Do I have my 4670k safe?

Rowan88 · 21 May 2017 at 19:24

Hi,

After a lot of help on this forum with overclocking this is what I have ended up with for my i5-4670K 3.4GHz chip:

Overclocked to 4.3GHz (1.200v) UNCORE @ 4.1 (1.190v)

These are my test results:

IntelBurnTest: Standard (88 MAX and 111G Flops) & Maximum (94 MAX and 121G Flops)

CineBench: CPU (662 score) & OpenGL (126.61FPS)

MaxxMem2: PASSED!

My last two steps where running OCCT and Prime95, possibly leaving them overnight to run 4+hour tests.

However when I load up OCCT and change packet to Small Data Set, when I go to start the test it runs for around 2 minutes then after the heart attack! noise of the duck quack (lol) I get "Core #0 over maximum value! Value reached: 88, Max Value: 85".

Does this mean my overclock is not stable? or should I test in P95 first before lowering settings.

Brizzles · 21 May 2017 at 23:02

Those temps are killing your chip, stop now.

What cooling are you using ?

4.3 @ 1.2v is a moderate overclock and you shouldn't be hitting 90 degree +

Here's mine as a comparison ;

4.6 @ 1.25 ( watercooled )

Use RealBench for a more realistic stabilty test

pastymuncher · 22 May 2017 at 09:25

Please tell me you are not using a stock cooler? If you are remove the overclock until you replace it. The stock cooler is not meant for overclocking. Those temps are ridiculously high for a moderate overclock with fairly low voltage. If you are not using the stock cooler then something is seriously wrong with your cooler or the way it is mounted because even your idle temps are not very good. What cooler are you using? What case do you have, how many fans are in it and how are they set up?

mattyfez · 22 May 2017 at 09:54

Yeh those temps are too high by anyone's standards.
As above you need to back down and adress the temperature issues.

VKleita · 22 May 2017 at 13:03

Max temp is 85 for that Chip, the fact you go well over it makes my mechanical sympathy go into overdrive.

Stop and look into it

Rowan88 · 22 May 2017 at 17:22

I have a Be Quiet Shadow Rock Pro SR1 Cpu cooler installed and have around 3 case fans. The computer is up on a flat desk with plenty of airflow..

I have my memory in bios set to xmp profile 1 which boosted it to 1600mhz.. that wouldn't be an issue would it?

I will reduce to 4.0GHz and leave voltage as Auto and Uncore/VRIN as Auto and run realbench see how it goes with temperatures.

mattyfez · 22 May 2017 at 17:39

Does your cpu voltage fluctuate? What's the max voltage in HWmonitor?
I'd look at that and reseating the cooler with new paste.

Rowan88 · 22 May 2017 at 17:48

I changed to 4.0GHz @ AUTO and Uncore to 3.4GHz @ AUTO. I ran a 15 minute test on Real Bench, here are the results:

What I need to know though is:

1) I noticed during the test in HWMonitor that the VRIN voltage was jumping between 1.116 and 1.254, is that normal?
2) When I tried to run RealBench on 8GB ram (which I have) it would just come up Realbench.exe has stopped working, but it worked no problem on 4GB.

Does that result look safe and should I now upgrade to 4.1Ghz or should I be tweaking stuff??

mattyfez · 22 May 2017 at 18:19

Whilst the temps are now ok, it's still high for that speed.

What speed is your cpu fan reaching? I can't see from your screen shot.
Is it plugged into the CPU fan header?

mattyfez · 22 May 2017 at 18:33

I'd do 3 things before anything else..
1 Make sure cpu fan is actually working and attached to the CPU fan header
2 Figure out what speed /rpm it's hitting under load
3 take the cooler off and refit with new thermal paste

pastymuncher · 22 May 2017 at 18:57

Leaving the vcore on auto looks to be overvolting to me. A peak of 1.254v is a lot for only a 4Ghz overclock. Mine will do that on stock volts and only needs 1.275v for a 4.4Ghz overclock on the core and cache.

Rowan88 · 22 May 2017 at 19:46

mattyfez said:
I'd do 3 things before anything else..
1 Make sure cpu fan is actually working and attached to the CPU fan header
2 Figure out what speed /rpm it's hitting under load
3 take the cooler off and refit with new thermal paste

1. The fan is connected via CPU FAN on motherboard, fan is constantly spinning so it is working fine.
2. Idle fan speed is 575 RPM and during the RealBench 15min test at 100% (4 cores) it shows maximum of 1530 RPM.
3. I re-done the thermal paste about 2 weeks ago but I suppose I could take it apart and give it another go.

pastymuncher said:
Leaving the vcore on auto looks to be overvolting to me. A peak of 1.254v is a lot for only a 4Ghz overclock. Mine will do that on stock volts and only needs 1.275v for a 4.4Ghz overclock on the core and cache.

So you think I should manually set the Core voltage? and if so to what 1.200v? And then try moving up to 4.1/4.2GHz?

mattyfez · 22 May 2017 at 19:51

Yeh set vcore manually would be the next thing.

And ditch the 'burn' tests, just use realbench for now.

Rowan88 · 22 May 2017 at 19:53

Currently running a test at:

CPU CLOCK: 4.1GHz
VCORE: 1.200v
UNCORE: 3.4GHz
CPURING: AUTO

Will update this post with results when complete. I'm just running standard RealBench 15min/4GB RAM tests...

Any idea why RB crashes when I try to use the full 8GB?

UPDATE: Test completed with max temp of 69 and max RPM of 1573

Rowan88 · 22 May 2017 at 20:41

Just ran 4.2Ghz and got same outcome (Max 69 degree and RPM 1577)..

I will keep these settings and run a full 8 hour test overnight and if I wake up in morning then I can assume this overclock is stable and safe yes?

And from tomorrow I can try 4.3.

mattyfez · 22 May 2017 at 22:02

Yeh sounds reasonable, just be wary, you don't want to be hitting 80c on the CPU at all if you can help it.

But also bear in mind that you won't be thrashing your cpu with 100% loads with normal usage for any length of time, but that depends on your usage.

Maybe target a max of 70c to 75c under load.

Rowan88 · 23 May 2017 at 10:10

Ran full 8hr test last night and it passed.

So should I now try 4.3.. and if temps hit over 80 stop. Then go back to 4.2 and reduce voltage to get minimum required. Then after that start working on uncore?

Brizzles · 23 May 2017 at 13:45

leave voltage as it is , just raise multi to 4.3 , check temps.

if temps are ok keep raising the multi, without changing the voltage, see how high you can get whilst remaining stable and within acceptable temps.

A couple of passes on Realbench will give you a good guide to stability - no need to do another 8 hour test at this stage

Rowan88 · 23 May 2017 at 14:07

Brizzles said:
leave voltage as it is , just raise multi to 4.3 , check temps.

if temps are ok keep raising the multi, without changing the voltage, see how high you can get whilst remaining stable and within acceptable temps.

A couple of passes on Realbench will give you a good guide to stability - no need to do another 8 hour test at this stage

Cheers.. if I raise multi and temps are too high what is next step? Reducing core volt or raising uncore?

Brizzles · 23 May 2017 at 14:22

Rowan88 said:
Cheers.. if I raise multi and temps are too high what is next step? Reducing core volt or raising uncore?

If temps start to climb too much, then settle at 4.2 ghz and gradually reduce voltage until it becomes unstable - Find the balance between 4.2 and the minimum voltage required.

Leave Uncore for now - ( it has very little effect anyway )