Damage CPU from overclocking and/or too much vcore?

[Dave3000]

The only stress test I have issues with is Intel Burn Test 2.54. I have an i7-4820k. In the past I tried to overclock it to 4.6GHz with 1.33v vcore and LLC on high setting and it was not stable. I only had it at that setting for about 10 minutes. 4.5GHz at 1.28-1.3v (don't exactly remember the vcore amount), I didn't have any issues but only tried that setting for 10 minutes. Also one time I set a specific performance feature to on in the BIOS and it automatically overvolted and overclocked my CPU to 4.5GHz with 1.46v and I ran it for 5 minutes at that setting. That performance feature didn't say it would overvolt or overclock my CPU.

Now I'm having issues with Intel Burn Test and I'm not sure if it's because of my CPU not being stable anymore. After the last run in Intel Burn Test I get a critical error message saying something about Linpack binary stopped working, maybe caused by missing executables, software bug, or an unstable system. Also I can get it consistently to show that message after the last run if I run the test at a 16MB size for 20 runs. Also a few times while running this stress test, my system shut down with one time getting a BSOD saying that my system has encountered a problem and will shut down, at stock settings. I can get Intel Burn Test to display that critical error message consistently after the last run on the 16MB test even underclocked at 2GHz.

I can run OCCT Linpack for 6 hours without problems and Prime 95 for 8 hours without problems, but Intel Burn test some times has issues with my system even underclocked.

 

[Dave3000]

Correction on my last post: I ran OCCT Linpack for 10 hours without problems, not 6. Also in Intel Burn Test, even at 1.2GHz, I still have the Linpack binary stopped working message after the last run if I set the memory test size to 16MB for 20 runs.

 

[Liquid_Static]

Are you overclocking your RAM?

 

[moonbogg]

Did you try Intel Burn test before you OCed the chip? I ran for a few days with an auto OC at 4.5ghz with about 1.45 vcore as well and my chip is fine. It is now at 1.28v @ 4.3. I doubt you damaged your chip in any way from your OCing. If other testing programs are fine then I personally wouldn't let it bother me. Do you game? Do the games crash?

 

[BonzaiDuck]

Correction on my last post: I ran OCCT Linpack for 10 hours without problems, not 6. Also in Intel Burn Test, even at 1.2GHz, I still have the Linpack binary stopped working message after the last run if I set the memory test size to 16MB for 20 runs.

First, Liquid_Static's question is pertinent here.

If the IBT testing is showing this consistent error, then the reasons may well be those that were given.

Also, Moonbogg's question. Mine would be "Have you checked to see that you have the latest version of IBT? Did you try to download it again, uninstall then re-install?"

If there were something wrong with your hardware, I would think the other programs -- OCCT and Prime95 -- would also exhibit a problem.

Some of the test programs use the same software components: that is, Linpack. And that would include IBT, LinX and OCCT.

I might agree with MoonBogg that you haven't yet damaged your processor. That being said, the Ivy Bridge cores would operate under a lower "safe" voltage than the Sandy cores. 1.33V is on the high end, but nothing to worry about. I worry instead about the auto-overclock features of your software.

Seems more likely to be a software error, unless there are other things (like RAM) that you didn't mention.

 

[Liquid_Static]

this honestly sounds like a bad ram kit or one that is too highly clocked

 

[DigDog]

the issue is (possibly) with the intel test, and auto voltage regulation which the mobo has;
i'm at work right now so i cant find the video, but it was along the lines of "ASUS - how to overclock manually", on YT, maybe you can find it easily.

essentially, it said that *only* during some synthetics, namely: Intel Burn Test (and a few others), if the mobo is set to auto regulate voltage (to help haswell use less power when the frequencies drop due to non-usage), then when the test runs the cpu will receive an additional spike of voltage above what you set in bios.

this is just so that - i assume - cpus run at higher multipliers will receive a slightly higher voltage when idling, as the idle is a % of the clock, and the standard Intel voltage for idle might not be enough for a OCd idle.

solutions:
1) set your voltages to manual, and disable the low-power state
2) .. ermm ... see 1
3) stop running synthetics.

edit:
here, i found it
www.youtube.com/watch?v=Z7zPu9255ZI‎

at least, in this video, there is a specific mention of auto voltages and IBT giving issues. i dont know if this is what's happening to your cpu, but i hope you solve it.

 

[BonzaiDuck]

the issue is (possibly) with the intel test, and auto voltage regulation which the mobo has;
i'm at work right now so i cant find the video, but it was along the lines of "ASUS - how to overclock manually", on YT, maybe you can find it easily.

essentially, it said that *only* during some synthetics, namely: Intel Burn Test (and a few others), if the mobo is set to auto regulate voltage (to help haswell use less power when the frequencies drop due to non-usage), then when the test runs the cpu will receive an additional spike of voltage above what you set in bios.

this is just so that - i assume - cpus run at higher multipliers will receive a slightly higher voltage when idling, as the idle is a % of the clock, and the standard Intel voltage for idle might not be enough for a OCd idle.

solutions:
1) set your voltages to manual, and disable the low-power state
2) .. ermm ... see 1
3) stop running synthetics.

edit:
here, i found it
www.youtube.com/watch?v=Z7zPu9255ZI‎

at least, in this video, there is a specific mention of auto voltages and IBT giving issues. i dont know if this is what's happening to your cpu, but i hope you solve it.

On my Z68 board, "Auto" disappears when you change it to "offset" mode. Then you pick the offset sign. After that, you can choose between "auto" and "specific fixed-value" of the offset. After that, you can then take "Extra voltage for turbo-mode" off auto and move it to the value that gives you error-free marathon stress-testing.

I think there are "different degrees of 'AUTO'." But DigDog's observations are interesting and maybe the light at end of the tunnel.

 

[FalseChristian]

Id an intel i5/i7 overheats it will throttle the clock speed all over the place. Do not use more than 1.4GHz and don't try to overclock your RAM. I overclocked my Kingston ValueRam 1333MHz to 1600Mhz at 1.6 volts and everything was fine UNTIL I ran Prime95. Core 0 (main core) ran so slow it wasn't funny. Went back to 1333Mhz and Core 0 is fine now.

 

[BonzaiDuck]

Id an intel i5/i7 overheats it will throttle the clock speed all over the place. Do not use more than 1.4GHz and don't try to overclock your RAM. I overclocked my Kingston ValueRam 1333MHz to 1600Mhz at 1.6 volts and everything was fine UNTIL I ran Prime95. Core 0 (main core) ran so slow it wasn't funny. Went back to 1333Mhz and Core 0 is fine now.

Well, I never came close to that with my SB. The throttling will occur at the TJunction spec temperature -- somewhere just short of 100C. I think you can also enable some throttling in BIOS, but we had an argument about that here last year and I lost. Later testing would show that I needed to tweak the "Xtra Turbo voltage" a few steps to get full GFLOPs and no error-correction delays under LinX.

Right now, under LinX CPU-load, my core sensors never exceed a maximum of 72C. My RAM overclock didn't affect temperatures much that I ever noticed. It just caused my system to reset once a week . . . much puzzlement to me in sorting it out [and per a thread I posted here recently and will update once I confirm that it's fixed.]

But with my RAM, the stock spec was 1.5V with a warranty maximum of 1.6. I'd only run up my RAM voltage to 1.53. A whole extra tenth of a volt would definitely add some heat to your configuration -- but I can't imagine it causing your CPU to throttle.

 

[Dave3000]

The XMP settings of my RAM are 1600MHz at 9-9-9-24 2t. I lowered the timings to 8-9-8-24 and Intel Burn Test gave me a critical error message about my system not being stable on the 6th run with a different output result than the previous runs. Is it possible that my RAM is borderline unstable at it's XMP timings or the IMC of my CPU can't handle the XMP settings of my RAM? Also one time memtest froze on me but the cursor was still blinking with this memory with the same CPU at XMP timings. I know that the memory controller of the Ivy Bridge-E officially supports 1600MHz DDR3 but at what timings. I never had this issue with my 3930k or 3820 with Intel Burn Test with my memory at XMP settings.

 

[BonzaiDuck]

The XMP settings of my RAM are 1600MHz at 9-9-9-24 2t. I lowered the timings to 8-9-8-24 and Intel Burn Test gave me a critical error message about my system not being stable on the 6th run with a different output result than the previous runs. Is it possible that my RAM is borderline unstable at it's XMP timings or the IMC of my CPU can't handle the XMP settings of my RAM? Also one time memtest froze on me but the cursor was still blinking with this memory with the same CPU at XMP timings. I know that the memory controller of the Ivy Bridge-E officially supports 1600MHz DDR3 but at what timings. I never had this issue with my 3930k or 3820 with Intel Burn Test with my memory at XMP settings.

If you have the time, set the whole thing back to stock -- even better, leave the RAM at 1600/8-9-8-24. You could even choose to keep your CPU OC settings.

Download HCI-Memtest, make a CD of it to boot the program from a restart.

Now it's your choice according to patience, convenience and "other stuff:" 1,000% is the thorough test, and it might take a couple days with 8GB of DDR3 RAM. 500% is pretty extensive, but another thread I posted recently to be updated in a few more days -- will confirm that even the 1,000% doesn't "catch everything" -- like RAM behavior in EIST idle state.

If you just scan that thread, you should see why I came awake to post on this one.

Also -- you haven't told us what motherboard (and even BIOS revision) you're using. There are other different chipsets, but I can't imagine failure with RAM XMP settings under stock conditions (9-9-9-24 also) with Z68, Z77 or Z87 mobos -- certainly not with socket-2011 boards. UN-LESS! This is about your (8-9-8-24) timings (and/or your vDIMM voltage), maybe the VCCIO -- but just as easily, a bad module.

Which manufacture of RAM are you using? Are you filling all four slots, or just two?

Maybe the best thing to start with is to set everything back to stock, run the RAM at spec timings or XMP, and test with HCI. I could guess you could do this keeping your CPU OC, but maybe better to validate the RAM controlling for everything else.

Once beyond that, if RAM not bad, you can choose to tweak some voltages, or leave the stock 9-9-9-24 timings. How far to twist up the vDIMM depends on what kind you're using: there is a "spec" voltage and a higher "warranty limit" -- usually. Your VCCIO should be >= vDIMM - 0.5V. You shouldn't twist up the VCCIO beyond 1.2V. I had my own "1600-9/9/9/24's" 4x4GB running at 1866 and loosened timings with VCCIO between 1.118 and 1.135V, and increasing it didn't matter -- I still had troubles.

Also -- double-check the VCCIO specs on the IB-E cores, because what I cite here comes from my Sandy system and experience.

 

[Dave3000]

Overnight I ran CPU: OCCT for 8 hours successfully at stock settings. I just have issues with Intel Burn Test and they can crop up within 2-20 minutes and consistently at the end if I use a 16MB test size. I understand that my CPU supports up DDR3-1600 but is there a specific timings that my CPU officially supports at that frequency? Or it does not matter what the timings of the memory are as long as the frequency is officially supported by my CPU then I'm fine? Strange thing is that I never had problems with Intel Burn test on my 3820 and 3930k with this memory and I was wondering my 4820k has a weak memory controller and can't really handle the XMP timings of my DDR3-1600 or my memory is actually defective, or it's just a bug with Intel Burn Test in relation to Ivy Bridge E?

 

[BonzaiDuck]

Overnight I ran CPU: OCCT for 8 hours successfully at stock settings. I just have issues with Intel Burn Test and they can crop up within 2-20 minutes and consistently at the end if I use a 16MB test size. I understand that my CPU supports up DDR3-1600 but is there a specific timings that my CPU officially supports at that frequency? Or it does not matter what the timings of the memory are as long as the frequency is officially supported by my CPU then I'm fine? Strange thing is that I never had problems with Intel Burn test on my 3820 and 3930k with this memory and I was wondering my 4820k has a weak memory controller and can't really handle the XMP timings of my DDR3-1600 or my memory is actually defective, or it's just a bug with Intel Burn Test in relation to Ivy Bridge E?

Now I'm vaguely remembering the overall import of your posts.

When you bought the RAM, did you check the "Configurator" link of the manufacturer to see if was compatible with your motherboard? Chances are, it is, but you don't act like you know for sure.

The timings -- latencies -- of the RAM are not a function of the processor, assuming the RAM works with the motherboard. The timings are a feature of the quality and design of the RAM. I simply cannot accept the notion that your processor somehow is "too weak" for the RAM at DDR3-1600 speeds, and frankly, that processor can handle RAM up to the DDR3-2400 range as well as the slower modules at 1333, 1600, etc.

Didn't another poster mention something about problems with IBT with certain hardware features? Didn't DigDog suggest that?

There are other test softwares that provide the same thing as IBT. OCCT and LinX all use the same Linpack stress routines to equal usefulness. If you ran OCCT for 8 hours with stock settings, then use it to validate your overclock settings. Or try LinX.

If you tested your RAM when you built the machine, or as I suggested earlier and it doesn't fail at stock settings, then there's obviously nothing wrong with it. If you want to overclock the RAM, research the experiences of others here or at other forums with that particular make, model and size of RAM -- it will give you an idea of what timings to use at higher speeds.

If you're running these as DDR3-1600 and the stock timings are 9-9-9-24, those timings are better than average for that speed.

The way you sound in your posts, you seem a little unsure of yourself in this business. For now, just run the RAM at their stock, factory spec -- even if you have to manually input the settings to the BIOS. For stress-testing -- you don't need to use Intel Burn Test if either OCCT or LinX are good for 8 or 10 hours. And frankly, I don't think you need to run those programs for that long. You should probably shoot for 20 or 30 iterations.

Stressing with Prime95 usually takes longer, and 18 hours is considered reasonably thorough. The others should give you the same validation of your OC in less time. But follow the advice of the SW author/provider.

 

[Dave3000]

I know it's not my motherboard as I already replaced my motherboard with the same one and still Intel Burn Test sometimes acts weird on me, no matter if I'm at stock turbo or full turbo on all cores. I suspect it's a memory problem because I'm having problems with X-Plane 10.25 64-bit with a certain 3rd party aircraft kicking me back to the desktop and crashing the game and tech support of that 3rd party aircraft told me that based on the log file that the problem could be related to running out of available memory in my system which he mentioned is unusual because I have 16GB of memory in my system. Also one time Memtest 86+ froze on me even though the cursor was still blinking. Then again no problems with Prime 95, LinX, or OCCT with full turbo boost synced to all cores. I also still had problems with Xplane 10.25 with my RAM set to the Auto 1333 setting, no XMP, but the problem took longer to occur. Anyways I just ordered 32GB of Crucial Ballistix Sport DDR3-1600 to see if my RAM is the culprit.

 

[readymix]

Intel® Processor Diagnostic Tool

 

[Dave3000]

Everytime I ran the Intel Processor Diagnostic Tool it fails the bus speed test.

 

[BonzaiDuck]

I know it's not my motherboard as I already replaced my motherboard with the same one and still Intel Burn Test sometimes acts weird on me, no matter if I'm at stock turbo or full turbo on all cores. I suspect it's a memory problem because I'm having problems with X-Plane 10.25 64-bit with a certain 3rd party aircraft kicking me back to the desktop and crashing the game and tech support of that 3rd party aircraft told me that based on the log file that the problem could be related to running out of available memory in my system which he mentioned is unusual because I have 16GB of memory in my system. Also one time Memtest 86+ froze on me even though the cursor was still blinking. Then again no problems with Prime 95, LinX, or OCCT with full turbo boost synced to all cores. I also still had problems with Xplane 10.25 with my RAM set to the Auto 1333 setting, no XMP, but the problem took longer to occur. Anyways I just ordered 32GB of Crucial Ballistix Sport DDR3-1600 to see if my RAM is the culprit.

Hey! Don't mean to be rude, but do you know what happened to the FA-22 Raptor control panel in X-Plane 10? I talked to their tech-rep and they say they "still have to revise" some choices, then referred me to some websites for plug-ins, but I couldn't find the Raptor.

Also -- if you're sure it froze with Memtest, I'd say the RAM is suspect. I used to use MEMTEST86 and a similar or earlier version of it. If you can, try HCI Memtest.

You didn't say what sort of motherboard you're using. Or -- after quickly reviewing all the posts here, I missed any reference to it.

 

[Dave3000]

I just tried HCI Memtest and right after I launched a 5th instance of Memtest with a 2048MB size my PC crashed. I had to do a hard restart of my PC and got a message at post saying Overclocking failure, press F1 to enter setup. My PC was not overclocked and I was running my memory at advertised XMP settings but manually entered instead of using the XMP profile. I don't use the XMP profile because my motherboard runs all cores at full turbo boost if I set AI Overclock to XMP or the memory setting to XMP profile.

 

[Ed1]

I just tried HCI Memtest and right after I launched a 5th instance of Memtest with a 2048MB size my PC crashed. I had to do a hard restart of my PC and got a message at post saying Overclocking failure, press F1 to enter setup. My PC was not overclocked and I was running my memory at advertised XMP settings but manually entered instead of using the XMP profile. I don't use the XMP profile because my motherboard runs all cores at full turbo boost if I set AI Overclock to XMP or the memory setting to XMP profile.

You still didn't mention MB version , On Asus you can disable the Asus multi-core enhancement" that will make it not do OC when using XMP .

I would try lower speed on ram with lose timings (1333@ 9-9-9-24 2t ) and see if it then passes a test, if so you know its the ram for sure .
Also if your populating all 4 slots you might need bump on voltage for ram , 1.55v or so .

 

[BonzaiDuck]

You still didn't mention MB version , On Asus you can disable the Asus multi-core enhancement" that will make it not do OC when using XMP .

I would try lower speed on ram with lose timings (1333@ 9-9-9-24 2t ) and see if it then passes a test, if so you know its the ram for sure .
Also if your populating all 4 slots you might need bump on voltage for ram , 1.55v or so .

Sounds like good advice. My Ripjaws "Gerbils" (4x4) spec'd 1600 needed 1.55V to run stable at 1866 and looser timings. I had begun to worry that they were going bad, but -- no. In the OP's case, if they don't run at spec with the right voltage and timings, then he should consider getting a 2x8GB kit or something better (socket 1155 or 1150) or a 4x4 kit (socket 2011). At minimum, he should test each module until he finds the problem-child. With lifetime warranties offered by the RAM-makers, they could just replace it for him costing the postage -- less trouble for the wallet by far.

 

[Burpo]

Don't know what board you're using, but if you're using the Asus P9 X79 board you'll need at least Ver. 4210 bios. preferably 4302. Set to default settings for boot, then run Intel Processor Diagnostic Tool. I'm sure you didn't damage the processor in 10 minutes of overclocking.

 

[tweakboy]

No reason to worry about this right now... all is good

 

[tweakboy]

Nope. your fine until it reaches 95c ,, after that the cpu will degrade big time. gl

 

[Dave3000]

Today I replaced my 4820k with a 4930k to see if my CPU was the problem. I was going to upgrade to a 4930k anyways at some point. Still having the same problem with Xplane 10.25 and Intel Burn Test in the 16MB /20 run test with the same error message at the end. Xplane 10.25's log file says that my system was out of memory in the last several lines of the file when the crash to the desktop occurred. It's only when I fly a specific aircraft in this game.