Can you burn out 1 of four cores on a cpu core i7

[WINTERLORD]

hi was wondering if it is possable to burn out just one core on a core i7 920 cpu and if so would it still work? reason i ask is because my tempratures get ungodly high and core one always gets 8-13c higher then the rest. iv reseated the heatsink and fan several times with no luck also using aric silver and a zalman 9700 heatsink with bracket. so there nots really any other way to get it cooler by the amount i need so im not exactly overclocking but i have tweaked the voltages as low as possable to save heat.

i also believe i get poor performance for a core i7 my passmark score is 1751 and some core 2duals get a better score then that. disabling the tempurature stoppers for the test all but prochot. so i am wondering if it is possable one of my cores are alrdy burned out and what i see for core 1 is actualy just the sensor still taking a reading. is this at all possable? and is there anyway to check this

thanks
winter

 

[andy5174]

8-13C higher at 100% load with prime95 or idle?

Your CPU should be fine if this is only true at idle. You are getting this odd core #1 temp due to the bad temp sensor at core #1 if that's the case.

By the way, my Pentium E6300's Core#1 temp sensor will only work for temps above 31C and any temp below will shows 31C.

 

[megamanx00]

Possible, but not likely.

 

[SpidersWeb]

It shouldn't burn out as such, but it will wear out faster than the other cores due to the extra heat. Weither this will be noticeable is up for debate. I doubt it.

But the temp of that core is the temp you should treat the system for, don't ignore it. If the dodgy core limits OC, then I'd order another i7 if possible and sell that one on ebay or similar. I'd also go for a newer high end cooler.

The first core wont be slower unless the core temp is getting well up around 80ish. Passmark (if I remember correctly) is for business apps etc, which are usually single core. A single i7 core at 2.6Ghz wont beat a E8500 single core at 4.2Ghz etc. If you were overclocking, you'd snot all over an overclocked dual core however This could explain that result as well.

You could also try rotating your Zalman 90 or 180 degrees, and seeing if the temp increases on another core, you may find it's just not sitting quite flat enough. My cores vary 2-3C, but 8-13C would be a bit much.

 

[WINTERLORD]

well underload in prime core1 gets upto 100celcius and the rest only get up to 80celcius. before it would crash because of heat now just normal everyday operating mainly playing a cpu intensive game the one core gets max 78-79ish the others about 68-69ish. it tends to crash around 83celcius but for some reason when i ran prime i saw it all the way up to 100celcius and stayed there for about 60seconds before it crashed. so yea around 10ish on average difference betweeen cores.

i know it aint spose to be anywhere near that but iv truly tried everything. nowq one thing i did notice and a new bracket is on its way from zalman was the other day i went to check the screws to see if maybe if i losened them maybe the tempo would go down a tad (was a long shot but worth a try) well the one screw poped. so clearly it must of been stripped or is now stripped. im wopnderin if it was stripped and just not noticable till i felt the definant pop while tighting it a tad. wich could be the casse but the rest of the screws are really tight so im not sure how much it will help.

my main question would be if the core did burn out would the other 3cores still work.

 

[andy5174]

my main question would be if the core did burn out would the other 3cores still work.

No idea! However, DO RMA it if possible.

 

[SpidersWeb]

100 celsius?
The fact it still boots is a miracle.

I did some reading. Zalman 9700 will not be suitable for an i7 920 overclock.
Buy a cooler that's designed for the i7 on 1366 that shows good reviews.
At 100C the thermal throttling should have kicked in to save the processor. Never turn this feature off.

If that core dies I sincerely doubt your CPU will still work. The damage caused by the excessive heat will likely also extend to other parts of the chip since they're only a few microns away.

Sorry I would've jumped on this earlier if I knew you made it to 100C. I thought we were talking 60-70C here. 100C is beyond insane, its also the highest the temp sensor can even read, so if you went over 100 we'd never know. The crash is likely the CPU turning itself off in order to save itself from a firey death.

Order a new cooler, try again, if problem persists you'll want to RMA the chip. 10C between cores is too much variance, and in all honesty you may have damaged/degraded it by running it up to thermal shutdown point.

 

[john_29]

and yes it is possible to burn out just one core

 

[Shadow703793]

I'l wait until CompuTronix respond before adding what I think.....

 

[WINTERLORD]

ok, one thin i hate with my motherboard it seems to have a generic bios its a ex-58-ds4 i left proc hot enabled, but i dont think it would throttle with thermal monitopr and eist disabled. though i had to disable thermal monitor for it to not throttle at when it goes past 130watts. for i dont see any other feature that does this. but i could be missing something. as for the 100celcius i know that was real bad but luckly it went up and came back down was only up over 90-95 for 60seconds tops. unfortunatly i just let it go because iv been use to high temps since i got the thing. i only get it up there if i try to run prime so i dont anymore. but i know it still runs way to hot even at 70-80 max when im just gameing. also i am wondering could my motherboard at all be causing these extreme temps? iv had problems left right and center with this board, and soon im going to give gigabyte *** over it. now a memory controler dont work (atleast i hope its the board) but iv tested all the ram one by one in bay in the last few bays and they work just fine, so im guessing that shows somethin wrong with the board not sure if this could induce those kinda temps or not? as for the processor, i probably will try and rma it since it looks like imma have to rma the motherboard.

and as for the thermal compound, iv tried AS% also have tried super grease from zalman very thin layer. found the as5 seems best but i use the super greases brush sionce it dfoes a great job at layin it thin

 

[hundredislandsboy]

If one side or one corner of your CPU chip is hotter than the rest, it could be a poorly seated heatsink.

 

[SpidersWeb]

Do not give Gigabyte *** over your motherboard.

The temperatures are set INSIDE the CPU, those core temps you get in RealTemp are what the sensors in your i7 chip are reporting, the motherboard has nothing to do with it. The memory controller is also INSIDE the i7 chip which may be in bad shape now.

- buy a proper LGA1366 designed heatsink
- see if things work properly (with thermal throttling turned on)
- if they don't, then get a new i7 chip, weither you RMA it or pay for a new one (Intel may let you return it, but by disabling thermal throttling and all the other stuff your warranty is well well gone)
- if after installing a new CPU, you still have issues (that aren't temperature related), then RMA the board

Do them in that order.

Your CPU was way too hot, thats the most likely reason why you're having problems. Normally this is caused by a poor heatsink, or one that is poorly mounted. It sounds like you're aware of the correct application of the paste, but the mounting itself could be off which would explain the 10C difference too.

Leaving thermal throttling on is a good idea (this only kicks in at extreme temperatures, the wattage isn't measured by the chip). Don't run a CPU with core temps over 80C, especially when it's not a cheap processor.

 

[SpidersWeb]

I'l wait until CompuTronix respond before adding what I think.....

I want to know what he says as well.

 

[WINTERLORD]

i checked the memory, the memory works just fine in all the other banks. but i confirmed banks 0 and 1 do not work this is a recent problem that just cropped up. but id assume that, the reason banks 0 and 1 dont work now are because of the motherboard or memory slots. iv just had problems with this board since i got it last year. i suppose it could be the memory controller but iv never run the mem over 1.6volts, so idk one other wierd thing i noticed and i took the snapshot before the banks 0 and 1 went out is that it says 1way channle interleaving in everest. i dont have the full version just trial so i dont have alot of info but heres what i got
Field Value
Memory Controller Properties
Error Detection Method 8-bit Parity
Error Correction None
Supported Memory Interleave 1-Way
Current Memory Interleave 1-Way
Supported Memory Voltages 5V
Maximum Memory Module Size 1024 MB
Memory Slots 6

now the memory size is off my computer at the time under properties said i had 5GB installed cause i have 3x 2gb sticks but now im only running 4gb in ddr2 mode since banks 0 and 1 went out. i checked all the memory sticks and they are fine just first 2 banks dont work now

 

[SpidersWeb]

The memory controller is integrated in the i7 CPU, which has been run at extreme temperatures (that are about 25-30C above the maximum Intel spec). If you are having RAM problems now, this is the most likely cause.

Replace cooler + CPU.
If you still have problems then change the motherboard.
Never run over 80C. Never turn off Thermal Throttling.

 

[rodney_ws]

Prime95 is an excellent tool, but it's totally not something to be used on a PC that already has thermal issues. I wouldn't touch it until you think you've got your issues sorted out.

In regards to the problematic memory... that totally could be processor related since your i7 has its memory controller on-die... and well, 100C isn't safe for any processor out today. If you're going to continue to game with your system in this state, you need some sort of audible alert when your temperature passes a certain threshold... 75C maybe?

I think SpidersWeb gave some excellent suggestions although if Intel gives you crap about your processor being baked and won't let you RMA it, try to RMA the board first. If the board you get back behaves in the same manner, well... you know your answer. Burned out CPU.

Now just back away from the Prime95 executable! I'm just having a hard time picturing myself staring at triple digit temps on my computer while running a stress test. I'd have pulled the plug on testing long before it got to that point.

 

[SpidersWeb]

Yeah I hit 83C doing 4.75.
Near sh*t myself. 100C (even just on 1 core) is crazy.

Anyway wanted to add, when you get the system running properly, to run in Triple Channel Memory mode (which is the reason for using 3 DIMMs), the RAM will need to be:

Bank 0: INSTALLED
Bank 1: empty
Bank 2: INSTALLED
Bank 3: empty
Bank 4: INSTALLED
Bank 5: empty

When you power up, it should say "Triple Channel" or "3 way memory interleave".
If you install all three next to each other, they just work as normal single chips.

 

[WINTERLORD]

yea it is insane. but intel should'nt give me crap i never overclocked it first couple months i had it, and it always ran way freakin high never below 60c during normal operation non overclocked never under 50idle non oc. i reseated for the 5th time now and i looked at the underbelly, suprisinly theres no burn marks looks shiny in the center, but the motherboard in the cente of socket was brown. not sure if this is normal i expected it to be the otherway around.

on the motherboard, i know due to the economy crash and all, that gigabyte realy skimped on this motherboard. number one they stopp selling the board completly 2months after it was out. then if you look the board only has 3bios updates in total and when you look at boards that are near idintical and came out at the same time they all have 8 plus bios updates. also this board has already recieved a end of life policy i believe. iv had problems with the board ever since, mostly unconfirmed but highly speculated. also the extream lack of bios support and bios options is just nonsense. lowest qpi will go on this board x36 so after a 3.6ghz oc youv already maxed the qpi buss at 6.67 qpi speed. so i never ran over 3.4 wont boot any higher dos'nt seem to boot if you go above 6.47QT's qpi speed and with only x36 or x48 theres realy no other option but 3.4 or below

also on the memory are you sure banks 0-2-4 even on a gigabyte ex58-ds4? reson i ask is because unless a bios update has changed it the original book says for triple channle banks 1-3-5. maybe thats just a gigabyte thing? gosh i wish i could try it to see with the newer bios but banks 0 and 1 dont work gahh

 

[SpidersWeb]

It's probably not the CPU that's crap, it's probably more to do with your heatsink. Those chips are tested and binned quite well out of the Intel factory, and they do throw out the bad ones. If it was a bad CPU I'd be quite suprised (but do admit it's possible).

I don't really know much about QPI on those boards, so really can't say much. Personally though I've gone off Gigabyte completely myself.

Normally with RAM it starts at 0, maybe it's different with yours. As long as they're spaced, sounds like you knew that though.

In other news, just had an earthquake, on the 7th floor and near shat myself again!

 

[rodney_ws]

So Intel shouldn't give you crap? You overclocked it. You ran a stress test on it that pushed it WELL past its thermal limits. And you think Intel owes you something? Really? If Intel won't take it back, well the blame rests on your shoulders for this one... if they do, count yourself as lucky... you beat the system.

 

[hundredislandsboy]

If it ever got ot the point that Intel started losing too much money from replacing overclocked CPUs, they would figure a way to stop the overclcoking.
1. A 1kb rom cache in the CPU that only they can access and it records any overclocking attempts, successful or not, by the user.
2. Or, an instruction set that detects if the chip is being OC and then you get BSOD that says "If you continue to OC your warranty will be voided and this BSOD event is permanently recorded on this chip and you have no way to access the data!"

 

[rodney_ws]

In regards to what HundredIslandsBoy just said... this sort of thing is coming in a variety of devices. Apple has taken out a series of patents that will help them deny warranty claims for devices that have been dropped or submerged in water. If it becomes a problem for Intel, I imagine they will take steps to differentiate legitimately burned out CPUs vs. one that were overclocked to hell and back.

 

[Shadow703793]

^True. In that case however, I see AMD becoming more prominent on the OCing crowd

 

[hundredislandsboy]

^True. In that case however, I see AMD becoming more prominent on the OCing crowd

Neither Intel nor AMD is in the CPU business to lose money by being tricked by the overclocking crowd. Oh poor me, I f*$#ed up didn't really know how to overclock and guessed that 110 C and 1.78 Vcore was ok. well please replace my chip.
They build a little headroom but certainly the CPUs can be inadvertently abused by noobs and when AMD's and Intel's bottom line starts stinking, we as the mass consumers get punished by higher prices. If you're gonna overclock and burn your chip out in the process, suck it up and buy a new one!

 

[SpidersWeb]

If you're gonna overclock and burn your chip out in the process, suck it up and buy a new one!

x2
I blew another P45 board last night, no plans on returning it. Nature of the hobby IMO.