i7-6850k Easily degraded (My fourth replacement)

[Dukelaking]

I run an x99 asus strix board, 32 gb of kingston hyper X ram.
My i7-6850k suffered heavy degradation over a week somehow at 1.3 volts at 4.2 ghz. I found out it was going bad because chrome and any application would not work without crashing.

I was wondering if anyone else had issues with the i7-6850k.
My replacement is able to do 4.5ghz at 1.33 volts stable. Temps stay acceptable at around 90 degrees max on one core and average 79 degrees across all cores with a custom water loop when running Intel burn test at maximum load.

My first i7-6850k was faulty but I did now know from the get go, and I ended up rmaing my ram multiple times and swapping power supplies and gpus. Then finally rmaing my cpu which solved the problem.

My 2nd i7-6850k just couldn't overclock at all, it was ridiculous so I got an advance replacement.

My 3rd i7-6850k overclocked to 4.4ghz right off the bat, then that became unstable after 2 months, so I lowered it to 4.2 ghz and then after a good 8 months it degraded heavily.

Now i'm onto my 4rth and I'm hoping it will last me with no issues. Intel has been really helpful and let me skip all the troubleshooting measures and even offered me a refund. I've never had a processor degrade like this before. My i7-4790k used to clock at 5ghz at a low voltage for 2 years. I have my doubts about my motherboard possibly being the culprit killing the cpus, but asus has not helped me at all they put me around in circles.

Has anyone else experienced this?

 

[XavierMace]

My replacement is able to do 4.5ghz at 1.33 volts stable. Temps stay acceptable at around 90 degrees max on one core and average 79 degrees across all cores with a custom water loop when running Intel burn test at maximum load.

I would personally never consider a CPU running that hot "acceptable".

 

[Ken g6]

Have you ever heard that the definition of insanity is trying the same thing over and over and expecting a different result? Have you ever considered trying a different chip? Like a 5820k, or another 5000-series chip? I never recommended those 6000-series chips. They seemed to have no benefit over the previous generation.

 

[Indus]

Hrm.. I wonder if intel's 14nm cpu's are all going to degrade badly considering the heat.

My 22nm 3470 never went above 75C and never degraded.

My 14nm Ryzen is staying well below 60C even overlocked.

So I'm thinking its the temps creating the problem.. 90C is high.

What kinda cooler have you been using? Noctua?

 

[tamz_msc]

79 degrees on IBT would mean 85 degrees on P95 or Linx, or something around that? Plus I've heard things like Asus X99 boards allowing modifiable Tjmax, so temperature readings for the throttling point could be off. Also, this is what the Asus rep at OC.net has to say:

It's simply a variable range, with a min and max that are defined by Intel. It allows you to fine tune the throttling point of the CPU. That's all. I suspect your confusion stems from the assumption that the value is being overridden to an extent that breaches Intel guidelines. Not the case, as they still define the absolute max it can be set to. If one is overclocking the system, a lot of this guidelines/safety stuff is out of the window, anyway. You can degrade a chip without hitting TjMax.

 

[Dukelaking]

Well 90 degrees is during intel burn test at maximum stress levels for that one odd core. During gaming it stays below 50 degrees in comparison. I'm still tweaking the voltages lower and I might need to reapply thermal paste for a more even application. My cpu doesn't throttle and it passed IBT. I'm using a custom watercooling loop, with both a 360mm rad and 240mm rad just for the cpu. It's overkill I know, I used to have two r9 290s in the loop until I upgraded to a gtx 1080. I might just lower the OC to 4.4ghz with a lower voltage to be safe anyways.

Pictures of rig is shown here.
https://imgur.com/a/37faE

 

[Dukelaking]

I got my i7-6850k for only 400$ cad from a dude who was an intel retail edge member a year ago. I rather not swap my chip for a previous gen and the 6850k already exceeds my needs for productivity and gaming. The chip I got is part of the top 10% in terms of overclocking performance. When overclocked it outdoes the stock 5960X by a bit. Comparing passmark scores and cpu-z scores.

 

[Charlie22911]

Yes, and you will probably notice that it is the OC socket equipped ASUS boards doing most of the killing; my x99A USB 3.1 has killed a couple before figuring it out. No issues with my other MSI x99 rig.
Follow guidelines on safe voltages and set everything manually, don’t use XMP. You shouldn’t have a problem if you do that.

Also 1.275v should be about enough for 4.2GHz. Anything over 4.2GHz heats up quick and is not worth the extra voltage, heat, and power consumption.

 

[ZGR]

Yes, and you will probably notice that it is the OC socket equipped ASUS boards doing most of the killing; my x99A USB 3.1 has killed a couple before figuring it out. No issues with my other MSI x99 rig.
Follow guidelines on safe voltages and set everything manually, don’t use XMP. You shouldn’t have a problem if you do that.

Also 1.275v should be about enough for 4.2GHz. Anything over 4.2GHz heats up quick and is not worth the extra voltage, heat, and power consumption.

I heard about this. I believe you are right. Asus really messed up here with X99! OP please follow Charlie's advice or change motherboards (not another X99 from Asus).

Hrm.. I wonder if intel's 14nm cpu's are all going to degrade badly considering the heat.

I've been running my 5775C at 4.2 GHz at nearly 1.37v for a long, long time, but my Noctua DH15 keeps it nice and cool below 70c at max load. That is the first gen 14nm correct?

 

[Shmee]

My 5930k has been fine with my Asus OC socket, but I never pushed over 1.3Vcore, and I have a HW-E not a BW-E. Even at 4.6 GHz, though that was not stable.

 

[Insomniator]

Never heard of it, been running my 6850k at 4.2 for a couple months with no issues (gigabyte board). 95 and 79 is really hot though. Don't think I've ever seen this CPU in the 90's. P95 is about 65 for me.

 

[LTC8K6]

He's counting the second one simply because it didn't overclock well...

 

[Dufus]

79 degrees on IBT would mean 85 degrees on P95 or Linx, or something around that? Plus I've heard things like Asus X99 boards allowing modifiable Tjmax, so temperature readings for the throttling point could be off. Also, this is what the Asus rep at OC.net has to say:

My old i7 6800k could be set to a Tjmax of 120C. Good to know running at 119C is perfectly fine.

As for killing chips IMHO the degradation would point to user error or bad CPU. Given the number of failed CPU's it would err to the former.

 

[Charlie22911]

My old i7 6800k could be set to a Tjmax of 120C. Good to know running at 119C is perfectly fine.

As for killing chips IMHO the degradation would point to user error or bad CPU. Given the number of failed CPU's it would err to the former.

I can't vouch for OP, but in my case it was two separate ASUS boards and two separate CPUs. This was for my first unRAID build so it was not overclocked, everything was set to stock to ensure stability. IIRC the ASUS board was setting my system agent voltage far higher than it should have been, causing degradation and eventual failure to boot reliably; even when set to stock it did this. Haswell-E is more tolerant to higher voltage than Broadwell-E, which simply gets hotter with more voltage without much gain*.

*from my experience with 3 6850k and 2 6900k.

 

[IEC]

I am not surprised to see it's an Asus board. They are responsible for an abnormally high number of dead chips on the X99 platform.

See very long OCN thread with many reports of dead CPUs. Some even killed >6 CPUs before figuring out what the common factor was...:
http://www.overclock.net/t/1561131/5960x-dead

And one on Guru3D with your motherboard as the culprit:
https://forums.guru3d.com/threads/killed-a-6950x.412720/

 

[Madpacket]

Had a MSI X99 board that was stable enough but my 5930K was a dud (4.0/4.1Ghz max overclock). Glad I got rid of that system but looking at Intel's newer HEDT wares makes me sad.

 

[weevilone]

My Asus board will automatically overvolt my 8700k CPU at default settings, so I'm a little disappointed. I loaded things up and was getting Windows and my software squared away before doing any overclocking. Little did I know Asus had everything locked at 4.7 with Vcore, and the memory controller voltages crazy high.

 

[Dukelaking]

My 2nd cpu couldn't even reach 4.2 ghz stable. I wasn't trying to settle on a "sub-par" chip. I bought the platform to overclock and make use of my watercooling loop lol.

 

[LTC8K6]

My 2nd cpu couldn't even reach 4.2 ghz stable. I wasn't trying to settle on a <redacted> chip. I bought the platform to overclock and make use of my watercooling loop lol.

But it wasn't "degraded" and there was nothing wrong with it, so "fourth" is inaccurate.

All cores at 79 degrees with a custom water loop doesn't make me think the chips are bad. It makes me think something is wrong elsewhere, such as unintended overvolting.

 

[Charlie22911]

But it wasn't "degraded" and there was nothing wrong with it, so "fourth" is inaccurate.

All cores at 79 degrees with a custom water loop doesn't make me think the chips are bad. It makes me think something is wrong elsewhere, such as unintended overvolting.

Depends on the load. With a full custom loop In Prime 95 my setup hits mid-high 70s, Broadwell was not made to be a high clocking high performance part; that's just the way it is; even the 5775C was a terrible overclocker.

https://www.anandtech.com/show/9482/intel-broadwell-pt2-overclocking-ipc/2

 

[Headfoot]

I've never had a faulty CPU or had a CPU degrade more than a few hundred MHz max. The commonality amongst all failures is the same Motherboard. Its extraordinarily unlikely you'd get 4 bad/bad-ish CPUs. Dollars to donuts your motherboard is the problem here. I've had zero CPU problems but I've had at least 2 different ASUS motherboard failures...

The eDRAM broadwell is a terrible comparison point for overclocking. It's clearly quite different being the only chip with embedded eDRAM....

 

[Charlie22911]

I've never had a faulty CPU or had a CPU degrade more than a few hundred MHz max. The commonality amongst all failures is the same Motherboard. Its extraordinarily unlikely you'd get 4 bad/bad-ish CPUs. Dollars to donuts your motherboard is the problem here. I've had zero CPU problems but I've had at least 2 different ASUS motherboard failures...

The eDRAM broadwell is a terrible comparison point for overclocking. It's clearly quite different being the only chip with embedded eDRAM....

I think we can all agrees ASUS is the culprit here.
As for the comparison, my intuition says eDRAM has little bearing on the ability of the core to overclock. I concede I may very well be wrong, though I think it unlikely that the multiple BW-E CPUs I’ve had hands on have had the exact same results as in the article. The average for stability with Broadwell in general seems to be around 4.2 with anything above being above average.
Going above 4.2 results in huge increases in power consumption and voltage requirements regardless of ASIC quality and stability.
But that really isn’t the point of this topic, so I won’t continue to derail it.

 

[LTC8K6]

5775C was a 65W TDP chip...and a typical overclock was 4.2ghz, pretty good given it's low standard clocks, the edram die under the heat spreader, and it's TDP.

 

[Insomniator]

Even the best BW chips get what, 4.5? Who cares if you only get 4.2... that's an embarrassing reason to send back to Intel.

 

[Dukelaking]

Well stock turbo is 4ghz. And it couldn't maintain 4.2ghz regardless of the voltage. It beats the purpose of a K series processor if I can't overclock it since I would keep it stock then. I decided to send it back and that was my decision. You're entitled to your opinion though. It was either that or I sell the processor and to get a different one.