Should/Can I Overclock?

Different CPU instruction sets for performing fast maths functions. Hardly any real world programs use FMA and only a few things use avx and none of them as intensely as prime. When you get to the limit of prime oc you can switch to asus real bench to find your max oc for real world uses.

I like to start off with prime so I can be sure I have a 100% stable base line to fall back on

Giz
 
I see so I'm using avx prime as a 100% limit.. which if we get under 90 and it passes all working u can use a more realistic program to test everyday gaming use.
 
Brilliant cheers for all the information so far. Only attempted overclocking yesterday but kind of understanding it a bit more now.

Running it on 4.3 just now and will let u know how it goes.. if it goes too high or fails would I reduce multiplier or vcore?

Update... it froze
 
Last edited:
Depends on temps - if you have headroom to increase vcore then do so

It may not seem like much but just going up by 0.005 may be all you need. if it fails very quickly though then you know you are miles out and go up by 0.02

response to update - are you sure your uncore is staying at 36
 
I've upped vcore to 1.185.. still running so far..

I'm sure uncore is still 3.6.

As far as vrin goes does that need to be higher than vcore? So if my vcore goes to 1.19 I'm assuming 1.9 is still OK?
 
yeah 1.9 should be fine right up to 4.5ghz at least

Here are my settings for 4.5ghz

input 1.9
vcore 45 @ 1.285
cache (uncore) 42 @ 1.180
I also have an increase to system agent voltage of 0.020 as I have 2400mhz memory

Giz
 
Last edited:
4.3GHz @ 1.185v (36 uncore, vrin 1.9) = PASS max temperature 82.
4.5GHz @ 1.190v (36 uncore, vrin 1.9) = FAILED blue screen before windows.
4.5GHz @ 1.20v (36 uncore, vrin 1.9) = FAILED blue screen before windows
4.5GHz @ 1.285 (36 uncore, vrin 1.9) = FAILED temperature reached 100 so I stopped it but it didn't crash
 
Here are some updated results and am I right and assuming this:

4.5GHz @ 1.190v - FAILED blue screen before loading windows
4.5GHz @ 1.200v - FAILED blue screen before loading windows
4.5GHz @ 1.265v - PASSED but temperatures reaching 95 degree so I force stopped it.
4.5GHz @ 1.285v - PASSED but temperatures reaching 100 degree so I force stopped it.

Based on these I cannot get to 4.5GHz without better cooling?

4.4GHz @ 1.185 - FAILED blue screen during early Prime95 testing
4.4GHz @ 1.190 - FAILED blue screen during early Prime95 testing
4.4GHz @ 1.195 - FAILED workers stopped very early into testing
4.4GHz @ 1.200 - FAILED workers stopped very early into testing
4.4GHz @ 1.225 - PASSED temperatures hitting max 92 degree (jumping between 80-89)

Based on this does it mean I can safely run 4.4GHz at 1.225?? I will leave this test running for a couple of hours to see if it rises in temperature or freezes. I could even find a middle ground between the 1.200 - 1.225.
 
Yep 4.4ghz looks good just find the right voltage start from 1.21v

Your max temp is fine below 95... It wouldn't of failed when it got to 100 because the CPU would have been throttling

When you think you have the right voltage leave it testing for longer

What settings are you using in prime?
 
I tested 4.4GHz @ 1.220, 1.215 and 1.210 but it froze so I think the 1.225 is my sweet spot for it. I have it running that again and the maximum it has hit so far is 93 degree. It jumps around the 70s but that was the maximum it had hit.

I will leave it on Prime95 while im at football (3-4 hours) and hopefully I come back to it not crashed or temperature risen.

As for setting in Prime I haven't really changed anything other than the edit in the local.txt log. I run Blend I think it was, the one that says something about using RAM.


Here is a screenshot of it currently at 20mins in:

image.png
 
Last edited:
All looks good to me, hope it doesn't crash!

You can now either move to real bench to increase your oc or if you're happy increase the uncore as much as you can
 
It hit 95 degree so ill leave this test running a bit longer...

I wouldn't be able to hit 4.5GHz though as the earlier post I shown I only got it running at 1.265v but it hit 95 degree very early on, and any increase in volts made it go up to 100... surely that isn't safe? If the 95 one was ok then I would need to find the middle ground between 1.200 - 1.265 for it to work into windows without crashing.
 
Current one 4.4GHz @ 1.225 crashed at 45 mins :(

Will leave 4.3GHz @ 1.185 running while I am away and see if that works then I have something to go off from..

But considering my 4.4 stats where going from 1.225 to 1.200 showed workers failing so I don't have much room to play about with in that, looks like I will need to settle for 4.3 (if im lucky).
 
Last edited:
Real bench runs much cooler that is why you'll be able to get 4.5. It also uses less voltage to get there than prime would

But yes get 4.3 nice and stable and then you have a good print stable base to work from
 
Am I not taking a big risk using settings that fail in prime though.. I know it kicks the arse out it but surely thats the reason we run it through prime.
 
No not at all

My 4.5 GHz is real bench stable and I never ever have crashes. I also use my PC to run excel sheets with array formulas and I have no issues
 
By testing 4.4 in prime you now have an idea what voltage you'd need to be stable ( probably about 1.23) I can guarantee that will be real bench stable and way under 95 degrees. As you can see from my settings I can do 4.5 at that voltage ( although all CPUs are different)
 
I came back to my computer frozen on 4.3GHz @ 1.185v. The maximum the temperature got was 86 degree. So couldnt have been an overheating issue. Earlier i got a blue screen by going to 1.190v.

How should i progress from here?
 
You must log in or register to reply here.
Back
Top Bottom