My 6850X just died! Just like that!

tomb18

Junior Member
May 12, 2009
5
0
61
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
 

.vodka

Golden Member
Dec 5, 2014
1,203
1,537
136

krumme

Diamond Member
Oct 9, 2009
5,952
1,585
136
Its a fate that is comming to all 6850 on thursday.
Its...
Thursday the 2

Scary stuff.

Trolling is not allowed
Markfw
Anandtech Moderator
 
Last edited by a moderator:

VirtualLarry

No Lifer
Aug 25, 2001
56,229
9,990
126
I don't know how common that is, but if you were running stock CPU clocks, and only XMP, and the CPU died on you, I would be calling Intel for an RMA. That does sound a bit... unusual.
 
  • Like
Reactions: Drazick and utmode

tomb18

Junior Member
May 12, 2009
5
0
61
You are darn right it's unusual....
This is my main development pc and can't afford to be without it...hence the expensive purchase today to just keep going....Sometimes I wonder just how worth it it is to buy top of the line....
I use an Samsung 950 pro as my boot drive and a intel 750 u2 drive.....what if those went??
Yikes...
 

dark zero

Platinum Member
Jun 2, 2015
2,655
138
106
Guess it's time to test another brand (looking to AMD)

Also, better to put it on warranty.
 

IEC

Elite Member
Super Moderator
Jun 10, 2004
14,323
4,904
136
RMA time for the CPU. I seem to recall Broadwell-E + Asus X99 leading to an unusual number of dead CPUs when I was researching building a BD-E system.
 

rchunter

Senior member
Feb 26, 2015
933
72
91
That sucks big time. That's the first i've heard of broadwell cpu's dying just like that. X99 motherboards yeah...
I've got a broadwell-ep. e5-1650v4. Hopefully it doesn't suffer the same fate. The xeons are really supposed to be better binned chips though so hopefully it won't happen.
 

bigboxes

Lifer
Apr 6, 2002
38,487
11,943
146
Never had a CPU die. It's always the other parts that fail. Eventually, I stop putting money into an old platform and put the old CPU in a box.
 

Charlie22911

Senior member
Mar 19, 2005
614
228
116
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?
 
  • Like
Reactions: formulav8

ReignQuake

Member
Dec 8, 2015
86
5
11
The motherboard has an Asus O.C. socket. The board did it my theory and not absolute fact. It'd be interesting to see if Intel actually attempts to diagnose the processor instead of just recycling it/binning it and replacing it.

A whole bunch of people tackled Asus who've not done the right thing in my opinion, people are sitting there, or rebooting, or adjusting elements of their CPU and suddenly the board decides to fry the CPU with higher voltage settings, almost at random and a lot higher than they've set.
 
Last edited:
  • Like
Reactions: formulav8

jihe

Senior member
Nov 6, 2009
747
97
91
Check your voltages, some boards overvolt on auto setting and kill your chip after a while. Check VCCSA etc and make sure they are within spec.
 

PingSpike

Lifer
Feb 25, 2004
21,729
559
126
I was reading a thread on another forum where Intel denied a warranty claim for a cpu because of running XMP.
 

tomb18

Junior Member
May 12, 2009
5
0
61
It's an Asus X99 Deluxe II.
If they deny the warranty based on XMP, then that would kill their overclocking market I would think.
I'll check the voltages in use.
 

Justinbaileyman

Golden Member
Aug 17, 2013
1,980
249
106
This is why I hate spending the big bucks on CPU's :( I would be totally heart broken if my high end rig took a dirt nap out of the blue like that. Specially if you weren't doing any O/C'ing at all.
 

.vodka

Golden Member
Dec 5, 2014
1,203
1,537
136
This is why I hate spending the big bucks on CPU's :( I would be totally heart broken if my high end rig took a dirt nap out of the blue like that. Specially if you weren't doing any O/C'ing at all.

To be honest I haven't heard of this happening to other CPU families.

It's related to LGA2011 somewhat, I've also heard reports of ASUS' own "OC Socket" screwing things up and killing CPUs now that I remember (around Ivy-E era, a while back), as it does not adhere to Intel's specifications. They've added a few pins that aren't there in a stock LGA2011v1/v3 socket, doing who knows what with the CPU.

This is not a coincidence.

Technically using XMP (and overclockg memory in general) involves overclocking memory controller as well.

Correct. But XMP is an Intel sanctioned system that allows memory to run out of JEDEC specs without voiding warranty.

It's absurd for them to invalidate warranty because someone used a feature inherent to a platform.
 

lolfail9001

Golden Member
Sep 9, 2016
1,056
353
96
Correct. But XMP is an Intel sanctioned system that allows memory to run out of JEDEC specs without voiding warranty.
Not even once there was a statement about "not voiding warranty" that i have heard of, got a source?
It's absurd for them to invalidate warranty because someone used a feature inherent to a platform.
Unlocked multiplier is a feature inherent to a platform, they do not honor overclocked CPU warranty without separate purchase either.
 

mxnerd

Diamond Member
Jul 6, 2007
6,799
1,100
126
One AT member already think Intel HEDT processors will soon be Highly Expensive Door sTops.
 

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
Technically using XMP (and overclockg memory in general) involves overclocking memory controller as well.

Ehhh I'm not gonna say 100% no, but I am gonna say, "not quite".

Any time you raise vDIMM in a modern or even semi-modern platform with a CPU featuring an IMC, you stress the IMC unless you also increase its voltage. I remember k8s could suffer a damaged/dead IMC from raising vDIMM without raising vcore, since IMC voltage tracked off vcore. In fact, I damaged the memory controller on my x2-3600+ .

That doesn't constitute "overclocking" the memory controller, though. Its clockspeed is going to stay the same, XMP profile or no. XMP profiles will raise vDIMM though, and that can burn out a controller. Intel IMCs seem quite finicky in that regard.

To be honest I haven't heard of this happening to other CPU families.

Newer FM2+ boards/UEFI revs will severely overvolt APUs and CPUs, though that only leads to some throttling, sometimes. Maybe.

Correct. But XMP is an Intel sanctioned system that allows memory to run out of JEDEC specs without voiding warranty.

It's absurd for them to invalidate warranty because someone used a feature inherent to a platform.

That's my thought on it. I'm not sure of the warranty implications of using XMP. It doesn't void the warranty of the RAM itself if you use it, but the potential chain of failure that can kill the IMC makes things more complicated.