PDA

View Full Version : Need Help Overclocking



Nisal
11-25-2012, 03:27 AM
Hi Everyone,

I recently bought and Maximus V Gene and now i'm wondering whether or not to overclock it. I haven't done OCing before and i just need help or advice Ocing my RIG on stable 24/7. my current stock load temp. is 40-45 C and as I'm doing 3D Animation and rendering I'll be running at full load sometimes for 12 to 18 hours.

My RIG:

Intel i7 3770k
MB: Asus Maximus V Gene
PSU: Corsair CX430
HDD: Samsung HD322HJ
VGA: Intel HD 4000
RAM: Corsair Vengeance PC3 -10700H (1600Mhz) 8GB
Cooling: Corsair H80

I'm thinking of just a mild OC of about 4.2-4.4GHz for now till I get a Asus GTX 660Ti because sometimes I'll be playing high-end Games sure as Battlefiled 3.

I also want to know about the correct way to OC and also want to know whether it will affect on the warrenty.
Thanks In Advance.:)

Arne Saknussemm
11-25-2012, 08:59 AM
HI there Nisal! :)

For a mild OC you can just follow this guide: http://rog.asus.com/forum/showthread.php?5836-ASUS-Rampage-IV-Extreme-Easy-Overclock-Guide&country=&status=

It is written for another board but the BIOSes are very similar and the principle just the same

For a more complex guide you can read through this one in the Maximus section: http://rog.asus.com/forum/showthread.php?19186-ASUS-ROG-Maximus-V-Formula-UEFI-OC-Tuning-Guide&country=&status=

Again the board is different but the UEFI BIOS for both boards will be the same in most important aspects.

HiVizMan
11-25-2012, 10:26 AM
Read and try things in small steps. Do not just jump in an go for 5GHz your first try. Keen one eye on temps and make notes of your default voltages.

For a OC of 4.4 you would only need to change the multiplier to 44 and press F10. The rest can be on auto and it will OC to 4.4.

Check the temps and voltages applied and then go back into the BIOS and manually set them the same or slightly lower.

Run a short stress test to check temps and if your system falls over.

Repeat if successful with slightly lower voltage again. Keep repeating until your system does freeze or fall over. Then simply go back one or two steps voltage wise and you should be good to go.

Nisal
11-27-2012, 02:06 PM
Thanks alot for the help guys,

So here's what I'm going to start with. Please correct me if my steps are wrong.
For a start I'm going for 42x
1)go into UEFI extreme tweaker
2)Set "Ai Overclock Tuner" from Auto to Manual
3)By all cores - 42
4)Save and restart :)
5) For stress testing I'm going to use 'Cinebench'
Will be using 'real temp' and Ai suite II for temperature monitoring

Here's some questions I really what to ask before I do this

1) I heard that overclocking can also be done in Ai Suite II, if so is it safer than using UEFI?
2) In UFEI should I keep the "Memory frequency" at Auto?
3) What is CPU Speard spectrum - should I keep it on or off (it's currently off)?
4) What are the worst case scenarios I'll be facing when overclocking?
5) I heard that the Vcore - should be kept under 1.300 in ivy-bridge because of temperature issue, so what kind of temps. Should i be looking for when over clocking from stock to 4.2GHz :D

Thanks again in advance :)

HiVizMan
11-27-2012, 02:32 PM
Please do not install AIsuite unless you desperately want to. And certainly do not have two monitoring applications open at the same time. Use HWmonitor or Coretemp.

1 Do all you overclocking in the BIOS, stay away from AISuite
2 XMP profile in mode of overclocking then the rest as per your plan
3 Do not bother with it now.
4 Your system freezes and you have to reboot.
5 Do not believe all you hear. Let temp be your guide.

I doubt that you will need more than 1.26 for 4.2GHz but no two CPU's are the same so it is impossible to predict what your system will need.

TenBlade
11-27-2012, 03:14 PM
Recommended temperatures would be a matter of debate. Temp.s during stress testing is one thing, but every day use is interesting. I would (justified or not) be sweating bullets if my cpu crossed the 90 C mark on a stress test, temperatures up to 105 C are mentioned "out there" as being within limits, but every CPU is different, hard to say how it would affect degradation of CPU (and thermal paste for that matter).

In everyday use Id try to stay below 70 C (or am I too worried?!)

How it affects warranty I have no idea. My philosophy is to clock to the limits of my wallet (which at the moment is no clocking at all...).

Nisal
11-28-2012, 01:23 AM
Ok guys OC it's at 4.2GHz. Pretty stable upto now.:) But the temperatures are betweem 60- 75 when in full load

Changes made in Bios are:-

1)Set "Ai Overclock Tuner" X.M.P
2)Cores to 42
3)CPU Spread spectrum from Auto to Disable

didn't touch the Vcore values
persent vcore value at 100% is 1.248
Also run Battlefield 3 and got a improvement from Low to Medium graphics

Im Still concern about the temperatures. Is it safe to keep running at these values?
Is there anyway to lower it to around 55 C - 65 C?

Menthol
11-28-2012, 03:54 AM
IMHO that is good temps at full load. Yes you can lower, either lower clocks or better cooling is the simplest way, you can lower your vcore in very small amounts until it becomes unstable then raise to the lowest it was stable at.

TenBlade
11-28-2012, 05:13 AM
Wonder if the new Corsair H100i in push-pull will make a significant difference to the H80 youre using, 3-5 degr.?! Anyone swapped an H80 to the H100i yet? Otherwise custom water cooling would be it, but the expenses could easily buy you a spare CPU ;)

sectionate
11-28-2012, 10:18 AM
Nisal if your running your machine at full load for 12-14 hours your corsair will get heat soaked and temps will increase. If you use cpuids temp program,make sure you get the pro

HiVizMan
11-28-2012, 10:41 AM
Now that you have a base line to work from. We can start by lowering some settings.

First you can go into Digi settings and set your LLC to regular.

Then run some stability tests and or play your game, both are good. Check temps and you should notice an improvement.

Next would be to lower CPU PLL a little currently it is at default and that is 1.8 or 1.9 depending on your CPU. In small steps lower the voltage until your system is no longer able to pass the stress test or play your game. Then increase that fail voltage by two notches upwards.

Temps will improve.

There are other things but let us do these two first. LLC and then once stable CPU PLL.

Nisal
11-28-2012, 11:37 AM
Guys without running all the cores at 4.2 is it safe to run the 4 cores different clocks, let say keeping one core at 4.2Ghz (max) and others reducing one so to make the other cores have lets say 3.9, 4.0, 4.1. Is it safe, will there be a reduction in temp. will there be any side effects, I also tried lowering CPU Voltage to 1.200 from 1.250 but no major notable changes in temp, but the cinebench mark values got slightly increased

CPU :- from 8.24 to 8.34
OpenGL :- 21.27 to 21.30
MP Ratio :- 4.66 to 4.72

Sadly the Battlefield medium graphics FPS rate is 5-12 :( i guess it because of the onboard VGA

Arne Saknussemm
11-28-2012, 11:55 AM
Everything I have seen suggests running per core OCs actually increase voltage requirement and temps. But as always different systems are their own worlds...you would have to try it.

Cinebench scores vary quite a bit depending on background program activity so score changes at same frequency are common.

HiVizMan
11-28-2012, 11:56 AM
You can set each of the four cores at their own multiplier, in fact that is what Intel do for you.

You do know you can overclock your IGP in the bios, not sure if there is software to do this as I do not use IGP.

Nisal
11-28-2012, 12:27 PM
Ok correct it didnt quite went to 1.200v now its work. The temps which were between 65 to 75 are now mantaining under 70 with max of 72 occasionally.

Yeah i was wondering about the iGPU, has anyone overclocked the iGPU of the i7 3770K before? If so how are they doing afterward? :)

sectionate
11-28-2012, 12:50 PM
Nisal I couldn't think of a way to explain it, so I copy/paste from this site

http://www.overclockersclub.com/reviews/corsair_h60/5.htm

I have no problem making the statement that this water cooling unit offers a lot of "bang for the buck" as well as other features that can appeal to many. For all tests except overclocked load, it performed within a few degrees of the Noctua NH-D14, which is more expensive than the Corsair H60. It's very compact, requires little to no maintenance, is quiet, and performs well even for those who plan to overclock. However, it may not be for those looking to perform very extreme overclocks, as the performance and effectiveness of the cooler begins to degrade as the heat output of the processor increases, and one would be better off purchasing a high-end air cooler that can keep up with the extreme heat loads. This is caused by what is known as "heat soak", which is when the radiator and the rest of the components heat up, which significantly reduces the effectiveness of the unit. Perhaps having another fan on this unit in a push/pull setup would help this, but with a radiator on the level of size of this one, as with any radiators, it's only effective up to a maximum wattage input at a given ambient temperature. Once the rate of heat input to the system exceeds the maximum effective dissipation rate of the radiator, this leftover heat input goes into increasing the temperature of the system until it is at equilibrium. This simply means a higher overall coolant (water) temperature, meaning a higher processor temperature. Long story short, don't expect excellent temperatures on an i7 930 at 4.5GHz!

Overall, I like this unit. It is very compact, quiet, self-explanatory, and little or no maintenance is needed (i.e. cleaning dust out of heatsink fins, draining/refilling the unit, etc). I am also one of those folks who love to work on the presentation of the interior of the computer case, and staring at the top of a huge heatsink takes away from the look slightly. I love the cleanliness and sleekness of the Corsair H60 and the way it makes the inside of the HAF932 look. With all that said, for the price, it is well worth it and I hope Corsair keeps up the good work!

Nisal
11-28-2012, 11:09 PM
Ok new change. Last time I only changed the manual CPU voltage to 1.200 , this time the voltage was set to offset mode, and as HiVizMan said i change the LLC to regular, temps are now around 55-65 a real change from been at 65-70. Cinebench is holding no change, battlefield 3 FPS slightly increased by about 5frame. :D What's next?

Nisal
11-29-2012, 12:43 AM
I tried lowering the CPU PLL Voltage from 1.8 (Auto) to 1.5 no notable change. So guys what should I do? should I keep it at 1.5, lower it more or keep it as it is at Auto?

HiVizMan
11-29-2012, 09:52 AM
Find a balance, any lower and it becomes unstable. I would go for 1.65 CPU PLL and consider my work done.

Save that OC profile to your BIOS and take screenshots F12 on a Fat formatted USB stick in bios will do the job for you.

That way if and when you update bios you have a base line to work from. And you have a base line to tinker from. Tinkering and fine tuning is where all the fun is.

Nisal
11-29-2012, 12:09 PM
Thanks guys really appreciate all your help. Thanks :) I'm going to love this forum :D

Nisal
11-29-2012, 02:00 PM
Next up MEMORY FREQUENCIES

14376

as I'm going to run in full loads upto 12- 18 hour a day should I keep the frequency at 1600MHz or can I increase it?
If so what are the advantages and disadvantages I'll be facing?

I have 2x4GB of Corsair Vengeance 1600MHz RAMs

HiVizMan
11-29-2012, 07:04 PM
Keep it at 1600MHz. You do not want to have a problem 11 hours into your run. No real gain for huge risk

Nisal
11-30-2012, 11:45 AM
1st crash guys :( today after running for 12hour due to an animation rendering, the monitor usually goes into sleep mode, aftercoming back from it the system crashed to blue screen. i'm wondering whether it has something to do with Internal PLL Overvoltage at Auto?

HiVizMan
11-30-2012, 11:50 AM
What did the OS event viewer say?

And you would be better served to have it enabled in any case.

Nisal
11-30-2012, 12:07 PM
The event level 'critical' at the time of the crash is kernel-power with several warning before that about \Driver\AiCharger failing to load

Nisal
11-30-2012, 12:12 PM
Followed by a event a few seconds later saying that the computer has rebooted from a bugcheck error (0x00000101) memory dump

Nisal
11-30-2012, 12:25 PM
Major problem now the processr doesn't go to 4.2, it remain in 1.6Ghz :( please tell me it can be fixed

HiVizMan
11-30-2012, 12:49 PM
LOL that is fine.

Clear CMOS and start again, your CPU needs more voltage 101 error.

Nisal
11-30-2012, 01:06 PM
Ok :D well it's back now, but what changes should I make now to prevent it from happening again. I already put the PLL back to Auto

HiVizMan
11-30-2012, 01:28 PM
You might need to bump up Vcore one notch.

Nisal
11-30-2012, 01:47 PM
As it's in offset mode, I gave it an offset value of +0.050 hope that would be enough. LLC is still at regular by the way.

HiVizMan
11-30-2012, 02:18 PM
Outstanding time will now tell if that is enough. But you have the method perfect

Nisal
11-30-2012, 02:21 PM
Ok, then if it crashes the same route should be taken then. OK thanks,we'll see want will happen next :)

Nisal
12-15-2012, 12:46 AM
Ok guys 4.2 GHz was pretty stable so this will be my base clock speed to return to.

In the beginning of this post I mentioned that I'm thinking of an OC of 4.4GHz so now I started to head to that goal.

step 1) check the stability when it overclock to 4.3GHz
- It was also stable with no crashes
step 2) increase the multiplier up to 44x

It is stable but with a cost. to make it stable I had to increase the LLC to high.
4.2 and 4.3 worked on LLC been in medium

In 4.4 GHz if I lower LLC to regular or medium some programs for example I'm using Vue 10 will crash when rendering an animation
But with the LLC changed to High it works smoothly.

Because of the LLC been at High the load temp. are now at 65 to 70.

My question is:-
Is there any way of lowering the LLC to get the temps. below 65 and still let the app run smoothly
Vue is the only software that I have which make the processor hit this temperature, all the other softwares run on temps below 65

My current PLL settings in 1.65
Vcore at full load is 1.144 V
CPU voltage s on offset mode with an offset to +0.02

Guys Thanks in advance for any suggections and advise

P.S. 4.4GHz is the limit that I'm going on overclocking. I have no intention on going further.
:)

Arne Saknussemm
12-15-2012, 09:37 AM
LLC is just a way of allowing more juice to the CPU when it asks for it so basically you could set 1.2 vcore and run high LLC or 1.3 vcore and medium or 1.4 and regular (these numbers are pure made up just showing how it works kind of...;)) the LLC is simply making up the difference in what the CPU actually needs....

So, basically for higher clocks you will need more voltage. LLC on medium and a bump in your Vcore offset is probably a good way to go.

HiVizMan
12-15-2012, 04:12 PM
I would increase the amount of vcore (or increase the amount of offset if you go that route) and drop the LLC to medium.

Nisal
12-16-2012, 01:31 AM
Okay guys I increased to the Vcore and lowered the LLC to medium. no major changes in temperature.

The load temperatures are around 60- 70 on full load. Is it okay to run at this temperatures? Don't want to the cooling to get heat soaked.

After a long animation rendering the system tends to slow down. Any reason why this happen.
Is it because of the RAM? Their currently at 1600MHz X.M.P profile.

Arne Saknussemm
12-16-2012, 09:41 AM
What voltage do you have on load now?

But increasing voltage will give you higher temps though 60 to 70 is OK. But if you are running the CPU constantly at full whack I would think about getting some better cooling.

What voltage is XMP setting for VCCSA, VTT, PLL, VDIMM.....maybe you could shave a degree off setting the RAM up manually but the major thing there is Vcore

Do you have something directly cooling the VRM of the motherboard? If not I would set up a fan to do so....maybe the constant heat is warming up the VRM and if that happens the CPU can throttle....

Nisal
12-16-2012, 10:22 AM
The Vcore is aprox. 1.200 V at full load
VCCSA 0.95 V
VCCIO 1.05 V
CPU PLL 1.65 V
PCH 1.05 V
VTTDDR 0.75 V
2nd VCCIO 1.05V

Well I currently dont have any liquid cooling on the VRM

Nisal
12-22-2012, 12:44 AM
By the way guys will there be any major change in temp. if I change the stock cooling fans which came with the H80 and replace them with an AF120 for pull and SP120 for push?

Seithan
12-25-2012, 06:44 PM
There is good information in this thread.

Makes me wanna go back on my 2500k and fiddle with the LLC and PPL.

Im running 4k, a boost of +700mhz, but the problem is not so much on the aircooler, a TX3, but the box is pretty bad quality for OCing (a cheap box from Asus :P no offense ).
Temperatures hit 70c while gaming, but not over it and im thinking of playing with the LLC and PPL to reduce heat a bit more rather than push the clock more.

My vcore is at 1.13-1.14 and it runs pretty stable too.

Nisal
02-14-2013, 12:11 AM
Guys what's the optimum Vcore for 4.4Ghz, because I added a Asus GTX 660 Ti, and Intel SSD and changed my PSU to Antec 520 high current gaming.

After adding these it seems my animation render time has increased a bit but gaming is fine(awesome).
I'm running at off-set voltage and offsetted to +60
my current full load Vcore = 1.1600 V (shown by CPU-Z)
LLC - medium
temps at 100% - 60 - 70 celcius