Diagnostic advice needed: PC reseting during boot.

[ikonia]

I recently did a reasonable upgrade to my desktop, new case, new cpu, new ram, new PSU, new motherboard, the only thing that stayed the same was my Nvidia Graphics card.

All components are bought from scan and have been working great for approx 2 months, maybe a little longer.

I left my PC on overnight running Windows 10 and handbrake doing some video encoding ( approx 14 hours from start to when I returned to it).

When I returned to it the Windows 10 blue flag/window boot logos was on screen and static. something happened during the night to cause it to try to reboot, I'd assume an update as I can't remember if auto updates where enabled or not, or maybe it just crashed, I don't know.

No big deal, I rebooted the machine, the machine passes post, and starts to boot windows 10, it gets to the windows 10 blue flag/logo and does a soft reset back and starts the reboot process again, passes post, boots to windows 10 blue flag logo, and soft resets again, and keeps doing that until the machine is powered off.

If I attempt to boot from the windows 10 DVD it boots to the blue flag and resets, exactly the same as booting from the M2 drive where the OS is. ( yes I am %101 certain it's booting from the DVD not the hard disk).

if I boot a Linux liveCD (Fedora/Ubuntu - tried both) it starts the boot process, (I boot without the splash so I can see what's going on) and soft resets again, with no warning or error.

I can boot memtest and did a 4 pass test (approx 6 hours run time) totally stable, no errors, using all 8 cores and all ram. - while I appreciate this is not an extensive test, it seems odd that it will boot and run memtest just fine for 6+ hours but as soon as it tries to boot an OS it resets.

I've stripped the machine back to minimal, took the nvidia card out and using the on chip graphics, removed any other hard disks, and any USB devices. gone back to one memory dimm (tried both in single dimm setup alternating them).

it seems pretty clear there is a hardware fault at this stage but the question is what and how to narrow it down before raising an RMA with scan, as scans answer will be send it all back, which means stopping all the kit down finding about 10 boxes to send all the components back, paying shipping and then waiting over a month for scan to do the tests and process the RMA process. I'd rather try to get some confidence in what's going wrong and just focus on returning that part for RMA.


The components that are currently in the case are.

Asus Intel Z170 Pro
Intel 540s 1TB M.2 SATA
Corsair 32GB White Vengeance LPX DDR4 2666MHz RAM (only 1 Dimm at a time in)
Intel Core i7 6700K Unlocked Skylake Desktop Processor
be quiet! Power Zone 850W Modular Power Supply

that is what it is currently stripped down to and what is still failing.

http://bender.dnsdata.co.uk/video/bootfailVideo.MOV

Any suggestion on how to narrow this down and get the best info for scan so that I can minimise the impact on this RMA ?
Kit is approx 2 months old and has worked great up until yesterday.

Suggestions welcome and appreciated.

[ikonia]

additional info, removed M2 drive, managed to get a spare PSU and test, no difference at all, gut it now leaning towards cpu/mother board.

ram seems to pass all possible tests, so while it's a possibility it seems less likely as running both sticks together or individual sticks on their own makes no difference, so that would mean both ram sticks would be faulty.

I assume the best option is to get an RMA for the motherboard and CPU, send them both back and let scan work it out.

I'd welcome any other opinions on how to narrow this down to help get solid info for scan.

[Ferral]

Could be the modular cables on the PSU, mine goes funny if I use the VGA cables that connect to the side of my PSU for my graphics card. Other than that maybe a hard drive fault, I had to replace my secondary drive a few weeks back as it was causing my machine to crash when I was running more intensive games like Doom.

Could even be a Windows installation issue, have you tried a clean install to see if that resolves the issue, again had an issue after a AMD driver update where windows wouldn't recognise my model of card, made a right mess of things and I had to clean install Windows?

[ikonia]

Could be the modular cables on the PSU, mine goes funny if I use the VGA cables that connect to the side of my PSU for my graphics card. Other than that maybe a hard drive fault, I had to replace my secondary drive a few weeks back as it was causing my machine to crash when I was running more intensive games like Doom.

Could even be a Windows installation issue, have you tried a clean install to see if that resolves the issue, again had an issue after a AMD driver update where windows wouldn't recognise my model of card, made a right mess of things and I had to clean install Windows?

apologies - but did you actually read my main post properly ?

there are no hard disks in this other than the m2 drive - I cannot boot the hard disk, or windows 10 install media, or linux liveCDs as the machine resets.

I've tried a totally different PSU (including the cables with that different PSU) and the machine has been working fine for 2 months before this weekend.

I appreciate input, but it would be better if the info I supplied was digested before offering suggestions.

[carved_eye]

Have you tried booting into safe mode? It could also be the drive has possibly turned bad/malfunctioned?

[ikonia]

I can't boot into safe mode - I can't boot from livecd's / installCD's even with no hard disks in the system.

This is all detailed in the post I made at the start

[peterb]

I was about to suggest a hardware fault in the reset timer that holds the CPU reset low until the system voltages have stabilised, but that would force a hard reset, and it seems to be resetting at the same point in the boot process. The other thought is corruption of the BIOS and the system call pointers at the very start of the boot process are getting corrupted, although that wouldn't explain why it failed with an OS already up and running (unless Windows doesn't overwrite those once it starts)

But that is just clutching at straws.

I can't really see how it would be CPU though, unless it is overheating and forcing a reboot, but unless the heat sink was off completely, it wouldn't be that quick. (The fan is working? ) so my thoughts tend to point to a mobo fault, although proving it is faulty except by substitution would be tricky.

(Good to see you back here, although pity about the circumstances)

[ikonia]

I was about to suggest a hardware fault in the reset timer that holds the CPU reset low until the system voltages have stabilised, but that would force a hard reset, and it seems to be resetting at the same point in the boot process. The other thought is corruption of the BIOS and the system call pointers at the very start of the boot process are getting corrupted, although that wouldn't explain why it failed with an OS already up and running (unless Windows doesn't overwrite those once it starts)

But that is just clutching at straws.

I can't really see how it would be CPU though, unless it is overheating and forcing a reboot, but unless the heat sink was off completely, it wouldn't be that quick. (The fan is working? ) so my thoughts tend to point to a mobo fault, although proving it is faulty except by substitution would be tricky.

(Good to see you back here, although pity about the circumstances)

so I walked through the similar thought process as you, reset the bios to be safe, which no surprise didn't do anything, but it was an easy thing to rule out. I managed to get an MCE event out of it for a split second, which suggested problems on core 0 and core 5 - but it was only on screen for a second or two, so it could have been the memory bank that was serving core 0 and 5 and there could have been more that I missed.

I have Grafana logging system statistics in general and I looked back over the data of the last few days, hottest the cpu got was 72 degree's while encoding, and it has a monster set of fans on it so that's good news there. I've pulled the motherboard out of the case ready for an RMA to scan on Monday when they open and removed the heat sync (I"lll send it back with CPU and motherboard as I'm not sure which one is playing up) and the thermal paste looks in a nice even spread on both the fan and the cpu - I've not cleaned that off so scan can see the exact state it came out of the chassis to return it to them, so I think over heating is not really an option. The board temperatures show 24 - 29 consistently, and it shows the power as a stable supply, although my psu monitoring stats are not great, so I'm not depending on them %100, but it's good to use as a guide to see if anything obvious is wrong.

the ram checks out in this system as best it can, so I'm not going to send that back in the RMA unless scan recommend it (then they can test board/cpu/ram together exactly the same as me, I'll wait for them to open on Monday and ask them what they want me to RMA.

I've put a link to this thread in their forum so they can see the full picture and hopefully advise me what to RMA to make the process quick and simple for both me and them.

[peterb]

One last thought, a high resistant joint somewhere on the power supply lines on the mobo. Me test doesn't load the processor much, but running an OS could increase the current drawn to the point where the voltage drop across the joint causes it to reboot, but again I'd expect a hard reboot, and why would that show itself now? (Unless it's been getting gradually worse over time). Any signs of discolouration or overheating on any of the tracks?

[noveltylamp]

Hey, sorry if i've missed it in the above posts, I can see you've reset the bios - have you tried reflashing it? How about removing the CPU and checking the pins are in tact on the board and there's nothing shorting the points on the LGA under the CPU?

The other thing is possible shorts on the motherboard, checked no screws or unneeded motherboard supports underneath?

[lljoe]

Maybe you could check the BIOS battery!

[Bonebreaker777]

Funny enough, last week I was doing a miniaturisation project (leftover for sale) and we had a very similar experience. Eventually when we took everything out, we came to the conclusion that we must reseat the CPU (or something went very bad somewhere along the way). After 3x attempts we managed to get it going and apparently it is still going.

Try everything, even repeatedly. Never mind Einstein and his famous line: "Insanity is doing the same thing over and over again and expecting different results."