i5 2500K overclock question

[DarkBahamut]

Interesting to see people seeing some degrading in the overclocks on these chips. I always wondered if it might happen, but I can't say I've noticed much on my 2700k so far. Currently holding 4.6GHz with a reported 1.304v during benchmarks. I've actually been tempted to start pushing a bit harder recently

 

[eddiew]

I always knew that chip degradation was a possibility when overclocking, but I never expected to be running the same CPU for 5 years either...

That said, I discovered recently that an instability that's been dogging me for months was down to memory speed. Backed it down from 1866 to 1600 and my rig's totally solid again. Wonder if I could get the overclock back up if I got some fresh paste in there now...

 

[Beasty]

My 4.6 that was clocked by OCUK when I ordered it back in 2011 became unstable last year, I've got it back stable at 4.5 now at 1.352, could still do some more work on it but its rock steady and temps dont go above 70.

What a boss CPU!

 

[zipp0r]

I always knew that chip degradation was a possibility when overclocking, but I never expected to be running the same CPU for 5 years either...

That said, I discovered recently that an instability that's been dogging me for months was down to memory speed. Backed it down from 1866 to 1600 and my rig's totally solid again. Wonder if I could get the overclock back up if I got some fresh paste in there now...

mine needed vccio increasing for 1866 ram

 

[eddiew]

mine needed vccio increasing for 1866 ram

Cheers for the tip, I'll have a look at it

 

[Seanoog]

101 does tend to be vcore and small ftt is the cpu test. large is more ram and imc than cpu. llc and pll ov can help but those shouldnt really be needed where you're at.

check the actual vcore it gets under load on the small ftt test with cpuz(dont look at VID in other monitoring apps, thats not the vcore its getting). with dvid you would be at around 1.35v with +0.100v ??

Hi zipp0r.
So to clarify, the vcore on CPUz is the best/most accurate way of measuring the voltage used by the cpu when idle or load?

I have been itching to push my i5 2500K harder but unsure how to do it with my temperamental ASrock Z68 Extreme4 Gen3 Motherboard.
Still according too cpuz vcore ideal is 1.000 +-0.05
While load is 1.300 at 4.6Ghz.

I'll be honest, my system doesn't seem to have deteriorated all that much, from a vcore perspective, where the system ran at 0.900 for some time, while ideal.
Any good suggestions to push this CPU higher?

 

[zipp0r]

Hi zipp0r.
So to clarify, the vcore on CPUz is the best/most accurate way of measuring the voltage used by the cpu when idle or load?

I have been itching to push my i5 2500K harder but unsure how to do it with my temperamental ASrock Z68 Extreme4 Gen3 Motherboard.
Still according too cpuz vcore ideal is 1.000 +-0.05
While load is 1.300 at 4.6Ghz.

I'll be honest, my system doesn't seem to have deteriorated all that much, from a vcore perspective, where the system ran at 0.900 for some time, while ideal.
Any good suggestions to push this CPU higher?

cpuz is one way, HWmonitor is another, there are others too, all read the same sensor. sometimes people see the VID in other apps and think its the actual voltage when it isnt, thats why i suggested cpuz.
anyway your cpu sounds better than mine if it needs only 1.300v for 4.6

just do the usual, increase multplier and test with p95 small ftt to see if it BSOD's, add more offset vcore untill it passes.

 

[Seanoog]

cpuz is one way, HWmonitor is another, there are others too, all read the same sensor. sometimes people see the VID in other apps and think its the actual voltage when it isnt, thats why i suggested cpuz.
anyway your cpu sounds better than mine if it needs only 1.300v for 4.6

just do the usual, increase multplier and test with p95 small ftt to see if it BSOD's, add more offset vcore untill it passes.

So how does VID vary from Vcore?
I am going to assume that VID is the voltage supplied to the memory controller and iGPU components in the cpu?

Thanks for the compliment, my CPU showed good potential when I first purchased it all those years ago.
As for over clocking it's all automatically done;
Forced the multiple to 46 and left the motherboard to sort the voltages, any higher 47-48 and the system crashes, temperature's rarely exceed 55C on constand load.
But it's a big case, big air cooler and lots of quiet fans.

I'll certainly look at over clocking manually, since I play CPU limited games, and I am not ready yet to purchase a new system with the new GPU.

 

[eddiew]

I believe the VID is the voltage the CPU has been internally labelled with, and it varies a little between units. If you set voltage on auto, the motherboard will read that to see what it should supply. Lower VID chips tend to make better overclockers, since they needed less voltage to run stock clocks. High VID units may end up binned as lower speed processors at the factory.

Unless I've totally mixed up my acronyms

 

[DarkBahamut]

The VID is the vcore the CPU is requesting, but it isn't necessarily supplied with the voltage it requests. LLC and using offset voltage adjustments change the voltage that is actually supplied by the motherboard. This is the important value as it's the supply voltage you want to know as that's whats actually delivered and this is what CPUz/HWMonitor report.

The only think to watch if you push more is you might need to enable PLL overvoltage from around ~4.8GHz and that tends to break sleep support if you use it.

 

[zipp0r]

^^
exactly. mine has a vid request of 1.42v @ 4.6ghz, i have it set manually via offset and llc so it will hit ~1.35v.

 

[eddiew]

Ah, well, I was half right. The more you know

 

[Sampsy]

Well I've been at the same OC for the most part since I bought the 2500k shortly after it was released - 4.8Ghz and ~1.408v at load. The overclocking guide says anything over 1.38v would limit lifespan, but it's still going strong after what, 5 years?