To those who crunch on overclocked rigs

[dkm777]

I figured this would be a better place to ask since CPU is mostly AMD trolling nowadays. If you crunch on an overclocked rig how long did it take for your processor to degrade? I don't crunch 24/7, but crunch pretty often nonetheless and after updating F@H client to the latest version I had to bump the voltage of my i7 980@4GHz ever so slightly. This makes me wonder - is my processor showing first signs of degradation or was that just the latest F@H being more picky? I've been running this rig for two years already and I figured with all the scary stories it was about time to see degradation.

 

[sangyup81]

Hard to say because it could be your power supply degrading

 

[VirtualLarry]

I don't run my CPUs at extreme temp or voltage, so I've never experienced degradation on my own gear. I did see a friend's E5200 degrade slightly, after running at 1.425v (BIOS) for a few years.

 

[ZipSpeed]

I'm not an extreme overclocking either. I basically overclock as far as I can on stock voltage. Sometimes I get lucky and I can undervolt. My two gems in my collection of crunchers are my 970 @ 3.9 GHz running at 1.2V and my 3770K @ 4.2 GHz running at 1.1V. The rest run around stock voltage.

 

[dkm777]

Hard to say because it could be your power supply degrading

This is a possibility, since my Seasonic X-750 went through 3 builds and was tortured by GTX570 SLI.

 

[Drsignguy]

I have oc'ed all my rigs and ran them 24/7 for a long time and while I did that, no signs of degrading at all. Although I must mention, I did not over volt nor did I push the extreme out of any of my cpu's or vid cards. Now, I have set all back to default levels and run only 2 rigs (for now). F@H can show crazy things to your setup if you don't have it properly over clocked. Personally, I think it's is a hell of a stress test.

By you bumping up the voltage, doesn't seem strange at all. I had to do that to most my rigs when I oc'ed them while running F@H. Prime passed but when I ran a work unit of F@H, failed. Had to bump the volts up just a tick and all worked just fine.

 

[Kougar]

You shouldn't expect to see any CPU degradation unless the CPU is being cooked and/or the voltages are pushed too far. As others have said, it can be the motherboard VRM, the PSU, or maybe the OC was only 99.9% stable to begin with... lots of possibilities.

I've run some heavy overclocks under 24/7 folding and just not seen it happen. The one time I thought I had where I was increasing CPU voltage, long story short turned out to be an intermittent issue with overly aggressive RAM settings beyond what the 920's memory controllers could handle reliably under a sustained load. Memtest simply isn't a substitute for a sustained CPU-wide/system-wide load so it never failed the RAM settings.

The longest running OC would be an E6300 from launch day 2006, with a 100% overclock 1.86Ghz to 3.8Ghz after I began using water cooling. The chip never needed voltage adjustments despite the 24/7 F@H load for several years, and both the mainboard & E6300 work fine today with a milder, power-saving 3Ghz OC as my father's PC.

 

[StitchExperimen]

My pass fail now has been updated to run 2 programs at the same time. One is IBT (intel burn test) and Prime 95. This will determine for me the Frequency I can run at and if boosting in the voltage will make it stable. Also it depends on the motherboard which enabled me to get a 3770K to 4.5GHz an ASRock Z77 OC formula but the cost is $225 which is inexpensive compared to some of their all inclusive boards. But there are some older boards at cheaper prices to cut build cost. Asrock Z77 Extreme 4 has quite a following at a price of $135 its not the new bells and whistles but $90 cheaper.
I've only seen only one close to o/c degradation failure it was a i7 3930 that failed in the first month. Under no load at stock voltage @ 4.5GHz it would run Windows 7 pro. Since 4.2GHz is a standard for this chip to run 24/7 I sat it for this and bumped the voltage twice for stability and long term use. I tried 4.3Ghz to get more hcc1 points and would get random errors so I tried 4.5GHz and no start, upped the voltage and no start 4.4GHZ no start, back to 4.2 and then later random errors (and I was wondering if it was the video cards)... end of story I was at base frequency of 3.2GHz and scientific programs were producing computational errors. I sent the chip back to Newegg and whoever finally tested it didn't contest validity.
I've seen 10's of thousands of older style cpu's in production and saw almost zilch failure except op-amps and old design spec incompatible roll memory. But early end of life failure for components that had connectors causing low voltage or heat so high caps let through AC into a DC circuit on high loading times. Many a repair was just re-seating the card connector or cable and if you cared you tightened the pinch on the pins. There was very much a dark art DC circuits sometimes as little as a 40mv AC signal riding on DC could shut a circuit down. Before we had UL switching power supplies and any nice thing that would regulate 100 amps they had 1,000 amp DC @ 5V with no transformer straight from the power line running computers this was recessed in the sub-flooring with a man standing over the top monitoring and by standing he was standing on the floor joists.