First Overclock has produced weird error

[neo-saitek]

Hi...

My first overclock has produced a weird error.. im using i5-2500k on asrock pro 67se mobo with 16gb vengeance ram at 1600mhz and the asaka venom voodoo cooler..

got my overclock to 5ghz stable on prime 95 for 24hrs.. no errors. cpu temps were 55-62degrees tops under prime95 idling at 28degrees when i closed it down

when i restarted my pc later on i had a "machine check error"..

ive not had this until i did my first overclock.. so im back on the default bios and not going to do it again..

any ideas why this might have happened and should i be worried?

 

[ShintaiDK]

Its not stable then. Running prime 95 doesnt tell much Since it only uses a small amount of instructions.

 

[neo-saitek]

Its not stable then. Running prime 95 doesnt tell much Since it only uses a small amount of instructions.

any ideas why this might have happened and should i be worried?

 

[ShintaiDK]

any ideas why this might have happened and should i be worried?

Because its not stable. Data corruption and crashes is the result.

 

[BonzaiDuck]

You're going to find forum veterans who are very proud that their Sandy Bridge K's could reach 5 Ghz. But even with water-cooled strategies, I've never seen any of our brethren end their proclamations confirming they were comfortable or even capable of making their 5 Ghz benchmark stable for 24/7 operation. And if such a user exists, I missed their thread.

While your temperature profile suggests to me that you might be water-cooling, that's still not going to cut it. And since this is your first over-clocking adventure, I could guess you haven't found the patience needed to accurately assess the limitations of the processor.

Some have boasted that their i7-2700Ks were running 24/7 @ 4.8, and in my exchanges with them I never thought to ask if they had enabled Hyper-Threading. This is a moot point for the 2500K chip, since that model has no Hyper-Threading option. Since I can run my 2600K and 2700K systems @ 4.7 HT-enabled, I also find that 4.8 means too much voltage for my comfort. And I actually never tried finding my over-clock comfort-zone with disabled HT.

Yes -- you may be able to boot your system into Windows with a 5GHZ clock-setting. I just don't think you'll get there for 24/7 operation. You should go through the painstaking process of increasing your clock speed 100 Mhz at a time, and at some point -- doing a thorough affinitized LinX for over 30 iterations or OCCT:CPU for 4 hours at one or more of those experimental clock settings. Ultimately, if you use PRIME95 as your main, marathon stress-tester (I don't), you should expect it to go 12 to 18 hours error-free.

And that's the reason many of us don't depend on PRIME95 for those assurances: it takes too long, while it submits the processor to load-level stress similar to the shorter tests.

Looking closer at your post, I see you found "24-hours Prime-stable" in your testing. Which test -- sFFT or lFFT -- were you using?

Also, if you asked me, I'd say that the drooped voltage for that processor shouldn't exceed 1.37V. Many folks may disagree, and in the early Sandy Bridge days, some "overclock advisor" hot-dawgs were encouraging users to push the voltage as high as 1.44V. If it were me, I'd see how far I get before loaded VCORE is bouncing between 1.35 and 1.39, and then -- call it a day -- refining my settings for that particular clock level. I might even drop it back 100 Mhz once I'd achieved "stability" with that sort of self-imposed VCORE limit.