Overclock i5 4670k to 4 Ghz @ 1.08 volts

Vragensteller

Honorable
Oct 28, 2013
19
0
10,510
I've just overclocked my Intel i5 4670k to 4 Ghz with 1.08 fixed core voltage (also measured as such in Aida64). I've been running Aida for an hour now and it doesn't seem to give an error, and average temperatures of the CPU cores are between 65-75 under 100% load according to RealTemp, with them usually being below 70, and below low 60's for Core 3 and 4. So far no throttle according to Aida.

My main question is if the 1.08 core voltage for the 40 multiplier is enough or if this is too much, but feedback about the above setup is more than welcome! (I'm new to overclocking, and building in general)

PS: I'm aiming for higher FPS and a more future proof set for future games. Especially because we're on the verge of new gaming consoles coming out, PC games will get more demanding very quickly. I'm trying to bridge that gap a bit for the next hardware generation or two.
 
I have a Corsair H100i with stock thermal paste. This is my first overclock ever, so I wanted to go for a "safe" 4.0 Ghz first. Is the voltage alright? Feedback would be more than welcome!
 


There is no end all "safe" voltage for ANY cpu. It is all dependent on the chip but most people with an h100i (not push/pull) can get between 4.4-4.6 (voltage dependent on the individual chip)
 
If you are truly stable @ 1.08v you're doing pretty good so far. You have a stellar chip. You should EASILY be able to go much farther. I just put in a new i5-4670K a couple days ago. Got mine stable @ 4.6GHz @ 1.25v

Tried to get 4.7GHz all the way up to 1.263v but the temps were getting higher than I wanted with my Antec Kuhler H20 620. And it wouldn't run Prime 95 for more than 30 minutes. So I dialed back to 4.6GHz and I've been playing CIV 5 and Batman Arkham City for a couple days with it.

Also, I think I read that 1.3v was the recommended max on this CPU from Intel. But I could be remembering that wrong.
 


I just put it on math and based my volts on the 4.0Ghz setting (1.08 volts), so now it's (1.08 / 40) * 42 = 1.134 volts for 4.2 Ghz. In this case (1.08 / 40) * 44 = 1.188 volts for 4.4 Ghz, which is very close to your 1.2 volts. Is that the right way of approaching it?
 
I'm running at 4.2 Ghz on 1.134 volts right now, and so far it seems stable with Aida64, with temperature cores 1 and 2 circling around 70-80 at full load with the program mentioned earlier.
 


Actually 1.2v was a rough guess
 


Alright, so 1.2 it is for 4.4Ghz. What would you recommend in the term of stress testing? Aida64 or something like Prime95 for around 6 hours?

PS: Last question, promise. Thanks for your time, I appreciate it!
 
Stress testing should be done with different programs. Prime95, Intelburntest, Aida64 are all good options. Then play some games as well. The thing is each of these software titles uses different things to load the CPU in different ways. If you can run Aida64 or Prime95 for 6 hours I'd say you're pretty solid.

Keep in mind that the higher voltage you go the hotter the CPU gets. So just watch max temperatures. Also, the faster you push the CPU the greater the demand for voltage. So it's not always a linear equation to determine what it will actually take to remain stable. Some chips can go to 4.8GHz. Some to 5GHz. Some only to 4.3GHz. Every chip is unique.

But, based on the voltages you're stable at so far this chip should be a high end one. :)

That said you have a more capable All In One cooler than I have so you should be able to push further than me. :)
 
BTW: Bit-Tech's review on the 4670K shows that their chip OC'd to 4.6GHz @ 1.26v. That's about where mine is (I think mine is 1.25? Something close to that.)
http://www.bit-tech.net/hardware/2013/06/12/intel-core-i5-4670k-haswell-cpu-review/2

Overclock3D.net hit 4.8GHz on theirs @ 1.276v
http://www.overclock3d.net/reviews/cpu_mainboard/intel_4670k_i5_haswell_review_overclocking/2

TechPowerUp only hit 4.2GHz @ 1.26v
http://www.techpowerup.com/reviews/Intel/Core_i5_4670K_and_i7_4770K_Comparison/3.html

Every chip is different I guess is the point.