My 6850X just died! Just like that!

[Dufus]

Maybe better to say "operating out of specification" rather than overclocking.

Reminds me of a P67 BIOS bug that when loading a profile could set VCore offset to 0.8V potentially supplying over 2V to the CPU core.

 

[PingSpike]

I was reading a thread on another forum where Intel denied a warranty claim for a cpu because of running XMP.

. . . wat?

https://www.reddit.com/r/hardware/comments/5p891v/tfw_when_intel_support_tells_you_xmp_may_void_the/

Just noticed the OP deleted his post. But the discussion was much the same as this thread has.

 

[rchunter]

I don't have xmp enabled on my board anyway. My registered ecc is default clocked at 2400mhz and that's what I run it at. My bios recognized all the correct timings automatically even without xmp enabled.

 

[dave_the_nerd]

I seem to be the only person who's done enough desktop support/repair that I've also seen CPU failures.

It's rare but it happens. Sell a billion of something, you'll get a few.

 

[.vodka]

I seem to be the only person who's done enough desktop support/repair that I've also seen CPU failures.

It's rare but it happens. Sell a billion of something, you'll get a few.

Sure, there are cases here and there, I've had a skt 754 Sempron die in a client's machine (yeah, they seemed to be getting good mileage out of that.. I don't understand how).

Apart from the anecdotes, you can't deny there's a LOT more cases of dead Broadwell-E CPUs for whatever reason over what's usual, apart from ASUS' suspect "OC socket"

Maybe better to say "operating out of specification" rather than overclocking.

Reminds me of a P67 BIOS bug that when loading a profile could set VCore offset to 0.8V potentially supplying over 2V to the CPU core.

Yeah, that's a nasty one.

I double check CPU voltage on my P67 motherboard after reloading or saving profiles. Better safe than sorrry.

 

[bigboxes]

I seem to be the only person who's done enough desktop support/repair that I've also seen CPU failures.

It's rare but it happens. Sell a billion of something, you'll get a few.

You're not alone. I've run across dead CPUs in my work as a pc technician. I've just never had any of my CPUs die. All the other parts, just not the CPU. It's rare for that to happen.

 

[Dufus]

Yeah, that's a nasty one.

I double check CPU voltage on my P67 motherboard after reloading or saving profiles. Better safe than sorrry.

I use to set negative instead of positive since a -0.8V offset would presumably cause a crash instead of CPU death and the offset set to 5, never trusted 'auto' either. Extra voltage was done by 'additional turbo voltage'.

Anyway was just meant as an example of how a CPU could be killed without any frequency overclocking and would I imagine leave most people wondering what the heck just happened.

An excerpt from a 3 year boxed Intel processor warranty http://download.intel.com/support/processors/sb/Limited_Warranty_8.5x11_for_Web_English.pdf

Quote:
WHAT THIS LIMITED WARRANTY DOES NOT COVER:
• design defects or errors in the Product (Errata). Contact Intel for information on characterized errata.
• any costs associated with the repair or replacement of the Product, including costs of removal or replacement of the Product;
• damage to the Product, or errors or malfunctions in the Product, due to accident, abnormal electrical, mechanical or environmental conditions, use contrary to product instructions, misuse, neglect, alteration, mishandling, repair, improper installation or testing, combinations with incompatible products or any third party virus, infection, worm or similar malicious code;
• that the Product will protect against all possible security threats, including intentional misconduct by third parties;
• any Product which has been modified or operated outside of Intel’s publicly available specifications, including where clock frequencies or voltages have been altered, or where the original identification markings have been removed, altered or obliterated. Intel assumes no responsibility that the Product, including if used with altered clock frequencies or voltages, will be fit for any particular purpose and will not cause any damage or injury.

 

[lopri]

I've killed a total of 3 6850s, one was potentially my fault in my gaming rig w/MSI board.

The other two were killed by an ASUS board in my unRAID server. RMAd the board and CPUs and now I run it with every manual setting changed from AUTO to be within intel specifications, as well as disabling things like MCT.

Now it racks up months of uptime without issue between reboots/updates.

OP,do you by chance also have an ASUS board? If so which one?

Killing 3 CPUs of a kind on 2 different boards? That is quite a feat.

 

[vissarix]

In all my years of building, using and working with PC's I have never had a CPU fail just like that. I wend down to my PC this morning and noticed the computer was off. Trying to start it up, the ASUS Deluxe II just have a 00 on the q code and a red light saying CPU failure.
Changed the power supply, removed all ram etc, no difference.
I had no choice but to go out and buy a new motherboard and CPU to find where the issues was.
Started with the motherboard. No difference! Swapped the CPU and all is now OK.
The only overclock on this was the XMP profile from the ram. Nothing else. Voltage stock.
Just how common is this?
Tom

You went out and buyed a new motherboard and a cpu just to see what was the problem just like that?

and after all that excellent move you came here to ask what exactly?

dude you should get an award

 

[VirtualLarry]

You went out and buyed a new motherboard and a cpu just to see what was the problem just like that?

Some of us, er, have "backup rigs". Partially, so we don't have to go out suddenly and buy expensive parts for our "main" rigs. We can use the "backup" rig, to compute, while the parts from the "main" rig are being RMAed.

Then again, if your professional work depends on having a certain minimum horsepower in your rig, then that solution might not work for you.

 

[LTC8K6]

Yes, I have backup CPU's, backup boards, backup ram, backup PS, etc.

I build computers mainly as my hobby. I always have all sorts of spare parts laying around.

I generally have 2 or three whole computers laying around.

When a computer stops working, I can quickly find out which part is the problem.

When my few months old 4790K quit on me, I couldn't believe it either. But since I had other Haswell chips, and a spare board, it was quickly clear that the 4790K had indeed given up the ghost.

Fortunately the RMA was fast and painless and the replacement chip is still going strong.

 

[tomb18]

I'm the OP. I could not afford the down time on that system. Although I regularly do backups I hadn't done one in a few days and I was coding something quite involved. I didn't want to wait to find out where the issue was. Motherboard? Or CPU? Good question. I also use an M2 and U2 SSD, and my other systems do not take these drives. So I couldn't just swap the disk into another machine.
As it turns out I just received my replacement Intel CPU today. This is how it went:
Filled out the warranty request at Intel on Sunday. I had to provide various numbers from the CPU and the serial number that is only on the box. Only boxed processors are guaranteed. Luckily I still had the box.
Intel replied late Tuesday. They asked questions concerning how I troubleshooted this. They asked the make and model of the power supply and the ram. They didn't ask about the motherboard!
On Wednesday, I received an email from the warranty department with instructions and they asked if I wanted it cross-shipped. I chose the cross shipping. They charge you the retail price of the processor and $25 for two way courrier. You have 30 days to return it and get your money back. I got it today, Thursday.
Pretty fast! I'm sure if I had to send ASUS my motherboard, that it would be way way longer.
Now, I will be sitting there and wondering if the motherboard had anything to do with this or if I just had the one in a million CPU that just died.
As to having a spare machine....not so easy. I have a visual studio pro installation, along with licensed software libraries, one of which is tied to that machine. So yes, I could just swap out the U2 and M2 disks as long as I have a motherboard that can take these.
I guess I now have one...and a spare CPU.

 

[Puffnstuff]

In all my years of building, using and working with PC's I have never had a CPU fail just like that. I wend down to my PC this morning and noticed the computer was off. Trying to start it up, the ASUS Deluxe II just have a 00 on the q code and a red light saying CPU failure.
Changed the power supply, removed all ram etc, no difference.
I had no choice but to go out and buy a new motherboard and CPU to find where the issues was.
Started with the motherboard. No difference! Swapped the CPU and all is now OK.
The only overclock on this was the XMP profile from the ram. Nothing else. Voltage stock.
Just how common is this?
Tom

When my 930 died I found out with a bios error message the next time I tried to boot up. I pulled it out and dropped in my 920 backup, remounted the heatkiller waterblock and kept on rocking.

 

[111alan]

Be aware, most XMPs will make the MB to adjust vccio and vccsa voltages to a dangerous region(like 1.35v,etc). That's why I suggest manual overclocking over XMPs.

 

[deustroop]

I'm running the ASUS x99 board and i6850k combination 24/7/365 with no problemo.

 

[Carfax83]

Be aware, most XMPs will make the MB to adjust vccio and vccsa voltages to a dangerous region(like 1.35v,etc). That's why I suggest manual overclocking over XMPs.

THIS!! I noticed that my motherboard set my VCCSA to a ridiculous 1.35v when I applied XMP. That would definitely kill a CPU that is air cooled. Funnily enough, my CPU needs only 1.13v VCCSA to run DDR4 3200 CL14 in quad channel.

To the OP, I think you just had a defective chip. Intel CPUs are quite robust, and hard to kill.

 

[2blzd]

I've had two 6800k's and their respective motherboard's die within 60 days of launch. A few of the threads in the first reply are mine.

In the end, I figured out the aftermarket PSU cables I purchased were the issue. At the time though, I had the same symptoms as you.

I had the Asus X99 A-II motherboard, twice. Then upgraded to Gaming Strix

 

[moonbogg]

THIS!! I noticed that my motherboard set my VCCSA to a ridiculous 1.35v when I applied XMP. That would definitely kill a CPU that is air cooled. Funnily enough, my CPU needs only 1.13v VCCSA to run DDR4 3200 CL14 in quad channel.

To the OP, I think you just had a defective chip. Intel CPUs are quite robust, and hard to kill.

I saw this comment and checked my SA voltage. It was at 1.365 even with XMP off and ram speed and timings manually inputted. I manually set the SA voltage to 1.1 and I'm in windows with it but haven't done any stability testing. Couple questions since I can't find any good info on this voltage. Does it ramp up and down like CPU core voltage? Or is just entering a fixed manual value ok for this? I find it shocking that using stock ram settings with a supported ram kit would result in dangerously high voltages. I have no experience with SA voltage. I've only ever needed to tweak core voltage. I was at 1.365 SA voltage for a couple weeks. Glad the chip isn't dead.

 

[Carfax83]

I saw this comment and checked my SA voltage. It was at 1.365 even with XMP off and ram speed and timings manually inputted. I manually set the SA voltage to 1.1 and I'm in windows with it but haven't done any stability testing. Couple questions since I can't find any good info on this voltage. Does it ramp up and down like CPU core voltage? Or is just entering a fixed manual value ok for this? I find it shocking that using stock ram settings with a supported ram kit would result in dangerously high voltages. I have no experience with SA voltage. I've only ever needed to tweak core voltage. I was at 1.365 SA voltage for a couple weeks. Glad the chip isn't dead.

It's a good thing you saw that comment then!

To answer your question though, no it's not dynamic like CPU core voltage to my knowledge. It's basically fixed. I think 1.2v is the max for air cooling if I remember, and 1.25 for water cooling. Those were the recommended values for Haswell-E if I remember, but Broadwell-E's should be lower as it's 14nm.

The only time I would think values as high as 1.3v and above are acceptable is if you're using exotic cooling like LN2 or phase cooling.

 

[.vodka]

You sure dodged a bullet there man.

It boggles the mind that a stable platform like X99 is at this point in time, has this issue.

 

[Carfax83]

@ moonbogg, also check your VCCIO voltage to make sure it's at stock or near stock value and not ridiculously high like the VCCSA.

 

[moonbogg]

OK then I'll start with 1.1 for SA and check stability. VCCIO was like .96 but I'll make sure its at default stock level. Hope I didn't damage anything. This really pisses me off. Although there are a lot of people who install fast memory, select XMP and only tweak the core voltage and they just go with that. There should be thousands of dead chips out there I'd think. Maybe there is, lol. Thanks.