Multiple CPUs failing with same Mobo and PSU. Bad OC?

Elliott_ch

Commendable
Dec 10, 2016
6
0
1,510
Hey guys,

Posted here a little while back when my CPU died the first time and got a lot of great advice so thanks again for that. Long story short:

i7 5820k/i7 6800k
Gigabyte x99p-SLI
Corsair H100i
Samsung Evo 850 pro 500gb
Corsair Vengeance 24gb RAM
Corsair RM650i PSU

Had a 5820k running in my Gigabyte x99p-SLI. This was overclocked to 4.4ghz I believe, can't remember the voltage but RealTemp said it never exceeded 65 degrees during stress tests. After a few weeks of heavy use in audio production, the computer began hanging, then one of the screens stopped working, then blue screens, until finally the computer would not get to post and was in a permanent boot cycle. I reset BIOS settings to optimized defaults before the final blue screens, which either made no difference to what was causing the damage, or the damage had already been caused. After extensive testing and troubleshooting, I ended up taking it back to the shop I bought the components from and they diagnosed it as a dead CPU and offered me to upgrade to a 6800k for $50, rather than going through the whole RMA process.


After having installed the new 6800k, everything is back to normal, no issues for a couple of weeks. I had a project in which I had a lot of processing going on, which relied on the overclocked CPU to be able to playback audio in the session. I went through the OCing process again, and settled on 4.5ghz at 1.5v which was stable and temps never went above 69 degrees in stress testing. A week a two later brings us to now, and the same symptoms as before have happened. This time I managed to grab the blue screen error code before the computer stopped being able to POST. "Crypto Library Internal Error". A bit of research suggests this is again a CPU failure, as well as the fact that it is exactly the same story as last time. I have a few ideas as to what could be causing it and I'd really appreciate the input of all you guys.

I feel like 1.5v is quite a lot of voltage, and 4.5ghz is quite high for a 6800k. RealTemp and CPU-Z both read the correct clock speed and say the CPUs aren't overheating, but is there a chance that RealTemp was inaccurate, and both times I have been overheating the CPU, but not to the point that it shuts itself off and never knew about it? Or is it more likely that the PSU or MOBO have an issue? Also, what's the likelihood of receiving two CPUs that were both faulty?

I need to get this rig back up and running for work. My solution at this point is to replace the CPU one more time, see what I can get done without the overclock, and wait until the shop's technician returns from holidays and is available to test all the components.


Many thanks, I look forward to any and all replies.

Cheers

Elliott



 
Solution
Haswell-E you should keep it below 1.35v at 80C, Broadwell-E is the same way, keep below 1.35v at 80C. If you are on custom watercooling and able to keep it below 80C you can raise the voltage a bit more than 1.35v just make sure temps are in check or you risk messing up the chip or degrading its life span significantly.

1.5v on the core is way to much if you ask me, I remember some of the Asus board had a bug that would for some reason drive the CPU voltage to 1.5v when x99 was new, killed a few peoples chips, and others survived it.

A i7 5820k typically overclocks further than the 6800k with much less voltage, Im at 4.6ghz currently at just 1.33v ranging into the low 70c on the H100I GTX. I won't go any higher since my temps are...

Atomicdonut17

Reputable
Feb 4, 2017
737
0
5,360
It would be safe to replace it, and for the beginning of its life just use stock speed. However, if you're having to absolutely OC it, don't OC it as high as before. As well, a higher grade cooler is recommended, as I've heard of several instances of the H100i's pump failing within 6 months and causing leaks/overheating.
 
Haswell-E you should keep it below 1.35v at 80C, Broadwell-E is the same way, keep below 1.35v at 80C. If you are on custom watercooling and able to keep it below 80C you can raise the voltage a bit more than 1.35v just make sure temps are in check or you risk messing up the chip or degrading its life span significantly.

1.5v on the core is way to much if you ask me, I remember some of the Asus board had a bug that would for some reason drive the CPU voltage to 1.5v when x99 was new, killed a few peoples chips, and others survived it.

A i7 5820k typically overclocks further than the 6800k with much less voltage, Im at 4.6ghz currently at just 1.33v ranging into the low 70c on the H100I GTX. I won't go any higher since my temps are already sorta of pushing it and summer is right around the corner. The performance difference between the 5820k and the 6800k is not vary big and is not worth really looking at if you can get a 5820k. Stock frequency sure the 6800k will always win but its x99, I'd at least push to 4ghz.

Replace the chip and keep it under 1.35v for the time being, if you need the clock speed and can get the 5820k again, I'd go for that, 9 times out of 10 the 5820k will out perform the 6800k when overclocking, also look into updating your bios, there can be some fixes depending on how old your current bios is.
 
Solution

danielthegreate

Prominent
Apr 4, 2017
113
0
760
Chips degrade when you apply higher than the rated voltage. The degradation is faster in the beginning and then slows down a bit, so that's why when you push a chip too far it tends to bluescreen in the first couple of months, but if it doesn't you are usually good for a few years.
I wouldn't go above 1.35v for Broadwell-E, most chips should be able to hit 4.3 at that voltage and the gains above that aren't worth the voltage increase.