Overclock.net › Forums › Intel › Intel CPUs › Complete Overclocking Guide: Sandy Bridge & Ivy Bridge | *ASRock Edition*
New Posts  All Forums:Forum Nav:

Complete Overclocking Guide: Sandy Bridge & Ivy Bridge | *ASRock Edition* - Page 95

Poll Results: Was this guide helpful?

Poll expired: Oct 17, 2012  
  • 80% (237)
    Yes (and I DO have an ASRock motherboard).
  • 15% (45)
    Yes (but I DO NOT have an ASRock motherboard).
  • 2% (7)
    No (and I DO have an ASRock motherboard).
  • 2% (7)
    No (but I DO NOT have an ASRock motherboard).
296 Total Votes  
post #941 of 9601
Quote:
Originally Posted by kennyparker1337 View Post


Have you tried doing any other CPU intensive task? Like gaming, renicoding, or even watching a video?
Make sure that the current limits are set to max (type in 3000 and press enter).

The limits were up all the way, but I did not try other tasks.

2nd Rig
(17 items)
 
  
CPUMotherboardGraphicsGraphics
[Intel] 2500k [4.5ghz] [1.3 volts] [AsRock] P67 Extreme 6 [Sapphire] 6950 Toxic [Unlocked to 6970 spec] Powercolor reference 6950 [Unlocked to 6970 spec] 
RAMRAMHard DriveHard Drive
[Samsung] Low voltage memory [@1866 mHz 9-10-9-28] [GSkill] Sniper 1866 9-10-9-28 1.5V [Corsair] 120gb Force GT WD 500GB 
Hard DriveCoolingOSMonitor
2x Seagate Crapstellation ES 2TB [Raid 1] [Coolermaster] V8 Windows 7 [Gateway] FDH2402 
MonitorKeyboardPowerCase
[Dell] S2316M [CM Storm] QuickFire TK [XFX] Core Edition PRO850W [In Win] Dragon Rider 
Mouse
Logitech G502 w/ stupid LEDs 
  hide details  
Reply
2nd Rig
(17 items)
 
  
CPUMotherboardGraphicsGraphics
[Intel] 2500k [4.5ghz] [1.3 volts] [AsRock] P67 Extreme 6 [Sapphire] 6950 Toxic [Unlocked to 6970 spec] Powercolor reference 6950 [Unlocked to 6970 spec] 
RAMRAMHard DriveHard Drive
[Samsung] Low voltage memory [@1866 mHz 9-10-9-28] [GSkill] Sniper 1866 9-10-9-28 1.5V [Corsair] 120gb Force GT WD 500GB 
Hard DriveCoolingOSMonitor
2x Seagate Crapstellation ES 2TB [Raid 1] [Coolermaster] V8 Windows 7 [Gateway] FDH2402 
MonitorKeyboardPowerCase
[Dell] S2316M [CM Storm] QuickFire TK [XFX] Core Edition PRO850W [In Win] Dragon Rider 
Mouse
Logitech G502 w/ stupid LEDs 
  hide details  
Reply
post #942 of 9601
Quote:
Originally Posted by kennyparker1337 View Post

As it stands with lowest positive setting with offset, you should get around 1.00v at idle and like 1.2v to 1.3v based on what multiplier you select.

That is pretty accurate, I would say. The problem is, that at 1.2v, RealTemp starts to give me the willies. Thus far I have only tried to OC to the lowest, stable speed while maintaining a sub 70C temperature. Sadly, if I were to go positive 0.05 on the offset and then the lowest on the turbo (which is 0.04 mad.gif) it would be way too much power for.... say.... ~4.1GHz. This is when I started to go lower on the offset.

If you suspect that to little can be a bad thing while at idle, do you perhaps think it would be best for me to disable C1E and EIST? If I do that, should I set the turbo voltage back to auto - since the option to disable turbo & turbo voltage is not something UEFI will allow?

Oh and here is another oddity I found out while using some recommended utilities:

Currently if I open up CPUZ or OCCT and leave my PC alone, the clock speed reported by these two applications is 1.6GHz. If I start up a game it then jumps up to 4.1GHz. If I open up Realtemp, however, it never stays at 1.6GHz. Not even on idle. The damn thing jumps to 3.3GHz and 3.6GHz. If I open up all three programs at the same time CPUZ and OCCT will report 1.6GHz while Realtemp still erratically jumps all over the place. Is this a known problem with Realtemp or is it that CPUZ and OCCT are just not reading my clock speed correctly?
post #943 of 9601
This thread helped me out a bunch. Thanks mate thumb.gif Now I'm rocking a solid 4.6ghz instead of 4.2ghz.
post #944 of 9601
Thread Starter 
Quote:
Originally Posted by rene mauricio View Post

Quote:
Originally Posted by kennyparker1337 View Post

As it stands with lowest positive setting with offset, you should get around 1.00v at idle and like 1.2v to 1.3v based on what multiplier you select.

That is pretty accurate, I would say. The problem is, that at 1.2v, RealTemp starts to give me the willies. Thus far I have only tried to OC to the lowest, stable speed while maintaining a sub 70C temperature. Sadly, if I were to go positive 0.05 on the offset and then the lowest on the turbo (which is 0.04 mad.gif) it would be way too much power for.... say.... ~4.1GHz. This is when I started to go lower on the offset.

If you suspect that to little can be a bad thing while at idle, do you perhaps think it would be best for me to disable C1E and EIST? If I do that, should I set the turbo voltage back to auto - since the option to disable turbo & turbo voltage is not something UEFI will allow?

Oh and here is another oddity I found out while using some recommended utilities:

Currently if I open up CPUZ or OCCT and leave my PC alone, the clock speed reported by these two applications is 1.6GHz. If I start up a game it then jumps up to 4.1GHz. If I open up Realtemp, however, it never stays at 1.6GHz. Not even on idle. The damn thing jumps to 3.3GHz and 3.6GHz. If I open up all three programs at the same time CPUZ and OCCT will report 1.6GHz while Realtemp still erratically jumps all over the place. Is this a known problem with Realtemp or is it that CPUZ and OCCT are just not reading my clock speed correctly?

Start by ditching OCCT and using prime95 x64 27.7. Select Blend, then Custom and change the Memory usage to 50% of installed memory (4096 if you have 8GB of RAM) and with stress testing programs, don't be afraid to hit 85C on Sandy and 90C on Ivy. It won't hurt your CPU. You will never see those temps except in stress testing programs (not even BF3 64 player game or reincoding a 50GB bluray).

You should go to offset +0.005v and then if you have too much power with turbo.. start making the turbo negative. That way you leave the idle clock voltage alone and only make the running state lower in voltage.

I believe Realtemp has a much higher refresh rate and tries to report every clock change, whereas CPU-Z only reports like 3 states... idle and max (running).

Quote:
Originally Posted by nvidiaftw12 View Post

Quote:
Originally Posted by kennyparker1337 View Post

Have you tried doing any other CPU intensive task? Like gaming, renicoding, or even watching a video?

Make sure that the current limits are set to max (type in 3000 and press enter).
The limits were up all the way, but I did not try other tasks.

That's probably it then. Unless you do something that actually requires 25%+ CPU usage then it won't jump to max state, because the idle state isn't even being fully utilized yet. Even having 20 tabs open in firefox may not trigger enough usage to require a max state.
Edited by kennyparker1337 - 7/10/12 at 1:20am
post #945 of 9601
Quote:
Originally Posted by kennyparker1337 View Post

You should go to offset +0.005v and then if you have too much power with turbo.. start making the turbo negative. That way you leave the idle clock voltage alone and only make the running state lower in voltage.

I think that is where our boards differ. If I leave the multi on auto, I am able to disable Turbo and doing so also removes the Turbo Voltage. If I disable Turbo (and again; Turbo Voltage subsequently) but then manually enter a multiplier, Turbo re-enables and is unable to be selected within UEFI to disable it. The Turbo Voltage however can be changed and it starts with the minimum value of "auto" and the next lowest is "0.04v". I am not 100% sure, but I did not see a way to get negative turbo integers. I will re-check again here in a minute. smile.gif

Also worth mentioning is that the LLC feature describe in your guide makes mention of numerous options - of which the Pro 4-M does not have. That I can see, it only has 3 options; 0%, 50% and 100%. Your recommendation of 50%, however, still applies and is what I am using at the moment.
Quote:
Originally Posted by kennyparker1337 View Post

I believe Realtemp has a much higher refresh rate and tries to report every clock change, whereas CPU-Z only reports like 3 states... idle and max (running).

I feel you may be correct on that once again. thumb.gif

As far as testing goes, what I have been using is LinX with the latest Intel library. It seems to stress my system harder than OCCT's LinX with AVC enable as well as Prime95. Perhaps you may want to include it in your package as it really is a great little tool.

http://software.intel.com/en-us/articles/intel-math-kernel-library-linpack-download/
Edited by rene mauricio - 7/10/12 at 2:13am
post #946 of 9601
Hello, my 2600k is running at 4.4ghz that reads [when i observe it] in CPUID and CPU-Z 1.28v core voltage but the max records 1.31-1.32. Does that sound normal or safe? Thanks
post #947 of 9601
Anyone know what can cause an FFT to skip like this (thread#2 finished 192K after just one test)? Is it instability or just a bug / glitch ?

493
post #948 of 9601
Quote:
Originally Posted by zerocraft View Post

Anyone know what can cause an FFT to skip like this (thread#2 finished 192K after just one test)? Is it instability or just a bug / glitch ?
493

I've had that happen to me before when I got a 15 hour stable run. I don't think it should be anything you should worry about.
post #949 of 9601
The Control Panel - Power Options - Minimum processor state controls whether your CPU idles down or not. Set that to a low number like 5% and RealTemp might agree with other monitoring utilities. RealTemp tells you exactly what multipliers your CPU is using. Some other apps like to fudge things a little when the CPU is lightly loaded.

C3 and C6 can also play a part in the real multiplier your CPU uses when lightly loaded.

RealTemp 3.70
http://www.techpowerup.com/downloads/2089/Real_Temp_3.70.html
Edited by unclewebb - 7/11/12 at 4:02pm
post #950 of 9601
The "Getting closer..." and "Finding the sweet spot" steps somewhat confuse me. In "Getting closer..." it seems like you're saying to increase the turbo voltage whenever we run into instability and keep increasing the multiplier at the same time until the vcore starts going over the max (which I believe is 1.55v, as stated in the voltage limits section?).

In the "Finding the sweet spot" step, you say one of the options, should instability occur, is to increase the turbo voltage. But didn't we already turn up the turbo voltage as high as it could safely go in the previous step? Or am I missing something important?

This is my first time overclocking anything, and I've gotten my 3570k to 4.4ghz stable so far with just one increase in the turbo voltage. I plan to do more once I get a new graphics card, but it's kind of pointless right now as I'm being bottle necked by a 9800GTX+
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Intel CPUs
Overclock.net › Forums › Intel › Intel CPUs › Complete Overclocking Guide: Sandy Bridge & Ivy Bridge | *ASRock Edition*