OC not stable over time, why?

[tuffluck]

hi all, i have a 4 year old PC that has the following:

1. e6750
2. g.skill 2gb 6400 RAM
3. radeon 5770
4. Gigabyte DS3L Mobo
5. non-stock cooling (forget the heatsink/fan combo, a popular one at the time)
6. NB fan

i OC'd the CPU about 3 years ago to 3.4ghz (425fsb). it ran stable for a long time, at least a year as i checked and ran benchmarks on it that next year. i set it up running p95 for 24 hours stable and it never got too hot.

i decided to buy bf3 recently and try it on my rig. while it plays fine, i noticed that cpu-z was showing no OC. i went into the bios and somehow the "cpu host clock control" had been disabled, but all other bios settings were intact. i tried to reset to 425mhz and then saved and rebooted, and the computer turned itself on then off immediately about 3 times in a row, then started up fine but defaulted out the OC again. i did some research and some people suggested this was a common problem with the DS3L board and you had to reset the cmos to fix it. tried this and it didn't work.

then i got the hunch to try a small OC and see if the PC would restart with even the smallest clock. well in doing this, i got the OC up to 360mhz FSB (up from 333 stock, down from 425 original OC), and this is all i could run stable at. over 365 and the PC would just turn itself off immediately. at 365 i booted to windows and crashed. i have 360 now as i type this, but no stress tests yet so not sure if this is even a stable clock.

but i'm baffled as to how my computer just lost the ability to OC. does this happen sometimes? or is it possible the problem is attributed to a specific hardware component like my RAM or other? i just want to know if i can get back to the 425 OC or if i'm going to have to live with a lower clock or even stock speed. if it's something like RAM, i could just spend $40 on new RAM to fix. if it's anything more expensive, i'm better off just upgrading the rig altogether.

just FYI, nothing fishy has occurred with my PC in the last 4 years to indicate there was something wrong with the OC. it just randomly reset itself somehow.

thanks in advance.

 

[farrengottu]

its not the ram, i know that much. your mother board could have a voltage regulator or a capacitor that is going bad. could also be your processor cant handle that frequency without shorting anymore. its time for an upgrade anyways. next month you have held out for 3 generations. time to buy some ivy bridge.

 

[elbert]

Over a span of 4 years will dry out most thermal grease. The only one able to do well this long is AS5. All thermal grease's dry out or the silver tarnish with AS5 over about 5~7 years making them less effective. Is your voltage settings defaulted to normal? Other possibility's is a weaker PSU due to aging is no longer able to push your overclock. Possibly you have added new part such as RAM or your RAM can no longer take that overclock. Lastly overclocking increase the rate of electrical migration that kills the CPU.

I suggest the cheapest route of re-appling some good thermal grease.

 

[tuffluck]

i have not added any new components since the OC. actually i forgot to mention to achieve 425mhz stable, i had to bump the vcore to 1.44375, which is significantly higher than 1.35 vcore stock. even though that is high, trust me that temps were well within a safe range with the OC.

for 360mhz i actually bumped it all the way back down to stock 1.35 and it's running fine so far (and stress testing). i may have to bump it up a little bit more from here, but definitely nowhere near 1.44ish.

i have AS5, but wouldn't fading AS5 mean higher temps? the temps have not increased, and also, the vcore has stayed at 1.44375 this whole time and has not defaulted. only the fsb has defaulted to stock.

 

[tuffluck]

thermaltake 500w

http://www.newegg.com/Product/Product.aspx?Item=N82E16817153028

 

[Soda-88]

most likely dried up thermal paste as elbert mentioned

 

[tuffluck]

so how exactly would dried up thermal paste be causing this? i just don't follow how it is causing my mobo to reset to default speed if the vcore is not also changing, and if temps are remaining the same.

 

[wolfram23]

Um... CPUs do degrade over time.

What is the stock voltage vs overclock voltage? If you left it on Auto, you'll want to run take CPUZ or HWMonitor readings and compare.

It's possible that it's something else. Could be the motherboard is dying, or the PSU. But it's not out of the question that such a big OC would degrade the CPU after 4 years. Usually you need to give it more voltage to overcome the degradation, but of course that just leads to faster degradation. Better to use a lower OC.

 

[Wisecracker]

My experience with the Thermaltake 500w is yah get clean power to 350w or so. Over time you may see some degrading.

Run OCCT and check the charts for ripple.

 

[tuffluck]

What is the stock voltage vs overclock voltage?

stock is 1.35. original OC was 1.44375 @ 425fsb. now it will run 360mhz @ 1.35v but will not run 361fsb @ 1.44375v. right now it doesn't appear the problem is VCORE but it is FSB. that doesn't make sense?

If you have another PSU to test with, I would highly recommend that. Also, change your CMOS battery.

i don't have another PSU unfortunately, but the vcore has been 1.44375 for 4 years now without problems...the mobo NEVER reset that function, only the FSB. wouldn't clock speed be irrelevant for PSU requirements and VCORE be what stressed the PSU? and wouldn't a failing CMOS battery mean all settings are being erased and not just fsb clock?

Run OCCT and check the charts for ripple.

that program gave me a BSOD (w7 64bit) after about 100 seconds. i had been running p95 for 8 hours up till that point and stopped it to run OCCT. i would think if i was stable with p95 for 8 hours but not with OCCT for 2 minutes, that doesn't necessarily point to my CPU being unstable? btw, in that 2 minutes there were no power spikes at all, if that's what you were asking.

 

[wolfram23]

Hey, just curious. Besides the potential for hardware malfunction, have you changed the FSBRAM ratio? Might be that the DRAM is going too high now if you forgot about that.

 

[elbert]

so how exactly would dried up thermal paste be causing this? i just don't follow how it is causing my mobo to reset to default speed if the vcore is not also changing, and if temps are remaining the same.

With most overclocking thermals play a part in how much voltage is needed for stable operation. IE LN2 overclocks can be far higher but on less voltage. Small changes in temperate can effect your max overclock in this way. Not sure this is the case but being its the lowest cost to check I suggest this first. Could your motherboard be setting a safe boot FSB due to instability?


Is it possible windows 7 may be adversely affecting your bios. Possible an incompatibility. Possible your windows 7 is unstable? Windows 7 original is very unstable with many new games causing kernal trap mode BSOD's.
http://www.gigabyte.com/products/product-page.aspx?pid=2629#bios

 

[tuffluck]

With most overclocking thermals play a part in how much voltage is needed for stable operation. IE LN2 overclocks can be far higher but on less voltage. Small changes in temperate can effect your max overclock in this way. Not sure this is the case but being its the lowest cost to check I suggest this first. Could your motherboard be setting a safe boot FSB due to instability?


Is it possible windows 7 may be adversely affecting your bios. Possible an incompatibility. Possible your windows 7 is unstable? Windows 7 original is very unstable with many new games causing kernal trap mode BSOD's.
http://www.gigabyte.com/products/product-page.aspx?pid=2629#bios

i am never booting into W7, so i can't see how it would affect the OC. if i set the FSB too high, my computer restarts then immediately shuts off, then restarts again. repeats this 2 or 3 times until it restarts and runs normally. i go into the bios at this point and the OC was completely turned off automatically. during this process W7 never came into play until the OC is first (somehow) disabled.

it could very well be the mobo disabling the OC (sounds like it is), but how would i know this is happening or why this is happening? i just don't really understand how the vcore has no impact on the max clock of the chip...360mhz is as high as it will go, 1.35v or 1.45v, etc. i did read this below--which sounded really promising since this article refers to my exact MB--but it did not help.

I have DS3L and this issue is common.

What happens is that the BIOS is set to error on the side of caution and pretty much assume that any computer reboot or system shutdown which isn't "clean" must be a bad overclock.

To "clear" the lingering memory of this false-positive you have to unplug the PSU from the outlet and set the mobo clear cmos jumper and leave it like that for about a minute.

Then set your clear cmos jumper back to default, then plug the PSU back into the wall. I have confirmed repeatedly that you must unplug the PSU from the wall to make the process work.

The next thing you must do when your system reboots is go into the BIOS and first select the load defaults option, save (F10) and reboot.

Then go back into the BIOS and setup your OC options. Save and reboot.

Now at this point in the timeline there is some kind of reboot/cold-boot counter in the BIOS and if it doesn't register a controlled shutdown followed by a clean cold-boot then it will assume that any subsequent reboots (initiated by you, windows, or a failed OC) are all counted as failed OC's.

So once you set your OC settings and the BIOS does it's mandatory reboot to get out of the BIOS screen the very next thing you need to do whenever the reboot is complete and you are in windows or linux is you must do a hard-shutdown (not a restart). Let the computer sit at least a minute after shutdown before booting again.

This sets whatever register Gigabyte uses for tracking "known good BIOS settings" as your OC settings and now you can initate reboots/restarts and not have the OC get mis-percieved as a bad OC.

 

[elbert]

i am never booting into W7, so i can't see how it would affect the OC. if i set the FSB too high, my computer restarts then immediately shuts off, then restarts again. repeats this 2 or 3 times until it restarts and runs normally. i go into the bios at this point and the OC was completely turned off automatically. during this process W7 never came into play until the OC is first (somehow) disabled.

it could very well be the mobo disabling the OC (sounds like it is), but how would i know this is happening or why this is happening? i just don't really understand how the vcore has no impact on the max clock of the chip...360mhz is as high as it will go, 1.35v or 1.45v, etc. i did read this below--which sounded really promising since this article refers to my exact MB--but it did not help.

Many motherboards are designed after a number of fails to clears cmos. Mine does this on 3 failed attempts. IE unstable boots which you will need to find a new stable overclock or attempt fixes. This may well be by design to keep you from pulling the cmos batter each time. Given the age of the system I suggest live with the max OC you can get stable. Maybe try some cheap fixes like the thermal grease or a RAM upgrade. With 4 years of a very high OC your CPU has defiantly went beyond its best years.

As to what you quoted if that where the case your 360 OC should have failed. I thinks its just to high a setting for some part of your system to handle given its age. IE what triggered this in the first place wasn't you increasing the overclock but your system was unable to stable boot with older settings.

If your running dual channel RAM try overclocking with only 1. If this works with either one of the sticks you may try a RAM upgrade. Just don't put to much cash in this old system. Here is about the best ddr2 RAM I have ever used.
http://www.newegg.com/Product/Product.aspx?Item=N82E16820231122
Lower latency this is a bit better but about to pricey.
http://www.newegg.com/Product/Product.aspx?Item=N82E16820231209

Save up for a new build as many games are recommending quad cores. Soon quad cores will be a requirement. BF3 is one game recommending a quad core.

 

[Wisecracker]

...

that program gave me a BSOD (w7 64bit) after about 100 seconds. i had been running p95 for 8 hours up till that point and stopped it to run OCCT. i would think if i was stable with p95 for 8 hours but not with OCCT for 2 minutes, that doesn't necessarily point to my CPU being unstable? btw, in that 2 minutes there were no power spikes at all, if that's what you were asking.

A drop in volts may have caused the blue screen, and was not picked up because OCCT (and the system) crashed.

 

[tuffluck]

elbert and wisecracker, thanks a lot for the information! is there any other PSU testing software out there i could try? i agree not to put much money in the system, and i don't plan to.

believe it or not, bf3 plays pretty well. as long as my hard drive isn't dying, i guess the other components can fail at their leisure. hopefully with bringing the vcore back down to 1.35 and just running what OC i can at that stock voltage, i won't have to worry about the system dying any time soon.