Can CPU be Fired if Prime stable at stock?

[rob_cunningham]

Hopefully I'll get a really simple answer to this really simple question so please post your views...

Background:

I have sucessfully overclocked my CPU Q6600 to 3.4ghz stable and also found stable settings for 3.2 & 3.0 and had a good go at getting 3.6 stable but gave up becasue i couldn't find a set of voltages that would run stable and cool enough for my liking. Anyway, i ran on 3.0 for most of time cranking it up to 3.4 if playing a game. i started crashing at 3.4 and getting random power loss to monitor or whole system. So i stress tested at 3.4 and failed. so dropped down to 3.0. some time months pass and i'm getting similar crashes (screen sleeps or system randomly shuts down) so i drop to stock speeds and voltages. Again some time passes and now i'm getting the same problems at stock speeds.

My Question:

I'm stress testing at stock speeds with Prime small ffts and all seems stable. core temps happy around the high 40s to mid 50s c. I have been testing for over 12 hours now and if i pass 24 hours is this sufficient that my CPU can be ruled out as being the problem?

if i find the problem i would like to try and bump my cpu back up to 3.0 or even 3.4ghz

My Specs are if it helps are:

P5N72-T Prem'
Q6600 (G0) VID 1.250v
Xigmatek HDT- S1283 Cooler
2x2gb OCZ Reaper 1066
BFG 1000w PSU

Thanks in advance

 

[ubernoobie]

it's likely your power supply is failing, get a quality pc power and cooling/antec/corsair/seasonic/ocz/xigmatek power supply and you'll be fine

Edit:Nvm the psu is good but it still could be not enough amps for the cpu or something since single rail systems are more stable

 

[rob_cunningham]

do you not rate this psu? I've got it slightly wrong sorry (which probaby won't improve your opinion of it!). its the BFG ES-800

bought it because it was the only psu i could ind that over 26a on 12v rails. link to it below:

http://www.bfgtech.com/bfgr800wespsu.aspx

 

[croc]

do you not rate this psu? I've got it slightly wrong sorry (which probaby won't improve your opinion of it!). its the BFG ES-800

bought it because it was the only psu i could ind that over 26a on 12v rails. link to it below:

http://www.bfgtech.com/bfgr800wespsu.aspx

You could be the victim of 'electron migration'... Basically, that means that there used to be a tendency in the high K gates for the dielectic to move from the gate area to the base area. You didn't mention the stepping of your CPU, and some of the C0 q6600's were prone to this. I am sure that there are many more knowledgable than I on this subject, but I explained it as best as I understand.

So, short answer is, it is possible that your CPU is failing.

However, as previously mentioned, your PSU could also be a prime candidate.

I have no firm answers for you, unfortunately...

 

[Gixbit]

So many factors. I find overclocking problems to be by far the most interesting problems. It always always peaks my interest.

ALRIGHT, Here is my 'thoughts' hehe. Like ubernoobie said, Power supply issue. If you have a spare one you might want to consider testing it. Make sure nothing is interfering with the power going into your power supply such as the electrical work in your house and such.

Clarification: Monitor goes into low power state (Like your computer has shut down or is off) or loses power?

Could be a ram issue. Sometimes when they heat up they fail.
Could be a heating issue.
Could be a mixture of problems. Faulty motherboard, faulty central processor, faulty ram, faulty video card.

Also, we need background info. Prior issues before overclocking? Hardware changes before overclocking but, relatively recent? Anything at all you can think of that (even logically) might be an overlooked factor by yourself that is causing you issues? Dropping a hammer on your motherboard type of thing and it still works as an example. Though I find that type of situation is a dead give away. So anything lesser than that you can think of, throw it up here.

EDIT: Also, 'Electromigration' may be the issue, but, I would think that the symptoms would be a little bit more profound and sudden. I would assume "sudden failure" as a symptom of electromigration in the best case scenario. Also, Electromigration is present even if you're not overclocking. It's just a really complicated way of saying that your Central processor is degrading. It's caused by heat, current density. No matter what you do, processors today as they get smaller, they will be victim to this phenomena. So overclocking is a really good way to speed up that process. At your own risk per se.

 

[rob_cunningham]

thanks both for your input. my cpu is G0 stepping if that helps.

since deciding to investigate the problem i have tried to see what i can do to cause my system to crash like before, but typically no luck!

Prime currently running small ffts, 4 instances of memtest & 3dmark06 all running at once and pc handling it!

annoyingly my pc is running the best it has in a long while!

i've moved my tower from upright under my desk to flat on top of my desk and turn power profile to high performance so that it would attempt to go into sleep mode. now thinking either a setting issue in power profile or mb bending and causing a connection issue from weight of cooler or gfx card???

 

[Gixbit]

If your motherboard is properly mounted and there is no damage to the mounting tray/platform in your case, that should not be an issue. The video card is screwed in and not using tool-less designs, most of the weight will be distributed to the case an an acceptable amount to your motherboard. What I'm saying is it should be fine unless there is some other factor that may be causing a warping effect. Your motherboard could also be warped, that is pretty common for heat to cause warping. So SOOOO many factors.

Look in your case, take the motherboard out, eye it. Eye your components. Examine it physically. Re-seat everything. So many things you can do man. Try ram in different slots. Clear dust out. Clean your case (70% alcohol is good). Then test it logically. Doing all of these things adequately will eliminate it as a factor.

ATX Motherboards and ATX Cases as an example, are designed within certain specifications.

If all cases were built to have that problem, what's the good?

 

[rob_cunningham]

So many factors. I find overclocking problems to be above far the most interesting problems. It always always peaks my interest.

ALRIGHT, Here is my 'thoughts' hehe. Like ubernoobie said, Power supply issue. If you have a spare one you might want to consider testing it. Make sure nothing is interfering with the power going into your power supply such as the electrical work in your house and such.

Clarification: Monitor goes into low power state (Like your computer has shut down or is off) or loses power?

Could be a ram issue. Sometimes when they heat up they fail.
Could be a heating issue.
Could be a mixture of problems. Faulty motherboard, faulty central processor, faulty ram, faulty video card.

Also, we need background info. Prior issues before overclocking? Hardware changes before overclocking but, relatively recent? Anything at all you can think of that (even logically) might be an overlooked factor by yourself that is causing you issues? Dropping a hammer on your motherboard type of thing and it still works as an example. Though I find that type of situation is a dead give away. So anything lesser than that you can think of, throw it up here.

gixbit you are thinking similar to me - could be a huge number of things. i'll try to answer your questions also see my previous post (while you were posting):

Home electircs fine.

Yes monitor goes into low power state. sometimes within a min or so system just turns off (not sleep or shut down). Sometimes system will just turn off (cuts off power) before monitor goes off.

Pc sometimes refuses to boot up into windows. Various problems:
won't post get a single long beep that repeats. i reset and then usually fine. sometimes just won't post at all. pc posts and gets half way through boot and turns off.

Currently testing Ram once one cycle has completed (taking ages with p95 running too) i will shut down and run memtest from boot cd for a few cycles.

Heat seems to be ok although higher idle temps (need to get dust out of fins on cooler) but still running well within limits. current idle is 30-35c across 4 cores and max loads temps 49-56c

No other hardware changes. Pc been untouched apart from reducing the Oc down to 3.0ghz and then later back to all stock levels.

If i get consistent running with tower horizontal i will check all connections. I have purchased a backing plate for cooler too. so will remove everything and check for connection issues.

 

[rob_cunningham]

If your motherboard is properly mounted and there is no damage to the mounting tray/platform in your case, that should not be an issue. The video card is screwed in and not using tool-less designs, most of the weight will be distributed to the case an an acceptable amount to your motherboard. What I'm saying is it should be fine unless there is some other factor that may be causing a warping effect. Your motherboard could also be warped, that is pretty common for heat to cause warping. So SOOOO many factors.

Look in your case, take the motherboard out, eye it. Eye your components. Examine it physically. Re-seat everything. So many things you can do man. Try ram in different slots. Clear dust out. Clean your case (70% alcohol is good). Then test it logically. Doing all of these things adequately will eliminate it as a factor.

ATX Motherboards and ATX Cases as an example, are designed within certain specifications.

If all cases were built to have that problem, what's the good?

when backing plate arrives i'll do a complete overhaul. also might look into dust filtering too.

 

[Gixbit]

Alright, Well. I think I'll just add in that you may want to consider your harddrive as an issue too. Try reinstalling windows and using Hdtune to check the smart. Might not be though, I'm just throwing ideas no matter how fruitless they are they will eliminate factors for you.

Check your cable management too. Sometimes something so small can be the issue. At least if it works after what you've recently posted, you'll have it narrowed down more.

 

[rob_cunningham]

Since the last coupe of days I have gutted my PC - removed all the dust, fitted a backing plate to my cooler, reseated CPU and applied fresh thermal paste. I have also thinned out all the excess hardware & cables in my case to improve airflow. Today i should also be getting an additional 120mm case which i am fitting to the under side of the case to blow up into the Graphics card and chipset (case will be raised slightly). Also will be fitting dust filters to all intakes too.

Already noticed difference in temps. i'm running at similar idle temps as before but my fans are currently set to Q-Fan mode.

Once filters arrive i'll re-test everything at stock speeds with case upright. and see what happens.

One thing i should mention upon inspection of my cpu after cleaning with Artic Clean i noticed an area of the chip's contacts were a shade darker. I mean only ever so slightly darker, at first i thought the area was wet from cleaning the top of the cpu but after leaving it for a while patch was still there. None of the contacts on the MB looked like they have been burned and i compared pics of fried cpu's i googled to see if mine looked similar but they weren't.

Should have taken a photo but competely forgot until i had just finished putting the heatsink back on!

Any ideas?

 

[Gixbit]

Well, It still turns on right? Then that's how you know it's not fried! Hehe.

Just hope for the best. Then again you're hoping for me to say more than that. Well, What can I say... It could be anything or it could be what you're thinking it is, a slight discoloration from overheating, maybe. I'd even say it could be due to the electrical current going through the pins. Your chip isn't fried, then it should be okay. I mean the most you can hope for is to do a stability test and leave it on for a long time on stock speeds. Large FFTs or Blend. If it lasts, then it should be okay. At that point I'd try to forget about it and hope it's okay unless it makes no difference to you and you'll use it as an excuse to buy a better processor/array of parts for a new build. And if it is okay, you can always sell it off when you buy a new processor.

 

[rob_cunningham]

Set Prime running small ffts before i left for work and came back just now and Pc had shut down/switched off. Powered up and got the start up menu you get after a crash, oh well. Crashed after about an 90min

I know what you are saying if it turns on its not fried. but its not stable either.

Any reason why you are suggesting large ffts and blend rather than small ffts?

If I can get get it running stable again it would be fine, its extremely annoying being in the middle of something and having a crash for no reason. If i could OC back to 3.0 or higher that would be fantastic! But a fried Cpu or motherboard is a great excuse to go out and buy a new Core i7 system! I would salvage the rest and sell it off to get some money back too!

I'll test the memory again, set the q-fan back to duty cycle (although doubt it is a temp issue) and put the system horizontal and re-test each time.

Another question.... when overclocking and setting the FSB higher will this effect the graphics card in any way? communication speeds. could overclocking the fsb or adjusting the NB/SB voltages damage a graphics card or the PCI-e port?

 

[rob_cunningham]

Prime puked on me after about 4hr30min but here's the interesting thing when i got up this moring i had one of the old symptons of pc being on but screen staying in standby.

before hitting the reset button i got the netbook out and tried connecting via remote desktop. It recognised the Pc was there and logged on but wouldn't go past the welcome screen. So i then tried searching for my Pc via my home network. PC is shown being there and i can access and open/download files from my PC on my netbook.

So could this be a graphics card fault??? and hopefully not the CPU?

 

[Gixbit]

That's interesting. I'm quite intrigued. Something else must be going on if your PC is going black and you can still access it. Hmm.

First, If you overclock your Northbridge speed (Front side bus hardly even exists anymore, stuff like Hyper Transport is the thing now) it will increase the connectivity between the motherboard, the cpu and the graphics card. So, yes. It will create more heat, proper case cooling is a necessity. I'd say a 10% overclock on northbridge is acceptable on case cooling. One thing you could explore is underclocking too. Try underclocking your video card and leaving prime95 on Large FFT. I say Large FFT because it uses more ram and pushes your PC harder than Small FFTs. I don't really feel like explaining it to you, look it up.

Underclocking can increase stability so, I don't want to see anyone blah blah at me for that one. That'll just show you that something isn't working right. I might consider your windows installation too. That's something else. A healthy windows O/S and a healthy stable system. I hope you mean you're trying this out on stock speeds and not still trying to o/c first.

Anyway, Keep trying things. Let me know how it goes

 

[rob_cunningham]

re-testing with an old 8500GT that i know is fully working. if it fails then i'll try the other pci-e slot and try again.

 

[Gixbit]

How's it working out for you buddy?

 

[rob_cunningham]

How's it working out for you buddy?

Hey!

Tested Ram @ 1066 with memtest and came up with Zero errors.

Swapped card for 8500GT and tested with small ffts (becasue that's what i had tried previously so thought would be a fair comparison) and it failed. Tried a second time and failed.

Put 8800GTS back in and screen kept going blank about 30sec after login (really annoying as i had to print off one thing before work!). Did this for about half an hour of trying and eventually stayed on. and PC has been on since.

Tested Prime large ffts and passed after 24 hours! Temps hit 60c but it is on Q-fan mode so not concerned.

Now testing Prime on blend and about 2 1/2 hours into it.

found a nice article on reasons for testing small/large ffts or blend. talks about how it can test memory, CPU and CPU/Memory interface and different heat productions and power usages for each. states large ffts produce most heat on single thread cpu but shows most heat is produced when running multi-instances of small ffts or a mix of small & large ffts. quite interesting.

if i pass the blend after 24 hours i'll go straight into a small ffts test. if that fails then i guess i have to assume the cpu is likely at fault.

i'll keep you updated.

 

[Gixbit]

You should reinstall windows. I can't help but feel if memtest has no issues that a software issue is likely. Viruses and whatnot can do that. So, reinstall windows and check out what happens.

 

[rob_cunningham]

i don't think i fresh install will do it because i have had the same problem going back to when i was still using vista, and win7 beta and now with win7 rc1

 

[Gixbit]

Personally, I think both of those operating systems are *** and I have personally opted back to WinXP on my system. I must be doing something wrong because, I find Windows 7 is *** so far. Memory leaks galore. I have no issues with XP. So, Something is going on there in their coding or my hardware with that windows o/s. I'd say don't look forward to anything better than XP for a while. Infact, I'm pretty much willing to say WinXP is the last good operating system you will see for quite some time.

If linux turned into a mainstream gaming operating system like windows... I would just completely drop windows forever. I wouldn't even need a reason to stay there would be none. Linux can do everything windows can it's pretty much the same and the only problem is user-friendliness which I have no problem with. The only thing you can't do with linux is native gaming, you have to emulate and that has it's limits sadly.

Anyway, I'd say try windows XP as I really don't have much faith in the vista clone and vista.

 

[Gixbit]

Honestly, I'd just like to go ahead and say something is probably faulty or damaged. But, you don't know what so, you have to figure out which one it is. Seeing as how your ram, central processor and motherboard all work together pretty synchronously, it will be hard to determine which one is the culprit.

So, When I say that, that assumes that your harddrive is in perfect operating condition within reason, your installed operating system is working correctly with all hardware and drivers correctly installed. Your computer has enough power and the power supply is not damaged/faulty. Your video card is running acceptably given the specifications of your hardware. As well as all other peripherals.

So, Ram. That's easier to test, memtest86. Motherboard...? I'm up for suggestions there. Central processor... Maybe underclocking it? As well as the northbridge?

These things are hard to troubleshoot. This is assuming you have no other hardware to swap and test. Like, if you can swap your processor into another system with all the other components and ram the same, bam it's the motherboard.

Or swap the processor with a different one, or swap ram sticks with different ones.

You see what I mean? So, you might have to take a guess and spend a pile of money to attempt to guess what is not working correctly.

So back to what I said, testing everything else -thoroughly- is nothing but, benefiting you. So, Reinstall your operating system and/or try windows XP. This is your money and hardware. I'm just the guy who's helping you troubleshoot.

 

[Gixbit]

Also, Having friends who are massive raging computer enthusiast types is helpful. They might be able to swap parts with you to test if they're not touchy on loaning parts.

 

[rob_cunningham]

Honestly, I'd just like to go ahead and say something is probably faulty or damaged. But, you don't know what so, you have to figure out which one it is. Seeing as how your ram, central processor and motherboard all work together pretty synchronously, it will be hard to determine which one is the culprit.

So, When I say that, that assumes that your harddrive is in perfect operating condition within reason, your installed operating system is working correctly with all hardware and drivers correctly installed. Your computer has enough power and the power supply is not damaged/faulty. Your video card is running acceptably given the specifications of your hardware. As well as all other peripherals.

So, Ram. That's easier to test, memtest86. Motherboard...? I'm up for suggestions there. Central processor... Maybe underclocking it? As well as the northbridge?

These things are hard to troubleshoot. This is assuming you have no other hardware to swap and test. Like, if you can swap your processor into another system with all the other components and ram the same, bam it's the motherboard.

Or swap the processor with a different one, or swap ram sticks with different ones.

You see what I mean? So, you might have to take a guess and spend a pile of money to attempt to guess what is not working correctly.

So back to what I said, testing everything else -thoroughly- is nothing but, benefiting you. So, Reinstall your operating system and/or try windows XP. This is your money and hardware. I'm just the guy who's helping you troubleshoot.

Hey,

Tested System with a different HDD that had a copy of XP Pro and same problem. Blank screen won't wake but system still on from running idle and during Prime test. So that knocks out two more things - its not the OS and its not the HDD. Which means there is only 3 things it could be now: PSU, CPU, or MB.

I'll see if my old PSU has enough juice to run my system and if that doesn't work then i may be able to swap CPU's with someone on the weekend.

its a little annoying as i was hoping that spending some time on this would unveil a simple (cheap) solution but now if its one of those 3 then its gonna cost me a fair whack to replace either of those components.

I'll keep you updated.

 

[rob_cunningham]

just further to above...

boot up this morning - posting failed initially, i got a single long beep that repeated itself after a few seconds until I reset. typically manual doesn't give a fault/beep code explaining what the problem is. contacted Asus Tecnicial services and they don't know either. had this problem before and tried researching it with no luck. Anyway, after reset i posted and got the bios screen and then the screen went blank (into standby) and nothing more happened. so i reset again and booted fine.

guess this confirms that its not OS related.