P8P67 Pro w/ Prime95 - One Core always has Hardware Failure/Fatal Error

[Havokator]

When I am overclocking my P8P67 Pro, it seems like it is always Core2 that gives me the Hardware Failure Fatal Error in Prime95.

I can run Prime95 for 48 hours and the other 3 cores will perform fine, but usually after a few hours Core2 will error.

This happens at various overclocking speeds/voltages, but always Core2.

Any ideas why this is happening? Should I consider the overclock not stable if Core2 is the only one that errors?

[Falkentyne]

Not all cores are created equal. If it happens at stock, RMA the chip. Why not just increase the voltages? You should have listed what overclocks, though.

If it happens at 4 ghz, even at 1.3v, I think it's safe to say that the CPU is defective, and you should send it to Intel. Even they will admit that there is no such thing as a 2500/2600K that can't do 4 ghz with 1.3v...that's just 200 mhz above the official turbo voltage....

[Havokator]

You are right. I should have mentioned that this is a Intel i5 2500k

It happens at 4.5ghz with 1.30v, 1.32v, 1.33v
All 3 other cores are stable at these voltages.

Not all cores are created equal. If it happens at stock, RMA the chip. Why not just increase the voltages? You should have listed what overclocks, though.

If it happens at 4 ghz, even at 1.3v, I think it's safe to say that the CPU is defective, and you should send it to Intel. Even they will admit that there is no such thing as a 2500/2600K that can't do 4 ghz with 1.3v...that's just 200 mhz above the official turbo voltage....

[Chickenfeed]

Bump up the voltage to 1.4 and work your way down from there. Failing that, as mentioned, test at lower clocks as well. In all likelihood you just need more voltage. It shouldn't need 1.4 but if it isn't stable there then its either a very weak cpu or there is another voltage / problem limiting stability.

It also depends on the board used and load line calibration employed. It is very hard, if not pointless, to compare overclocks with different samples when only mentioning the cpu voltage alone so when so and so claims such and such a clock speed with a certain voltage you can't compare it directly. Voltage set in the bios vs actual load voltage are 2 very different things. ( just had to include mention of this before someone comes in here saying you don't need that much voltage )

[Havokator]

Everything is at default settings except LLC is at +75% (ultra high I think). That is what it takes to keep the voltage steady when under full load.

Should I really need more than 1.33v for 4.5ghz? Is this a bad overclocking chip you think? What is the average voltage needed for 4.5ghz?

Bump up the voltage to 1.4 and work your way down from there. Failing that, as mentioned, test at lower clocks as well. In all likelihood you just need more voltage. It shouldn't need 1.4 but if it isn't stable there then its either a very weak cpu or there is another voltage / problem limiting stability.

It also depends on the board used and load line calibration employed. It is very hard, if not pointless, to compare overclocks with different samples when only mentioning the cpu voltage alone so when so and so claims such and such a clock speed with a certain voltage you can't compare it directly. Voltage set in the bios vs actual load voltage are 2 very different things. ( just had to include mention of this before someone comes in here saying you don't need that much voltage )

[Falkentyne]

Just try what the other poster said. You won't know if the CPU is defective or if you simply don't have enough vcore. Having one weak core tends to happen from time to time.....

[silis]

And make sure your memory is not faulty, had a same problem failing always in prime if memory were stressed, could be after 2 or 6 hrs. One of my sticks was bad..fastest way to test is run small ffts it errors faster than a blend test hope you get it sorted out !

@i5 2500k | 4,5ghz / Sabertooth P67 B3 (1502) / G-Skill Ripjaws X 1600mhz 8GB CL9 / HD 6950 2GB (shader unlocked)

Forget to add, one stick at time, fails pretty fast.

[Falkentyne]

Didn't consider that. However in my experience, RAM errors are ALWAYS caught fastest with prime blend. Whenever I used to overclock too far for the northbridge, prime small FFT would run forever, while Blend would crash within minutes or seconds.

Remember, the RAM isn't being stressed if the data all fits within the L2 or L3 cache. And 8k-64k FFT's all fit inside the cpu cache...

[st0ned]

Yup your chip is bad. Live with it. A friend of mine's have a 2500k that will need as much as 1.39v for 4.6Ghz... I've a 2600k that can do 4.5 with 1.21v

[Demo]

1.21v is AWESOME for 4.5Ghz!! Id say test your chip at stock speeds if it fails then RMA it.. if not then its a bad clocker..

@OP

My chip requires requires something like 1.350~1.360 to be stable at 4.5Ghz so yeah.. Id consider that a bad clocker like mine..

[Skratch]

Your chip is fine.up the voltage a tad and enable pll over voltage in the bios.

[st0ned]

1.21v is AWESOME for 4.5Ghz!! Id say test your chip at stock speeds if it fails then RMA it.. if not then its a bad clocker..

@OP

My chip requires requires something like 1.350~1.360 to be stable at 4.5Ghz so yeah.. Id consider that a bad clocker like mine..

Yup it also scales nicely with voltage. 1.368v for 5Ghz and 1.416v for 5.1 . All 20+ linx runs with 8gb RAM stable.

Your chip is fine.up the voltage a tad and enable pll over voltage in the bios.

Useless to enable Pll... If it needed it wouldn't boot

[Skratch]

My 2500k will boot fine with it off and is not stabke after 4600mhz without it enabled.

[st0ned]

My 2500k will boot fine with it off and is not stabke after 4600mhz without it enabled.

That's really odd... All chips I've own so far, for exemple the one I'm using right now. 50x no PLL all ok 1.368v linx stable ; 51x no PLL = no boot at all.

[Skratch]

That's really odd... All chips I've own so far, for exemple the one I'm using right now. 50x no PLL all ok 1.368v linx stable ; 51x no PLL = no boot at all.

IM using same board as op and I can boot fine up to 4700 ish in windows but will fail some tests and once its on the chip is rock stable.

Im running 4 sticks of 2133 ram so I dont know if its from the extra strain on the memory controller but fo9r me pll over volt makes a huge difference.

Im supprised you got a 50x multi with it off

[Havokator]

IM using same board as op and I can boot fine up to 4700 ish in windows but will fail some tests and once its on the chip is rock stable.

Im running 4 sticks of 2133 ram so I dont know if its from the extra strain on the memory controller but fo9r me pll over volt makes a huge difference.

Im supprised you got a 50x multi with it off

What vcore do you need to get 4.7 stable??

I wanna know how bad it is to need 1.35v for 4.5ghz on a 2500k on the p8p67 pro. Is it bad enough that I should return the chip and get a different one? I bought it from a local store so it would be easy to return.

[st0ned]

What vcore do you need to get 4.7 stable??

I wanna know how bad it is to need 1.35v for 4.5ghz on a 2500k on the p8p67 pro. Is it bad enough that I should return the chip and get a different one? I bought it from a local store so it would be easy to return.

It's bad, if you can trade it for another one I wouldn't see why not... And it would be hard to get a worst one.

[Skratch]

Mine needs about 1.390 to get 4.7 stable and 1.420 for 4.8 then 1.520 for 5.1 ghz

[PiLsY]

Set your memory to 10.66x while finding the cpu clock limit. Probably not cpu causing this.

Whats vccio set at and do you have current capability at 140%, switching to extreme etc? Wouldnt bother with LLC over 50% on the Pro.

[ghmz]

Not all cores are created equal. If it happens at stock, RMA the chip. Why not just increase the voltages? You should have listed what overclocks, though.

If it happens at 4 ghz, even at 1.3v, I think it's safe to say that the CPU is defective, and you should send it to Intel. Even they will admit that there is no such thing as a 2500/2600K that can't do 4 ghz with 1.3v...that's just 200 mhz above the official turbo voltage....

Incorrect. Failing prime is not an indicator of a faulty chip, even at stock speeds.

[ghmz]

When I am overclocking my P8P67 Pro, it seems like it is always Core2 that gives me the Hardware Failure Fatal Error in Prime95.

I can run Prime95 for 48 hours and the other 3 cores will perform fine, but usually after a few hours Core2 will error.

This happens at various overclocking speeds/voltages, but always Core2.

Any ideas why this is happening? Should I consider the overclock not stable if Core2 is the only one that errors?

Only Core4 fails on me though. How do you explain that? Increasing the voltage helped.

[Havokator]

Well I just went out and returned my 2500k for a new one, Costa Rica this time (last was Malay). I booted it at 4.5ghz with 1.28v and it has been running Prime95 solid for the last 4 hours.

What an improvement!

My last chip needed 1.35v. We will see if it passes Prime95 for 24 hours, but the last chip couldnt even RUN Prime at these settings.