6700K + AIO (sudden, dramatic overheating)

[Wilk0]

Strap in, this is a tedious one.

Silverstone Tundra TD03-lite (AIO short loop cooler, single-120mm-fan-supporting rad)
Scythe Silent Wings 3 (for static pressure against AIO radiator, auto modulation on CPU fan pin)
Thermal Grizzly Kryonaut
Intel 6700K (never delided) (seems to be a fairly good silicon lottery sample?)
Silverstone FT05 (rotated design chassis, using both 2x200mm bottom fans at various speeds)
R9 380 4GB (not much heat coming from this) / (formerly) 5970 HD (lots of heat came from this)

Upon first ever installation the performance was excellent with the 6700K at stock speeds and 1.2v voltage never passing 60 degree centigrade in late summer after prolonged and various stress tests. It was very impressive.

I cannot remember the perfectly-complete history of maintenance regarding this setup unfortunately, but I believe that my story will be sufficiently complete. I have replaced the paste at least once before my following problematic experiences. After said replacement of the paste I did forget to use a coffee filters, instead using some other material likely paper towels, to clean the cooling block so it has lost some of its sheen, this may have affected cooling slightly. However, using what I could remember of the same technique and guidance from online I found my cooler to again be at least very sufficient, with 70c being seen never even in tests unless I forget to turn the case airflow on or lower the pump speed too far.

Rather suddenly, twice, in the last month - I've found my CPU overheating, even to the point of shutdown. I was even overheating when idle and in BIOS, heat was clearly accumulating over time and eventually caused the system to shut down even in windows CPU speed capped at 25% (Needed my PC for some things). This was very odd as it happened suddenly with no physical disturbances to the system. I was perplexed, seeking any answer other than physical... After acquiring some new 99.9% alcohol as a volatile solvent, I used coffee filters to thoroughly clean the CPU and block then re-applied my paste and subsequently firmly secured the block to the CPU. The temperatures were then down again to 25c in idle as normal and some reasonable values when worked... Wow OK, I guess it was somehow, suddenly, a thermal interface issue.
A week or two later, I heard my CPU fan whirring loudly, what could this be? Temperatures were approaching 70s... 80s in games, 90 in a memory stress test, a week ago idling at 40. It was not crashing, but I wasn't happy with the temperatures of course. A bad paste application? (again(?)) But it worsened over time?
It's complex to describe my observations, the progression of the temperatures and the reading and temperatures on other components however I believe that:
This is probably not the fault of:
-cooling solution/setup/airflow (including dust) being insufficient (I've seen it performance excellently)
-the iGPU being wasteful (I've tested with it on and off)
-VRMs being broken (I'm seeing no strange voltages or power usage on the mobo or GPU)
-cooler (pump) or fans being broken nor unreliable (the behaviour is not erratic in that sense)
-sensors being broken (I've put my finger on the back of the mobo, it is certainly overheating)(although core #1 (2nd of 4) on my CPU consistently reads up to 5c hotter than the rest, and the whole set often jumps up and down by 10c between my 1sec set poll time)
Considerations for diagnosis:
Fan and pump speeds are all 
-Skylake is thinner than its predecessors which this cooler was designed for (Tundra 03 was released before Skylake) (although it feels stationary, tight and secure)(I noticed that when installing a H212+ on a friend's 6600k that the cooler was very loose and had to be tightened significantly but still wiggled)
-Cooler could be too tight, squeezing the TIM out (I have no way of knowing)
-Paste could be too little, or too much, or a combination with the above.
-There could be airpockets in the paste (I have no way of knowing but clearly I've successfully avoided them in the past and otherwise applied the paste well)
-I've tried stock motherboard settings and this has no positive affect on the thermal performance
-The bolts that pass through the motherboard of the tundra 03 are a slightly troublesome design, as they can easily rotate in their provided bracket, causing the block to sit unevenly. (However I am aware of this and am especially careful to install this cooler correctly every time)
-Recent intel security patches affect performance (this is really a long shot but I haven't tested my system with hyperthreading off)
-Being an AIO, waterbubbles in the loop seem fairly unlikely.
-I've had some BSOD lately, can't play a game (ARMA 3) as it causes a crash (which did scramble some windows settings on occasion, such as my start menu layout), very recently my PC is crashing when I'm AFK but never when I'm sitting at it, strangely. although I've seen no bluescreens since the zombieload etc. windows patch though I suspect no relation, could easily have been an audio driver.

Please can anyone you offer some advice on what could be suddenly wrong with my set up? I'm relatively certain that the issue concerns mechanical design and installation of the cooler which I'm unable to inspect, nor form insight on, nor assess.
Hope this is at least slightly interesting to someone out there. Holler if I need to clarify something.
Update, 3rd June – Idling at sub 30, even falling to sub 30 after torture. Stressing my 6700k: with CPUz and MemTest is resulting in 65c after 5 min, up to 88c after an hour with all fans and pumps spinning up to their highest. I've heard of mechanical materials (TIM) needing to settle but I retain suspicions that a sudden and critical situation might occur again... Unless the recent windows patch addressing zombieload really did resolve something eheh...

[jones177]

Make sure multicore enhancement did not get put on by accident in the Bios. My i7 6700k would get crazy hot even with a Noctua NH-D15 with it enabled. 

 

My i7 6700k also seemed to be more affected by ambient temperatures than other CPUs I have owned. When it was in the living room it was fine. When I put it in a bedroom it got hot.  I ended up having to open the case and use a floor fan to blow air into it. 

 

When I replaced the i7 6700k with a i7 8086k all the heat issues went away. 

[SpookyCitrus]

Possibly a bad pump, how old is the AIO?

[Wilk0]

12 hours ago, SpookyCitrus said:

Possibly a bad pump, how old is the AIO?

Ah, good question. Pump, mobo, CPU, case, PSU, 16gb of the ram - all bought in late 2015 and enjoyed heavy usage. CPU fan is newer.
I keep everything constantly under the watchful eye of HWinfo64 with colour coding and alerts for bad values and have never seen the pump to misbehave, whether using default PWM or a custom reactive schema.

13 hours ago, jones177 said:

"Make sure multicore enhancement did not get put on by accident in the Bios."

Interesting, I'll investigate this setting.

13 hours ago, jones177 said:

"My i7 6700k also seemed to be more affected by ambient temperatures than other CPUs "...

"When I replaced the i7 6700k with a i7 8086k all the heat issues went away. "

That's a curious observation. There could be a few explanations though. Different conductivity (mobo to cpu) / heat build up / heat dissipation / airflow exposure patterns, particularly of the motherboard. It does seem like something that would only improve with hardware generations.
But perhaps most relevantly: I have a pretty tight case with a strong airflow from bottom to top, though no airflow behind the motherboard. I also have an internal ambient air temp sensor clipped to the pump pipe, I don't check it often but can't remember it passing the very low 30s, pretty much the upper limit around here. It can't account for humidity of course but England in a city is pretty dry. After inspecting some logs I see that case air temp was 24/25c on days when critical overheats occurred.