Z170 System: Four months of troubleshooting system freezes, idk what to do now.

[Highlord]

Hi, I’ve been having trouble with system freezes/hangs on my Z170 system for 4+ months and I’m soon at the point that I don’t know what to do. I try to divide this post into subcategories with current settings (voltage, clock, history and so on) so it can be easily read.

1. The issue
I’ve been experiencing system hangs (freezes) without BSOD (as in never) under general use (browsing internet (firefox, opera)) and gaming (mostly heroes of the storm). Note: browsing internet – Higher occurrence rate when watching streams (twitch.tv), I’m using adblock.
Occurrence rate: 3-10 times a week (I spend 8-10 hours at my computer a day).

2. The System
MB: Asus Maximus VIII Hero
CPU: I7 6700k @ 4,6 Ghz
CPU Cooling: CM Nepton 280L
Memory: Corsair Vengeance DDR4 LPX 2400Mhz 2x8GB @ XMP Stock
GPU: Sapphire Nitro 390 @ 1150/1500 (current), Asus 280X CU II @ 1130/1550 (old)
PSU: Corsair RM 1000i
SSD/HHD: Sandisk 250gb SSD, WD 3TB Black
OS: Windows 10 Pro 64bit

3. Motherboard – Asus Maximus VIII Hero
When I “first” ran into this issue (2-3 month into troubleshooting) I RMAed my MB and got a new one (same MB) and both MBs behave exactly the same.
PCI Express: Switched from auto --> Gen 2 --> Gen 3 --> Auto. No/little observed difference (Gen 2 seemed more stable).

Screenshots from bios:

4. CPU – Intel I7 6700k
When I first got the system running, I got to overclocking it. In the first phase I got 4,6 running at 1,395V (my thoughts were: ok… derp cpu), at a later phase I upped the memory voltage from 1,2V to 1,3V and got the CPU voltage down to 1,345V.
I have responded by upping voltage with no effect.

4.1. Stress testing:
My current setting:
CPU: 100x46 @ 1,345V (see images above) Load line calibration lvl 5, CPU Current Capacity 130%)
Corsair Vengeance LPX 2400Mhz 2x8GB @ XMP Stock 1,3V
OCCT: 14hours and 20 min (stopped due to error) (this should be stable enough for normal use? Jayztwocents)

Intel burn test: 51 rounds on Maximum
Screenshots of settings:

4.1.1 Temperatures:
As you can see from the screenshot above my worst case temps never go above 82C on core and package. I've read that skylake becomes unstable at around 85C. Running OCCT my temps never got above 76C and under normal use it rarely goes above 65C.

5. Memory
At my first OC phase I ended up lowering the memory speeds to 2100 (at some point as low as 1800 troubleshooting) to keep my CPU voltage down and if I remember right these setting were tested for 8+ hours in OCCT. As mentioned in the CPU subcategory I later found out that my system became a lot more stable at lower voltages (cpu core) when I upped my memory voltage from 1.2 to 1.3. Other modules of this memory from corsair (usually running at higher speeds) run at 1.35V so I don’t think 1.3V is a problem. I currently run stock settings (apart from voltage) 2400mhz with XMP enabled, I tried running without XMP and the issue persists.
I have tested my memory using the windows tool (Memory diagnostics) and no errors were found.

6. GPU
I originally thought it was my GPU overclocks that were freezing the system and I responded by lowering the clocks on both core and memory with no observed difference.

My 280X was OCed to 1130/1550 at max and lowered to 1100/1500 before changing it out.
My current 390 was at its peak 1175/1600 and later lowered to 1150/1500 (which are my current settings).

Both GPUs passed multiple runs of 3Dmark Vantage (with no observed artifacts) which according to TTL himself is a good indicator of a stable OC on your GPU.

6.1. Testing if Flash (adobe) was freezing my system.
I do experience freezes watching streams (twitch.tv) and responded by first migrating to a difference browser (firefox to opera) and at a later point disabling hardware accelleration on both browsers. I'm using adblock on both.

6.2. GPU temps
Doesn't go above 70C, I'm currently locking max framerate to eliminate extra heat.

7. PSU - Corsair RM 1000i
I replaced my PSU incase my problems were due to bad power delivery. Things doesn't get much better than Corsair on the PSU market. No observed difference.

8. SSD/HDD
Seems fine:

9. OS - Windows 10 Pro 64bit
Running the latest official build.
I have formatted my system once for troubleshooting.
Clean install from windows 10 iso

10. Current Software/Driver versions
AMD GPU driver: Updated 16.1 hotfix
Chipset: Newest
Sound: Newest
USB: Newest
SATA: Newest
BIOS: 1302 (newest)

11. Other
Tried different USB combinations troubleshooting damaged usb.
Inspected CPU socket (nothing abnormal to my eye)
Inspected CPU (nothing abnormal to my eye)

12. My Current thoughts on further troubleshooting
- Running 4,5 Ghz on CPU
- Playing with bus and multiplier (CPU)
- Run GPU on stock
- Run CPU on stock
- Uninstall my antivirus
- Uninstall Razer synapse
- Uninstall Logitech gaming software.

Closing words:
I want to specify that all the different setting I've run result in the same problem "system freezes". I'm sorry if this post became abit long I will edit this post if I've forgotten something and note it as an edit at the start of the post.

 

[MrKambo]

So you've done everything but run things stock?

I mean in my opinion sounds to me like you're jumping the gun with trying to get your OC stable, also if a stress test fails, its not stable regardless of 2 hours in or 20 hours in!

How about you give this a watch



Its not for Skylake but the basic outline and the fundamentals are the same, then after you've watched it, reset your BIOS and make sure everything is stable at stock.

Then, work your way up the multiplier getting it stable, you have to bare in mind your CPU may not be able to get 4.6GHz no matter how much voltage you pump through it Silicone Lottery thats known as!

 

[Highlord]

Thanks for replying! I will try to drop the multiplier down and see if the problem persists. I've been really stubborn on getting the OC to 4,6, but if this fixes it, I'll leave it be.

About "also if a stress test fails, its not stable regardless of 2 hours in or 20 hours in!" you're right, but most things will fail it's just a matter of time. Say that your stresstest fails on day 341, is this system stable? probably, but it's a matter of perspective (i.e. is the system supposed to be on constantly?). Where do you personally draw the line on stresstesting (stating that a desktop system is fully stable)?

 

[coffeefoot]

I would second Mr. MKambo's advice and take it one step further. Take everything down to bone stock No XMP. Just Plain default settings. start with one stick of memory. Take out all but your boot drive. Remove your graphics card and use the onboard graphics. Boot up and go from there. See how long that is stable. Your system should continue unless there is a physical failure.
add one piece at a time. Beyond that, there are more knowledgeable blokes here who can help you.
Time consuming ? You've already spent 4 months on this, no ?

 

[Highlord]

Coffeefoot, what you are describing is certanly an option, however abit drastic for now. It would be the tidiest way of doing it, tho quite timeconsuming and I therefore see it as a last resort. I would also have to run the system for a day or two stripped down of it's features (gaming, my storage drive and so on).
It certainly could be the way to go if the problem persists.

 

[TheF34RChannel]

I would really follow their advice, it's solid mate easiest way to find the issue. Unplugging some cables isn't that time consuming

 

[AlienALX]

http://www.bit-tech.net/news/hardware/2016/01/11/intel-skylake-crash-erratum/1

I would wait until that is fixed and then update your bios and see if it makes a difference. I did post this in Quick News but it seems a massive Skylake bug is not big enough to make the front page. Either that or WYP didn't notice it.

And I would assume that it is also the cause of several other threads I have seen lately, all oddly enough on Asus boards.

 

[TheF34RChannel]

I don't think that issue is the cause of too many their problems and it's not a big issue either. Not many applications use AVX and games not at all; 99% of people will never run into the issue, and you can run prime just fine even, as I understand the issue pertains to a very specific set. That's what I got from it all anyway, could be wrong

I am interested in your mentioning seeing a lot of Asus boards related issues; similar issues?

 

[Davva2004]

Thanks for replying! I will try to drop the multiplier down and see if the problem persists. I've been really stubborn on getting the OC to 4,6, but if this fixes it, I'll leave it be.

About "also if a stress test fails, its not stable regardless of 2 hours in or 20 hours in!" you're right, but most things will fail it's just a matter of time. Say that your stresstest fails on day 341, is this system stable?

No, it's not. Stable means it can run indefinitely without a single component failing due to overheating or overvoltage. Everything is working together and will continue to do so indefinitely.

In the real world, stable means it passes all stress tests without errors.

 

[TheF34RChannel]

^ this!

I also think the op can only progress if he takes all OCs completely off and components as well and test them one by one; this is no time to be stubborn mate

 

[coffeefoot]

Time consuming ? you've already spent 4 months on this.