NUC will turn on or shutdown, doing a short between pins 6 and 8 (red pins).
On page 45 and 46 from the technical product specification PDF, it is available more information about the Front Panel Header.
I believe there must be a bug in the BIOS...
As a workaround, I found out that if I enable fastboot in Windows 8.1 power options, then I have no freeze and Windows boots when powering the NUC from the POWER_SWITCH# pin...
As soon as I disable fastboot in Windows 8.1, Windows doesn't boot anymore when powered on from the pin.
With the physical button, it works both with and without fastboot.
So there must be something wrong during the system initialization phase when powered on from the pin, which is circumvented by the hibernation file created by fastboot.
How can I get Intel support looking into it?
Intel motherboards need a short between pins 6 and 8 (red pins) to turn the system on and off. Pin 8 is the ground. Fast boot disable the Boot display options, boot from Network, Optical, Removable devices and RAID disable. Video and USB devices will not be available until after operating system boot.
1 of 1 people found this helpful
I am not from Intel, but one thing I am confused about is you say from S3 state, but I think that S3 state typically means Sleep/Suspend. I do not see Intel NUC logo screen when I am resuming from Sleep/Suspend. I think S4 typically means Hibernate (I don't do that), and S5 typically means power off. I do see Intel NUC logo when I am powering on from S5 (not S3). So it might make sense to clarify whether you really meant S3 in your original post. Windows 8.1 fastboot might be thought of sort of like a hibernate style resume but from S5 rather than S4. Of course mike was talking about a different fastboot (Intel's BIOS setting) - although he did not come out and suggest that if you changed the default setting of Intel's BIOS for general optimization or video optimization, you might try disabling those (the default) to verify that doesn't help.
Also I have never used Intel's Rapid Start or Smart Connect, one or both of which might be in their driver bundle and affect traditional meanings of some of those power states.
Once we clarify that, my comment might be to see if you have some way to try some different monitor connection while still somehow trying the POWER_SWITCH# pin situation you describe. There have been various reports at times about how Intel's NUC boot and/or graphics driver fail with certain connections of HDMI such as through Audio Visual Receiver and/or certain TVs. They do something like trying to read the display data and interpreting it which I guess is failing sometimes. Details of such procedure may vary between versions of BIOS and versions of graphics driver. I just thought it might be worth mentioning that in case you (or someone else reading this) might experiment as further diagnosis.
I realize that none of this directly addresses your statement about difference between pin and button, but they are things that I think could be related to sorting out the issue.
You're absolutely right, I've mistakenly written "S3" in my original post... I actually meant S5 of course. In the BIOS I have left unchecked the "Deep S4/S5" setting though, otherwise the internal usb port is not powered up...
Your comments make sense, it's indeed connected to my TV through an AV receiver (both supporting the HDMI-CEC protocol) so there must be some issue with reading the display data... But like you said, it still doesn't explain why the process differs between the physical button and the pin.