NEED HELP: Two P8P67 EVOs & 2 2600Ks DEAD within 10 months. PLEASE READ CAREFULLY.

[cadaveca]

Nah, if there is 1600MHz JEDEC profile, it is used. Confirmed by myself on nearly every Intel board I have reviewed in the past year or so. I havea set of stick that doesn't even have a 1333 MHz profile.

 

[phanbuey]

So the recommendations are:

Different MB model and different/compliant ram?

 

[MaxAwesome]

So, what do I do?

Get new RAM? Go as low as 1.35v to be one the super safe side? Or does it break warranty even I run 1.35v @ 1600mhz?

Should be easy to sell both my RAM kits though.

I just didn't know RAM could be such a pain in the arse, this is all new to me.


One other thing, I have used the same RAM since I have my LGA 1155 system, and the first time both Asus and Intel sent new hardware without saying a word. Hopefully, they will do the same.

Even though, the Asus process was smooth because of the recall, so I was going to have a new mobo either way.

Let's hope it's smooth again.

 

[Crap Daddy]

You are right here. One shouldn't run memory out of spec on Sandy Bridge but on the other hand it seems unlikely that this could effectively destroy the board and CPU - 1.55V is not so mucho.
Or is it?

IMC is on the chip so yes, you overvolt it. 1.5V is on spec

 

[newtekie1]

Is it possible that one of the other Auto settings is setting some voltage way too high and killing the CPU and board at the same time?

I don't think running memory out of spec would kill both, maybe kill the IMC on the CPU, but not kill the board...

 

[Completely Bonkers]

Overclocking isnt so much the problem. Overvolting is, esp. on IMC. And if the JEDEC profile or bios results in an overvolting, this can fry your CPU. Wasn't this exact same problem prevalent and well discussed on Nehalem? Hostory repeats itself. http://www.google.co.uk/search?hl=e...l9254l0l9441l16l16l1l0l0l0l198l2027l3.12l15l0

 

[MaxAwesome]

Is it possible that one of the other Auto settings is setting some voltage way too high and killing the CPU and board at the same time?

I don't think running memory out of spec would kill both, maybe kill the IMC on the CPU, but not kill the board...

This is exactly what I was about to bring to the table. Thank you

You are right, I had thought about it, since the IMC is on the CPU die, what could possible cause such a tremendous amount of stress that would take down the mobo and the CPU?

And the worst thing is: What is dying first, and taking the other component with it?

 

[MaxAwesome]

Overclocking isnt so much the problem. Overvolting is, esp. on IMC. And if the JEDEC profile or bios results in an overvolting, this can fry your CPU. Wasn't this exact same problem prevalent and well discussed on Nehalem? Hostory repeats itself. http://www.google.co.uk/search?hl=e...l9254l0l9441l16l16l1l0l0l0l198l2027l3.12l15l0

Is overvolting the RAM, the same as overvolting the IMC?

Isn't the IMC voltage controled by the VCSSIO voltage or something? That was left at stock settings.

 

[cadaveca]

I don't think running memory out of spec would kill both, maybe kill the IMC on the CPU, but not kill the board...

If the IMC "grounded out", and pulled excessive current from board, it definitely could happen. The power that killed the chip had to come from somewhere.

It could just be the PSU, there is no doubting that. And I'm not suggesting people do not overclock...I am simply suggesting that memory overclocking causing such issues cannot be ignored. It would be nice to have both the CPU and board to diagnose, but beucase they are from different OEMs, this is not likely to happen, so the actual cause of death may never be known.


What I do know is that for some people, myself included, these chips can be pretty indestrucible. But at the same time, my chip spends most of it's time @ stock, as when gaming with a single card, i can hardly tell the difference from overclocking, period.

Isn't the IMC voltage controled by the VCSSIO voltage or something? That was left at stock settings.

They are connected, memory and memory controller. The are components in place to step down or step up the voltage as required by each end of that link, but they are still directly connected to each other. The comment about "AUTO" settings causing the problem is an issue too.

 

[MaxAwesome]

If the IMC "grounded out", and pulled excessive current from board, it definitely could happen. The power that killed the chip had to come from somewhere.

It could just be the PSU, there is no doubting that. And I'm not suggesting people do not overclock...I am simply suggesting that memory overclocking causing such issues cannot be ignored. It would be nice to have both the CPU and board to diagnose, but beucase they are from different OEMs, this is not likely to happen, so the actual cause of death may never be known.


What I do know is that for some people, myself included, these chips can be pretty indestrucible. But at the same time, my chip spends most of it's time @ stock, as when gaming with a single card, i can hardly tell the difference from overclocking, period.

That is 100% true. The 2600k (and the 2500k, 2400, etc) are strong enough at stock speeds to not even bottlenck SLI or CF setups.

Overclocking is an added bonus, but I paid good money for that little "K" in front of the 2600, I want to use it.


I'm actually leaning towards getting new RAM. Suggestions?

 

[MaxAwesome]

Side note: Things are going to be sent to RMA tomorrow. I hope it goes through.

Since the hardware was only tested today, the technician is going to open the RMA process first thing in the morning.

 

[Crap Daddy]

What I do know is that for some people, myself included, these chips can be pretty indestrucible. But at the same time, my chip spends most of it's time @ stock, as when gaming with a single card, i can hardly tell the difference from overclocking, period

This is very true. I just did a quick bench today with Mafia II for another thread here at TPU. At stock 2500K gave me 77.3 FPs, at 4.2GHz under one frame more...

 

[erixx]

lol, good info!

BTW, Daddy, I saw you downclocking your HyperX1600. I have tested and I get better latency etc. staying at 1600 (CL8, 1,65v)

But of course, who knows in realworld ... but i am stable.

 

[thebluebumblebee]

Everyone should own one of these anyway, so I suggest getting one of these and testing the outlet. (you should be able to find them at any hardware store, home center, electrical supply) You have a newer house, but a human did the work and the inspectors don't catch everything. (looking for grounding problems)

 

[MaxAwesome]

That's enough, right there, methinks. I mentioned maybe last week in another thread, that it seems Intel is now denying RMAs based on using XMP-rated memory on SB.

Nevermind, Intel does have a list of XMP-compatible motherboards and memory, and neither of yours is on that list, which can be found here:

http://www.intel.com/content/www/us.../core-i7-memory-suppliers-0811-datasheet.html

Until you use "approved" parts, we can simply blame your overclocking for the premature death.

Could be the power in your house too, for sure. Could be many things, but until you stop overclocking, you'll never know. There's a reason overclocking voids your warranty, after all.

You know, this might have been the issue. The store technician told me the same thing, and my Kingston HyperX RAM was not XMP certified.

So, to be on the safe side, I ordered some XMP certified Ripjaws 2x4GB 1.5v, and sold my other kits to an AMD user.

Hopefully, this is what was causing the problem, and maybe killing the IMC on the CPU?

I just don't know if a dead IMC/CPU can take out a board with it.

 

[cadaveca]

Possible, but i do not know the likelyhood. Test socket in wall and PSU, for sure. Better safe than sorry.