KILL your i7 980X AT ONCE (and RMA it)...

[radaja]

well this is getting real scary.now i guess i wont disable HT(EVER)
or run with less than 6 cores(EVER).
i think i will be fine with basic OCing(4.5GHz to 4.8GHz) on water
with voltages in the 1.40v to 1.50v range(I HOPE)

[BaRaN]

not for 45nm if you done for 45nm cpu you do heavy oc and then start with out under cold chip in not boot and after boot show you only overclocking fail i test on 920 - 3540 - 3570 and same on all

but in 32nm sure its kill. i test on 670 too and its dead and its cool when you where in iran and there is no RMA for chip you can play football with dead chip

you dont need heavy oc for 32nm just 5.5 fine with 1.700 vcore and vtt upper than 1.440 and work 100% and you kill your 32nm chip easy

[Vienna]

Running an SR2 with X5650's with HT disabled, all Cores enabled, crunching Rosetta, NO PROBLEM.

[OBR]

this problem is related to disabling cores, not HT ...

[Vienna]

this problem is related to disabling cores, not HT ...

I realise that, but someone mentioned earlier that he didn't disable any cores, only HT and killed his chips, so I thought I'd throw in my experance with HT off on the Xeons to show that I haven't killed my chips(yet, knock on wood)

[quake6]

this problem is related to disabling cores, not HT ...

If you disable cores in windows, will it be the same? Maybe it has something to do with bios related problems while disabling?

[thebanik]

Can confirm its happening, I read the thread, and since I had been able to run 2 cores before without killing it I ignored it. But now after 2 weeks of testing on air(where I ran 2 cores) my 6Ghz chip is not able to run on 6 cores even @ 5Ghz. Would try and boot it with 1 core and kill it fast now, :p

Oh!!! and disabling HT doesnt degrade it(atleast not visible in a couple of weeks of testing, maybe it would degrade it after prolonged use)

[Falkentyne]

Ok, guys, I asked this before but everyone ignored me...

So I'll ask it again.
WHY HAS NO ONE BOTHERED TO CONTACT INTEL ABOUT THIS? And if you have, have they taken note of this problem?

Intel actually has links to the enthuiast community now, and some of you DO know people in the right places. If anyone wants this SERIOUS BUG (and this is why I held off on a 980x) fixed in time for the 990X, with a new stepping, then people NEED TO get Intel's attention. This is probably the MOST serious errata I've ever seen in a CPU, even worse than the Pentium 4's SNDS/GNDS problems at 1.65+ voltages...

I want to buy a 990x, but I sure WON'T buy one if you can kill the chip from disabling cores....some games don't even like multiple cores...

[The_Game]

I`m also wondering if the problem exists if the cores are disabled from windows, someone to try it ? .
Also wondering if this problem will still exist in the coming soon i7 970.

[sniper_sung]

I've just got a new retail 980X (batch 3013A524) in the UK. Still waiting for my motherboard. I am hesitating whether to test it or not...

[radaja]

has anyone experienced when loading optimized defaults with a 980x
the vdimm keeps getting set at 1.79v?i noticed when i loaded the Odefault
it keep setting the vdimm at 1.79v and one other person i talked too had it do it too.
is this some problem with the microcode of the bios's for the 980x's?any insight into this?

[xpatar]

Continuing testing with 1C/1T, current setup under water.

In this test I didn't exceed Intel's specifications safe limits, which means a maximum Vcore of 1.375v.

In that way I wanted to see if there is going to be any sign of degradation or a killing of my cpu with what Intel has issued in its data sheet, here it is an abstract of it:

"Absolute Maximum and Minimum Ratings
Table 2-6 specifies absolute maximum and minimum ratings, which lie outside the
functional limits of the processor. Only within specified operation limits can functionality
and long-term reliability be expected.
At conditions outside functional operation condition limits, but within absolute
maximum and minimum ratings, neither functionality nor long-term reliability can be
expected. If a device is returned to conditions within functional operation limits after
having been subjected to conditions outside these limits, but within the absolute
maximum and minimum ratings, the device may be functional, but with its lifetime
degraded depending on exposure to conditions exceeding the functional operation
condition limits.
At conditions exceeding absolute maximum and minimum ratings, neither functionality
nor l ong-term reliability can be expected. Moreover, if a device is subjected to these
conditions for any length of time then, when returned to conditions within the
functional operating condition limits, it will either not function or its reliability will be
severely degraded.
Although the processor contains protective circuitry to resist damage from Electro-
Static Discharge (ESD), precautions should always be taken to avoid high static
voltages or electric fields."


1C/1T, 1M Super Pi stable @ 4.54 GHz Vcore @ 1.35v





1C/1T, 32M Super Pi stable @ 4.54 GHz, Vcore @ 1.35v





1C/1T Prime95 1 hour stable @ 4.54 GHz, Vcore @ 1.35v





After finishing those tests I restarted, entered into bios, set everything back at default and booted into windows, everything went fine.

Searching a way to test if there was any sign of degradation I used Cinebech.

Everything @ default with 6C/12T :



Not any sign of degradation in the performance of the cpu.


I would understand those killings under extreme overclocking conditions with ln2 and high Vcore but want really happened and George's friend lost his cpu by just enabling 1C/1T?

I don't know if Microcode Update Revision plays a significant role but mine is 5,

check it out with Everest.

[radaja]

when you load optimized defaults does it set you memory vdimm at 1.79v?
i dont know if its my memory or my 980x or my MB that is doing this.
my Microcode Update Revision is 5 also.
my ET6 will show 1.50v vdimm after loading O-defaults but in everest
and the bios in the health status page(current temps and voltages)
it shows 1600MHz 1.79v,but in MIT page it shows auto at 1.50v vdimm and
memory frequency at 1333MHz?

[hipro5]

tsan's CPU died a couple of days ago by playing with it with 2 cores as well....It's not getting into windows now with all cores plus HT....

[sofos1990]

George, as far as I know both you and tsan have a EVGA Classified E760 . Your cpus died on it right ? Maybe is a motherboard's related problem ?

[hipro5]

No, my good one died by playing with 2 cores on a GIGABYTE X58 UD7 mobo....
Then my other one died on an 760 mobo.

Yesterday I tested another one (crappy) on the 760 with 1 core and then with 2 cores and then full cores on the 760 and nothing happent.....It works fine......

So I think that it's CPU's related problem....."Something" inside the cpu dies/degrades VERY FAST when palying with 2 or 1 cores without HT and this doesn't happen to all cpu's.....
My other one is still OK......

[Leeghoofd]

My other one is still OK......

The Andre Yang killer CPU ? Good to hear it's still okay

[xpatar]

when you load optimized defaults does it set you memory vdimm at 1.79v?
i dont know if its my memory or my 980x or my MB that is doing this.
my Microcode Update Revision is 5 also.
my ET6 will show 1.50v vdimm after loading O-defaults but in everest
and the bios in the health status page(current temps and voltages)
it shows 1600MHz 1.79v,but in MIT page it shows auto at 1.50v vdimm and
memory frequency at 1333MHz?

@ default I get a vdimm of 1.6v 1333MHz, with 9-9-9-24-1T, the exact same reading I get with a dmm, at boot time and also within windows.
@ XPM I get a vdimm of 1.64v 1866MHz with 7-8-7-20-2T, the same reading I get with a dmm again at boot and windows environment.

[radaja]

ok thanks.i just cant figure this out
my ram spd doesnt even have 1600 1.79v
so i think its my board thats confused

[Donnie27]

tsan's CPU died a couple of days ago by playing with it with 2 cores as well....It's not getting into windows now with all cores plus HT....

With all due respect, and mad props to your skillz, this thread is almost as stupid as the SNDS thread from years ago. Correction, his processor didn't die it, just like yours', was KILLED. That pretty much goes for all of the destroyed and or damaged processors.

""In this test I didn't exceeded Intel's specifications safe limits, which means a maximum Vcore of 1.375v.""

Then feeding one core how much voltage and still expecting it to live? I really hope any company, not just Intel, seeing their products abused tell folks killing these processors to piss-off!

[Falkentyne]

Excuse me, but if you can throw your arrogance out the window, SNDS cost me THREE northwood processors. Only one ever saw the "max" vcore of 1.7v. I had one that SLOWLY degraded with just a vcore of 1.625v which was considered safe by everyone. Lost over 50 mhz on the max overclock with just that vcore.

(BTW when I mean "cost", I mean the processors worked, but they were worthless for overclocking)

Correction: you can call it " GNDS" (gradual) instead of sudden, as SNDS is basically complete death. I never had that--just "Gradual" degradation.

I even saw just 1.6v degrade a northwood...My P4 EE has degraded slightly with just 1.60v only...(and that's without any sort of vdroop mod !); it won't even do 3.6 ghz stable now, when it used to do 3.7...

[Aberration]

I have the ability to get a new 980X at about half price so I am trying to get a clear picture here.

If I leave all cores enabled, and HT enabled I should be pretty safe?

[SteveRo]

I have the ability to get a new 980X at about half price so I am trying to get a clear picture here.

If I leave all cores enabled, and HT enabled I should be pretty safe?

yes, and ht off with all cores turned on is also very safe (i have done this a lot).

[Donnie27]

Excuse me, but if you can throw your arrogance out the window, SNDS cost me THREE northwood processors. Only one ever saw the "max" vcore of 1.7v. I had one that SLOWLY degraded with just a vcore of 1.625v which was considered safe by everyone. Lost over 50 mhz on the max overclock with just that vcore.

(BTW when I mean "cost", I mean the processors worked, but they were worthless for overclocking)

Correction: you can call it " GNDS" (gradual) instead of sudden, as SNDS is basically complete death. I never had that--just "Gradual" degradation.

I even saw just 1.6v degrade a northwood...My P4 EE has degraded slightly with just 1.60v only...(and that's without any sort of vdroop mod !); it won't even do 3.6 ghz stable now, when it used to do 3.7...

I get your point about useless for overclocking but some folks were warned! My own system was a 2.6GHz running at 3.46GHz with 1.575v. 1.65v got me to 3.6GHz so I didn't take the chance for 140MHz LOL! My Computer ran like that for about 1.5 years until it was replaced by an Athlon 3500+. The person I sold it to is still using it.

Why? I was told if the core needed more than 1.6v, it wasn't going to do much more anyway. 1.6+ = diminishing returns on all 5 Northwood I played with.

Yet, most folks here complaining about SNDS were using wayyyyyyyy over 1.7v and the thread is still here showing crazy folks running 1.8, 1.85 and even 1.9v after Volt Mods LOL! Then wondering why their processors died! Yours wasn't indicative of what was going on with the folks here killing processors and saying they died! You have a heart attack and die, you get shot or get hit by a car, you got killed! They were hitting it with a MAC truck.

Both that 2.6GHz Northwood and the Athlon3500+ are still being used by the folks I sold them to. Why would anyone not decrease the voltage before sending full voltage to one of four cores? With or without HT is moot. Killing processors this way has absolutely NO intrinsic value at all. Speaking of useMaybe some useless LN Xtreme overclocking stunt?

The name of the thread got it half way right, it is Kill your i7 980X at once but the RMA it part. I said I hoped someone at Intel was watching not for support but to see this as a form of fraud! This IS NOT a legit reason for an RMA! You may as well do this!

http://dl.maximumpc.com/galleries/co.../spark_415.jpg

Oh wait, never mind, the way I understand it this one still worked afterward-s Calling for a Fix for something that ain't broke is also kind of lame IMHO! Now folks with legit RMA's get lumped in with this lunacy!

[Erklat]

Read the first two pages of this thread before questioning the whether the RMAs were legit or not.

CPU dying on stock by only shutting a couple of cores off sure as hell sounds as legitimate RMA to me.
As does to all guys who got the CPUs which suffer from the same phenomenon. Whether they applied excessive voltage
is irrelevant. They were unable to shut cores off from the beginning, therefore it limited the CPUs functionality, regardless
of whether disabling any of the cores seems reasonable or not.