Help debugging a system

[dkanter]

I have a system that is misbehaving; it was working fine for about six months and suddenly become incredibly unstable. It can boot into windows, but hard reboots (no BSOD) after anywhere from a few seconds to a few minutes. The display also goes blank out for a few seconds at a time when Windows is running. It can run for hours in the BIOS or memtest86.

CPU: Intel Core i7-6700K (Skylake)
Motherboard: Asus Z170-A, using integrated graphics
Memory: Crucial Ballistix Sport 2x8GB kit
Power Supply: Corsair RM650i
OS storage: Seagate SSD

Steps I have taken so far:

0. Attempt to boot with both DIMMs, or each DIMM individually [suggested by posters]
1. Boot and run memtest86, everything checks out
2. Replace power supply
3. Replace motherboard
4. At this point, the only thing left to replace is the processor, but I’m hoping that it’s some other sort of problem.
5. Attempt to boot with the motherboard outside the case [to test for shorts - suggested by posters]

None of these things have solved the problem.

Does anyone have some suggestions on further steps to take? I could reinstall the OS, or starting using a discrete graphics card. Those seem like the only obvious options, but perhaps you guys have better insights than I do.

Thanks,

DK

 

[Tír na nÓg]

Have you tried mounting it on cardboard instead of the case?

It sounds like a short to me...

 

[dkanter]

Have you tried mounting it on cardboard instead of the case?

It sounds like a short to me...

Yes, I have tried removing from the case and running it on top of the bag. The same behavior was observed.

 

[LutaWicasa]

Boot and run with one stick....if no reboot, do the same with the other single stick.

 

[dkanter]

Boot and run with one stick....if no reboot, do the same with the other single stick.

I'm having problems in all configurations (stick 0, stick 1, both sticks). Can still boot into Windows, but getting machine check exceptions and it reboots itself within about 1-2 minutes of being in Windows.

 

[LutaWicasa]

Any other ram available to test with?
Noticed the Crucial you have isn't on their QVL

 

[dkanter]

I don't have any other DDR4 lying around.

The other thing I should mention is that the system was working fine for about 6 months, and then instability set in. I'll modify the top post to indicate the various things I've tried.

 

[Tír na nÓg]

Have you tried each ddr slot?

Try disabling the green stuff in bios (speedstep and C states).

 

[dkanter]

Have you tried each ddr slot?

Try disabling the green stuff in bios (speedstep and C states).

OK a question and an update:

Does it matter whether I use the A1 and B1 DIMM slots or A2 and B2?

Next, I found that disabling all Intel's power management gives me a more stable system. It is currently in Windows and has been running for half an hour. If this is the problem, is there anyway to get turbo, speedstep, etc. working again?

I believe I have the most recent BIOS update...

David

 

[Tír na nÓg]

Nope, doesn't matter.

You might lack a bit of juice at idle, that's why it reboots. Add 0.02v to the adaptive vCore base (it is you vCore when idling).

Check your vCore in CPUZ@idle, might be something between 0.85 and 1v (not sure about the Skylakes).

Then check on load.

In the bios you set the adaptive base to what your vCore is at idle. Substract load vCore - idle vCore (will be certainly something around 0.2v). Use this figure to set the adaptive offset.

That's for stock clocks.

 

[dkanter]

Nope, doesn't matter.

You might lack a bit of juice at idle, that's why it reboots. Add 0.02v to the adaptive vCore base (it is you vCore when idling).

Check your vCore in CPUZ@idle, might be something between 0.85 and 1v (not sure about the Skylakes).

Then check on load.

In the bios you set the adaptive base to what your vCore is at idle. Substract load vCore - idle vCore (will be certainly something around 0.2v). Use this figure to set the adaptive offset.

That's for stock clocks.

What is adaptive base? I'm pretty familiar with the design of DVFS systems. Do you mean Vmin (minimum voltage), or something specific to the Asus BIOS setup?

I don't really want to do any overclocking, perhaps I accidentally enabled it through the Asus BIOS?

DK

 

[Tír na nÓg]

Not too familiar with Asus bios anymore...

On my Asrock mobo, I set a "base" value for the adaptive voltage, which will be the value at idle. And then there is an offset value, which is the extra vCore to use when on load.

To confirm it is idle/vcore related, check what is the vcore on load (ceratinly around 1.1v), go into the bios, set vcore to manual, and put this value.

The green stuff won't work, but if after a couple of days your system is stable, it means that with the energy saving settings enabled, there is not enough juice to feed the CPU when idling.

If it is the case, it will give me a bit of time to have a look at you bios!

 

[dkanter]

Not too familiar with Asus bios anymore...

On my Asrock mobo, I set a "base" value for the adaptive voltage, which will be the value at idle. And then there is an offset value, which is the extra vCore to use when on load.

To confirm it is idle/vcore related, check what is the vcore on load (ceratinly around 1.1v), go into the bios, set vcore to manual, and put this value.

The green stuff won't work, but if after a couple of days your system is stable, it means that with the energy saving settings enabled, there is not enough juice to feed the CPU when idling.

If it is the case, it will give me a bit of time to have a look at you bios!

What processor do you have? Older processors might require some of that, but more recent Intel CPUs let the hardware power manager handle everything.

I'll check into what BIOS options I have, but generally I avoid touching anything relating to voltage. In theory, the vendors are supposed to get that right.

DK

 

[dkanter]

OK, I just tried to tinker around in the BIOS and when it restarted, it hung without booting into Windows. I hard reset and it then said "Overclocking failed, enter setup...".

I find this absolutely enraging as I don't want to overclock any part of my system. Do you have any idea how to set up the system for maximum stability?

 

[Tír na nÓg]

Well, I'd start with the saving features disabled. See if the rig is stable for a few days. That would confirm the problem is a too low vCore at idle.

If it's the case, changing the adaptive vcore settings doesn't mean you need to overclock. You can keep everything else at stock.

 

[wingman99]

I think since you have changed your motherboard and had the same results you should RMA your CPU, don't run it out of speciation or they wont RMA. Do you run the memory in XMP that is overclocking?

 

[Tír na nÓg]

I think since you have changed your motherboard and had the same results you should RMA your CPU, don't run it out of speciation or they wont RMA. Do you run the memory in XMP that is overclocking?

They'll never know unless you tell them!

 

[dkanter]

So I think I have the problem back to my SSD!!!! It's a seagate SATA SSD. I tried a clean win10 install on the Seagate SSD and it was still having the same problem. When I clean installed onto an Intel SSD, everything boots up again and I can use power management.

I'm very surprised and cannot figure out how the SSD would cause problems for the CPU...but that's what it seems to be.

Now, I still have a problem waking the system up and getting the display to come back to life, but power management works.