random results, the system could reinitialize fine or not and I still got the black screen.  Eventually, I noticed that it takes the monitors a few seconds to go through internal boot-up procedure before they can sense the signal from the computer. So I tried the following sequence for the morning power up:

- power the monitors
- wait until the monitors finish booting up
- power on the computer

That did the trick! After I started doing this the graphics card always re-initializes properly now. So in my case the GTX 1060 "failed to return from hibernate" (and from sleep if the monitors were turned off) problem had something to do with the computer booting procedure, with indications that it failed if the motherboard could not see HDMI monitors online as part of the POST procedure or something and as the result did not re-initialize the Windows Nvidia driver. If however the monitors were powered up AND completed their internal boot procedure before the computer starts, the motherboard happily picks them up and re-initialized the driver.

It seems to me that this may be the ASUS motherboard issue and not Windows driver issue because it only take the monitors about 5 seconds to boot, and they are ready quite a while before Windows kicks in. I believe my motherboard is ASUS Sabertooth X79 , but the problem may be common to many other ASUS motherboards if ASUS is using same or similar BIOS routines to scan the bus for video devices during boot-up.

Another solution would obviously be to never turn off the monitors, but I have a habit to shut all power off in the shack when I am not there for a prolonged period of time.

[Monkeh]

Have you tried power cycling the displays after the failed boot?

[Bud]

Can't remember 100% but I think I did try, as well as re-connecting the HDMI cables.

What is still sort of a problem is if one of the HDMI monitors is powered off, the computer does not pick it up upon boot and that messes up the other monitor because all open applications screens from the inactive monitor get pushed to the active monitor. This was not the case with DVI monitors. They did not mess each other and boot sequence did not matter.

[Monkeh]

DVI and HDMI are effectively the same - it seems your HDMI monitors are being too smart for their own good.

Try switching to DP. Those monitors can't run at their native res on HDMI anyway.

[Bud]

Sorry for the confusion  , I am running them on miniDP not HDMI.  The HDMI port on the card is connected to my Sony TV, which was not part of the investigation.

[Monkeh]

Ah. Well, with DP, you do have the problem that the monitor needs to be turned on for the computer to see it. With VGA, DVI, and HDMI, the PC powers the EEPROM (or microprocessor) over the cable and knows of its presence.

This is indeed some sort of BIOS and/or VBIOS limitation - plus perhaps some brain damage in the host OS (which should detect resume from hibernation and force a display probe). For example, I use a DP to VGA adapter - with my old motherboard, that display does not come up until the OS driver loads, with the new board, it comes up on boot.