How much can I safely overclock my i5-2500k?

[drhead]

I'm wondering if I can safely overclock my i5-2500k without a dramatic effect on the CPU's lifespan.
My setup is:
MSI P67A-GD65
Intel i5-2500k
EVGA 560 Ti 448 core FTW (factory overclocked for those unfamiliar with EVGA)
4x4GB Corsair Vengeance 1600MHz DDR3
My cooling setup involves a Cooler Master Hyper 212 with IC Diamond thermal paste. I am using a Thermaltake Level 10 GT case, which is big and has plenty of airflow.

I've collected some temperature data while running Prime95 and FurMark for the GPU (to see how much of an effect the gpu has on the rest of the computer when it comes to cooling ability). Here's what I've found:

Prime95+FurMark
57*C - avg. CPU core temp
80*C - maximum GPU temp observed after 15 minutes (It throttles from 800mhz to 400mhz immediately once it hits 80)

Prime95 after 15 mins of FurMark
51*C - avg. CPU core temp

Note that Turbo Boost had the chip running at 3.4GHz the entire time. I also think I remember running Prime95 a long time ago and getting around 70*C. Is this normal or am I probably remembering things wrong?

I recognize that every CPU is different and there is no definite answer. I'm really looking for temperature guidelines and an estimate for how much I might be able to overclock. I also plan to add a second GPU in an SLI setup later this year, and I'd like to know if that would complicate things (I have a 750W power supply, so I think it might be cutting it close to add 2 GPUs)

Thanks for any advice you can offer.

 

[jemm]

I5 - 2500K Overclocking Guide http://www.tomshardware.com/forum/265056-29-2600k-2500k-overclocking-guide

 

[Thompaam]

For Sandy Bridge CPU's, I'd say that the thermal roof hits about 75C. Just see how far you can push it, when either crashing or temperatures start messing, stop.

 

[jemm]

You won´t be able to push so hard with a Cooler Master Hyper 212 -- if you really want to get into serious overclock, then get a better CPU cooler.

 

[DJDeCiBeL]

4.5 is pretty much a lead pipe cinch guarantee with a 212+/Evo on a 2500K, unless you have a very unlucky chip.

If you have a better than average chip, 4.7-4.8 isn't out of the question. I can do 4.7 on my chip and still keep the temps under 72C (room temp 73F) with my 212 Evo, but I don't like the voltage, so I stick with 4.5, since I don't really need it to be any faster anyway.

 

[jemm]

The OP didn´t say how much it will be pushed.

 

[DJDeCiBeL]

The OP didn´t say how much it will be pushed.

Basically in the title, and "safely" is pretty much 4.5 on a 2500K with a 212+/Evo.

Of course, ultimate safety is to just not OC at all, but 4.5 is safe enough.

 

[amuffin]

Safely?

Well, that depends on your definition of safe.

 

[DJDeCiBeL]

Safely?

Well, that depends on your definition of safe.

True :lol:

Seriously though, with the voltage involved with 4.5 on a 2500K, you're not really shortening the lifespan to the point that it'll even matter.

 

[jemm]

Of course, ultimate safety is to just not OC at all

Ultimate even if OC is not involved it is not safety! 😉

 

[redeemer]

You need good cooling 4.8 Ghz will need 1.38-1.4v, 5.0 Ghz needs over 1.4 mostly around 1.45, bad chips need sometimes 1.5+ to hit 5.0Ghz. Definitely enable PLL for anything over 4.8Ghz. H100 pushpull for full out WC loop, Sandybridge can take heat but it will degrade faster. Under prime testing or IBT you shouldnt hit more than 80 degrees. As long as you keep it 75c on load and 1.4v or under your chip will last. 1.52v is the max !