Typically how many milivolts per 100Mhz?

beany_bot · 28 Dec 2015 at 15:14

I have a stable OC on my 4690K [4.5 Ghz @ 1.22v] (was unstable at 1.21v)
Just wondering. My temps are really good (typically high 60's under normal gaming load, low 70's under stress testing). Wondering typically how many volts you would likely increase per 100Mhz? i.e. If I wanted to go to 4.6 or 4.7 what I should be looking to put the voltage at? I know I have to trial and error all this but just as a quick benchmark to timesave slightly. Bear in mind. 4.5Ghz @ 1.22v stable. at 1.21v was not stable.

new boy · 28 Dec 2015 at 15:25

*shrug

Maybe 0.04v - 0.06v per 100MHz as a very very rough indication.

As you already know though, what my chip does and what yours does are likely to be very different things

BluD · 28 Dec 2015 at 16:35

It goes up with each 100 MHz so maybe .04 for 4.6 then .06 for 4.7 then .08 for 4.8 etc.

beany_bot · 28 Dec 2015 at 16:51

I see. Diminishing returns then. I think I will try 1.25v at 4.6Ghz see if stable and temps OK. Thanks all.

russell664 · 28 Dec 2015 at 19:23

Sometimes you just hit a brick wall on the clock limit, regardless of the voltage. Like on my i7 4930k.

Disco_P · 29 Dec 2015 at 08:27

russell664 said:
Sometimes you just hit a brick wall on the clock limit, regardless of the voltage. Like on my i7 4930k.

My 2550K is the same. 4.8Ghz seems to be a no go. Even 4.7 requires a big jump in volts from 4.6Ghz.

Caracus2k · 29 Dec 2015 at 12:15

If you were to plot frequency vs voltage required for stable running for any given CPU on a graph you would see that the voltage required would increase exponentially till a point were no amount of voltage would run a CPU stably past a certain frequency. Given that your also practically limited by what voltages wont kill a CPU fairly quickly as well for most modern intel CPU's I would suggest an upper voltage 'ceiling' or 1.3-1.35 volts with the caveat that your cooling system needs to keep the CPU at around 80 degrees max given we are currently in the middle of an albeit mild winter.

NoNameNoNumber · 29 Dec 2015 at 12:25

my old 2500k

4.8GHz @ 1.344v
5.0GHz @ 1.488v
5.5GHz @ 1.656v

Voltage and heat shoots up!

D3K · 29 Dec 2015 at 12:28

My 4790k:

4.5 @ 1.225
4.6 @ 1.275
4.7 @ 1.325

Couldn't get 4.8 at 1.375 and wasn't prepared to find out what it needed.
So 50mv per 100mhz up to 4.7

NoNameNoNumber · 29 Dec 2015 at 12:30

It isn't linear though D3K, check my post above. 4.7 probably needed nearer 1.35v for you to be fully stable which makes 4.8 well beyond another 0.05v

Either that or 4.5 could've run fine with slightly less.

D3K · 29 Dec 2015 at 13:01

4.5 running with less would mean 4.6 also has to run with less, otherwise the jump to 4.7 actually goes down in required volts.

What is more likely, was that 4.5 wasn't completely stable at 1.225 and required more. Not out of the question as I didn't spend a long time at 4.5.