Should I Stick or Twist

Rowan88

Rowan88

Rowan88

Associate
Joined
2 Jun 2013
Posts
1,286
Location
Glasgow
Hi,

I managed to run a overclock of 4.1GHz @ 1.280v for a full OCCT Large 6 hour test. These are the results:

41ghz6hr.jpg


However when I try for 4.2GHz @ 1.280v it got about 1 hour in then said one of my cores went over the maximum value.

Should I keep 4.1GHz and try reducing the volts to get a sweet spot or should I go back onto 4.2GHz and try lowering the volt to get a safer temperature? (I just cleaned out and re-applyed thermal paste to CPU so it could be it needs time to break in first)

Thanks
 
I would keep playing .

TBH 4.1 at 1.28 is not that great :(

I get 4.6/4.7Ghz at 1.25

wxpjU1p.png


What cooling are you using, as those temps are pretty high ?
 
BeQuiet SR1 Pro...

I decided to start at 1.280 as I read the max safe for my i5 chip was 1.300.

I ran 4.4GHz but it froze within minutes, I tried 4.3GHz but after 1hr5min I got one of the cores has reached max temperature in OCCT.

So I knocked it down to 4.2GHz which I thought was safe but I got a core temperature error after 2 hours..

So I tested 4.0GHz overnight and it passed a full 8 hour test, then I upped to 4.1GHz while I was at work and it passed 6 hours of that..
 
Last edited:
4.1Ghz at 1.28v is terrible to be honest. Normally you would be looking at hitting around 4.4Ghz with that much vcore. Your temps are poor as well for such a low overclock which is probably down to the vcore. Have you changed any other voltages such as cpu input voltage (probably called something else on Gigabyte boards)? My old 4670k needed a vcore of 1.275v and cpu input voltage of 1.90v for 4.4Ghz (I didn't try anything lower than that as that was the first overclock on it that I tried).
 
No I don't think I changed anything else, like I say I just changed the CPU Core Volt to 1.280 and then started tweaking the multiplier.. The only other thing I had enabled was XMP on my memory for 1600MHz.

Could it be OCCT that is being pushy? I used to use RealBench and got a lot more happier readings, but because it was failing OCCT I took that as it wasn't stable.

I dont know if this is any good information, but at the very start on stock settings I ran Cinebench and Valley Benchmark and took note of some of the results in HWMonitor. On stock at 3.4GHz my core was showing 1.116v. So I took it from that I would need obviously a bit more to be able to overclock up to 4.2 etc.
 
Last edited:
Rowan88 said:
Ok I will give that a go, but if it passes RealBench and fails OCCT, doesn't that mean its not stable?
Click to expand...

RealBench will give you a decent stability test more focussed to real life usage - I would use that as a good indication of stability

Run the Stress Test for a few 15minute blasts, if it's OK play a few games.
 
Ran 44 @ 1.26v for 15min (up to 4gb) stress test and maximum package temperature was 79 degree.

44126RB.jpg


So what do I do from here raise the multiplier till a 15min test fails or do I have to run longer tests on this current one to see if its stable?
 
Last edited:
You're probably at the limit of your cooling, but try 4.5 - do not change the vcore keep at 1.26

Run the stress test on loop for an hour, keep an eye on temps, see what happens.
 
Ran 45 @ 1.260 and it the screen froze around 3mins in at 76 degree. It then blue screened with error CLOCK_WATCHDOG_TIMEOUT.

I ran 44 @ 1.260 for the full hour and these were the results:

44126RB1hour.jpg


Most of the time it sat around the low-mid 70s but most it ever hit was 80.

So can this be confirmed as a stable overclock? Or does it require more testing?
 
Last edited:
Ok, we'll call 4.4 as the max with your current cooling. A hour of RealBench will be a good indication of stability. Personally that is all i would do - don't see the point of massive 8hour tests ( though i'm sure someone will come along and disagree ! )

Try running a few games aswell.

You can now fine tune it and try and reduce temps a little.

Drop down to 1.25 and try the same test as above ( keeping 4.4ghz )
 
Thanks,

Currently tested:
(I have my Frame Limited in AMD Catalyst to 60 FPS)

DiRT Rally - Ultra Settings 1080p Benchmark, was hitting a very nice stable 60 FPS
League of Legends - Ultra Settings 1080p, Practise Tool was hitting 300+FPS then went into a team fight and spammed skills, lowest it dropped to was 150FPS.

Currently installing Crysis 3, Batman Arkham Knight and GTAV so will test those out too.


If all goes well then as of tomorrow I will drop the volt down to 1.25 and re-run tests in RealBench to see if I can get the temperature down any further. Am I best running 15 min or full 1 hour for these?
 
Rowan88 said:
If all goes well then as of tomorrow I will drop the volt down to 1.25 and re-run tests in RealBench to see if I can get the temperature down any further. Am I best running 15 min or full 1 hour for these?
Click to expand...

Just give it a 15 minute blast to see if it is stable
 
Thanks for your help m8..

Will tweak core tomorrow.. I assume once thats found the minimum core needed to be stable I then just start raising the uncore up as high as I can get it?

Batman Arkham Low Settings benchmark passed on 36min, 82max and 51avg FPS..
 
Rowan88 said:
Thanks for your help m8..

Will tweak core tomorrow.. I assume once thats found the minimum core needed to be stable I then just start raising the uncore up as high as I can get it?
Click to expand...

Yep, though i wouldn't worry too much about Uncore, it has very little impact on performance
 
You must log in or register to reply here.
Back
Top Bottom