MSI Z170I Gaming Pro AC + i7-6700K freezing - Motherboard fries

[kaans]

Hi everyone,

 

I'm not sure how to approach this issue, so I decided to ask the ltt community

 

I built a system with these parts: http://pcpartpicker.com/p/MqnZYJ (really regret the Raid0 by the way)

 

After the Win10 installation, I installed the intel drivers, made the win10 updates, updated the nvidia drivers, ran 3DMark, the program crashed frequently, then the entire system crashed, at first I approached the issue as a GPU issue, however I'm 99% sure it's a CPU/Motherboard issue now, as I also get freezes when I boot to the Win10 install disk, and one time at bios

 

Researching things more, I realised that my ~85C temps at Cinebench weren't normal, turning off XMP pulled things to 70C's, manually setting voltage pulled things further back to 50-65C's

 

TL;DR: The motherboard's auto clock/voltage likely fried the CPU

More info here: https://forum-en.msi.com/index.php?topic=262420

 

Manually setting the VCore to 1.2 makes things really stable, however I got 2 more random shutdowns, mainly while playing Broforce this happens

 

Currently, I set everything manually to defaults, including all CPU/DRAM voltages, the "Auto" actually means that the motherboard is going to overvoltage things to his taste

 

What do you think about all these? Is it normal for motherboards to push a 95W/1.2V CPU to 120W/1.4V's

 

I bought my motherboard/CPU locally, they arrived 10 days ago, I might have 3-4 days to return replace them

I bought everything else from Amazon US internationally, they arrived 3 days ago

Both are very tedious processes, I just wanted to build and use a PC happily for once, but that didn't happen

 

Memtest86 (Edited)

I also 1-runned an 13 test Memtest86 at the above stable setup, there were 0 errors on 13 tests

I also ran it again using 8 CPU's with 10 tests, CPU3 throwed errors

When I stress test with Prime95, sometimes windows starts continually throwing DLL errors, I thought memory might be bad too, but it doesn't seem to be, probably all those random shutdowns screwed the win10 installation too, or it could be the regular behaviour of win10, in any case, I don't think the issue is OS related

[Master Disaster]

I've heard of this issue before, on this very forum, with an MSI board and a 6600K.

 

Some people claimed that user just got unlucky and purchased a very bad CPU which needs the extra Vcore while others said it was a faulty board.

 

I don't think that issue ever got fully resolved.

 

Have you updated the board to the latest UEFI?

 

As a reference point my 6700K will happily run at 1.24v at stock, at 1.4v I can push 4.7Ghz with no issues at all and even 4.8Ghz with everything except Cinebench seeming to be stable and I still don't get temps anywhere close to 85c.

[kaans]

I tested things further, moved the RAM's to another identical build, they failed there too, so I kind of eliminated CPU

 

It's either a motherboard shortcoming - or simply faulty RAM's

 

The challenging thing is, I can't single out a RAM stick, together they fail, individually they pass, which makes things complicated, I guess I'm just going to buy some DDR4's locally and see how it goes, there seems to be some HyperX Savage's available

[phoenix230197]

Just out of curiosity, what about thermal compound? Enough / not enough / too much?

Possibly a bad cup. Maybe take the whole thing to the place you got the cpu and mobo from and get them to test it for doa.

[kaans]

I'm pretty sure their testing wouldn't doa it, from my experience such places rarely detect issues well

 

Currently found a way to replicate the issue fast, run prime95 for 20 minutes, run memtest test6/7's with parallel CPU's - errors occur at phases 1-3 fast, before this, I ran 10 phases, no issues occurred, so it's kind of a persistent cumulative absurd issue

 

I think I applied my Noctua NH-D9L pretty well, watched countless youtube videos before doing anything, when the CPU is capped at 1.23V, the temps are around 60C with prime95 at high load

However, with XMP and Auto voltages, they reach 90C, which I consider a motherboard flaw, enabling XMP adds 15-20C to the temps, consistent on both systems

 

 

The same process doesn't produce errors on the second system that is identical, so It's probably not a mothorboard design flaw, but likely a faulty cpu/motherboard/ram

 

Going to buy some Kingston HyperX Savage's locally that should arrive tomorrow, as far as MSI, it's the RAM's that are faulty, they say they are aimed at X99 that are not guaranteed to work well at Z170, I kind of agree, as there are other reports of similar issues, yet a RAM such as Corsair Dominator should keep on working no matter what considering the price tag and hype