Official ASUS P8P67 Series Overclocking Guide and Information

Ceebs

n00b
Joined
Nov 14, 2011
Messages
4
As in the "OC Tuner" option? I tried it once, the system wouldn't boot =\ so went straight back to stock.
 

Ceebs

n00b
Joined
Nov 14, 2011
Messages
4
As in the "OC Tuner" option? I tried it once, the system wouldn't boot =\ so went straight back to stock.

*Sorry double post*
 
Last edited:

speedlever

Limp Gawd
Joined
Sep 29, 2006
Messages
447
On the EZ set up page, it's the setting on the far right of the system performance settings:

HQFAc.png
 

Ceebs

n00b
Joined
Nov 14, 2011
Messages
4
Ah k, I'll give it a go I guess. However, I would still like to overclock properly and reach that goal of 4.5.

I am still very curious about the voltage fluctuation I was getting if any one can shed some light on that.
 

Ex0duS5150

n00b
Joined
May 4, 2011
Messages
9
Ceebs, Off set voltage set to + x.xx and load line calibration (LLC) settigs are key to stable voltages under load.. The higher the LLC the less v-droop :) I had to change LLC from medium to ultra high then lower my + voltage off set so mine quit fluxing so much..

Just back to give a lil update, still @ 4.7Ghz 10 Hrs stable but i have change a few things as i became unstable after i dont know how may months.. You can see here i was stable 6 moths ago: http://hardforum.com/showthread.php?p=1037219904&posted=1#post1037219904

Recently i decided to mess with memory settings to see if i could get a lil more outta her and found my base overclock became unstable some where along the line..

After A BIOS update to 2001 and getting back to basics as in the settings suggested by Juan, i found a stable 4.7Ghz 10+ hour, 30 LinX run stable overclock again.. Of course also testing with 3Dmark 11 and vantage as well as more than a few hours gaming..

My current settings:

AI Overclock Tuner: Manual
BLCK Freq: 100.00
Turbo Ratio: Enabled (All Cores)
By All Cores: 47
Internal PLL OverVoltage: Enabled
Mem Freq: 1866
EPU Power Saving: Disabled

Load-Line Calibration: Ultra High
VRM Freq: Manual Set to 350
Phase Control: Extreme
Duty Contorl: Extreme
CPU Current Capability: 130%
CPU OverVoltage: Offset Mode
CPU Offset: + 0.020
DRAM Voltage: 1.55
VCCSA Voltage: .97
VCCIO Voltage: 1.15
CPU PLL Voltage: Auto
PCH Voltage: Auto
CPU Spread Spectrum: Disabled
CPU Ratio: Auto
Intel Adaptive Thermal Monitor: Enabled
Intel Virtualization: Disabled
Intel SpeedStep: Enabled
Turbo Mode: Enabled
CPU C1E, C3, C6: Enabled

This gives me 1.368 to 1.383volts on CPU full load.. No other setting seemed to be stable for me and believe you me, i tried every voltage variable that made scene imaginable.. Not sure if this is due to degradation or what but this is whats working for me now..

The 2x4 gig G-SkillRipJaws were @ 9-10-9-27 and now i have them fully stable at 9-9-9-24-t1 timings.. .1.65 volts and looser timings to achieve 2133 were not worth it..
 
Last edited:

speedlever

Limp Gawd
Joined
Sep 29, 2006
Messages
447
I'm not real crazy about running a 47x OC on my P8Z68-V Pro/2500k with CM Hyper 212+ and dual fans (push-pull). Temps are too high to suit me. I think I'll leave it at a mild OC at 43x.

This is after running LinX a few cycles. Note the max temp differences between Real Temp and CoreTemp with HW monitor in between. Note especially core 3 max temps. 18*C temp difference between RT and CT? How can that be?

3ruBj.png
 

tyreman

Weaksauce
Joined
Mar 31, 2009
Messages
71
One item I did do which helped quite a bit since I have the P180B Antec case.

I moved the hard drives to the lowest chamber ahead of the power supply in the pull out mount chamber which is actually for hard drives mounted vertically

There is a tricool fan in the lower chamber ahead of the PS on medium to aid cooling ,.it is blowing in from the lower grille and filter at the front

I then have a tricool fan in the middle chamber at the front. it is in the case of course behind the filter and door
Additionally fastened 2nd tricool fan on the other end where the hd's normally slide out on their trays horizontal mount, trays removed them now,.the tricool is held on with fan clips for that purpose so its a push pull config pulling air in both tricools on high

The top tricool fan is blowing out on high with the the top case removable angle deflector shield off, its far to restrictive.

The rear tricool fan is blowing out on medium.
The case PCI slot covers are all solid as well

Additionally the front grilles have been reworked neatly to reduce their high static pressure.
by removing alternate grille strips

My cpu temps are now 4-8 degrees lower at the least depending on room temps under load
The Coolermaster Hyper 212+ is blowing push pull with 2 pwm Coolermaster OEM fans for the Hyper up to the top case outlets tricool fan



2U5O7216jpg1.jpg


2U5O7213jpg1.jpg


2U5O7214jpg1.jpg


2U5O7218jpg1.jpg
 
Last edited:

speedlever

Limp Gawd
Joined
Sep 29, 2006
Messages
447
I'm tempted to liquid cool but I think I'd have to replace my Antec Solo case in order to do so. I have the single Tri-Cool exhaust fan in back running on low and two 92mm fans in front of the HDs pulling in cool air at the front plus the dual OEM fans on the Hyper 212+. Short of improving my TIM and/or application method, that's about as good as I'm gonna get and keep the fan noise acceptable.

I just built another Z68/2400 in a 400R Carbide case and it runs much cooler than my rig. Same HSF with dual fans too.
 

tyreman

Weaksauce
Joined
Mar 31, 2009
Messages
71
I'm tempted to liquid cool but I think I'd have to replace my Antec Solo case in order to do so. I have the single Tri-Cool exhaust fan in back running on low and two 92mm fans in front of the HDs pulling in cool air at the front plus the dual OEM fans on the Hyper 212+. Short of improving my TIM and/or application method, that's about as good as I'm gonna get and keep the fan noise acceptable.

I just built another Z68/2400 in a 400R Carbide case and it runs much cooler than my rig. Same HSF with dual fans too.


I am trying to remember methinks Antec Solo has the power supply mounting on the top of the case-+?
 

Sotiri

Limp Gawd
Joined
Feb 5, 2011
Messages
178
Here are my current results on my 2600K/P8P67 Deluxe. All other settings at default.

ocsettings.png
 

bigdogchris

Fully [H]
Joined
Feb 19, 2008
Messages
18,523
I have an OC question I want to ask w/o starting a new thread so any response would be appreciated.

I have a P8P67 Pro 3.0 with a 2500k and a S1283 w/bolt-through for cooling. I'd like to do a mild OC (maybe 4.3) without increasing the stock voltage. I'd also like to leave Speedstep enabled.

What settings should I look at to get me started other than just raising the multiplier? I don't mind raising the load line or other voltage tweaks, as long as I'm not raising the voltage itself. If a voltage increase is necessary, can I just adjust the CPU off-set? I've read that if you change only the off-set that you can leave Speedstep enabled while increasing voltage. So basically, what would be the minimal settings to look at?
 

speedlever

Limp Gawd
Joined
Sep 29, 2006
Messages
447
On my P8Z68-V Pro/2500k, I just bumped the multiplier to 43, set the BCLK at 100, set the DRAM voltage at 1.5 and pretty much left the rest on default/auto. I tweaked a few settings to prevent the double screen flash on boot, and it's running fine for me.
 

Sotiri

Limp Gawd
Joined
Feb 5, 2011
Messages
178
I agree. You should be able to reach 42-43 with just changing the multiplier.
Caution though, the MoBo increases the voltage automatically.
You may find your CPU is stable at a lower voltage than the MoBo throws at it, that's where setting LLC and offset comes in handy. I found the MoBo was giving way much more voltage than necessary. Setting the phases to extreme also helps keep the idle voltages down.

For example, In my results above with LLC and Offset in Auto, a 46 multiplier was hitting over 1.4v.
Trying out different combinations of LLC and offset got me to where I need to be.

Using this method allows the CPU to downclock in idle vs a manual Vcore where it runs at that voltage always.
 

bigdogchris

Fully [H]
Joined
Feb 19, 2008
Messages
18,523
I agree. You should be able to reach 42-43 with just changing the multiplier.
Caution though, the MoBo increases the voltage automatically.
You may find your CPU is stable at a lower voltage than the MoBo throws at it, that's where setting LLC and offset comes in handy. I found the MoBo was giving way much more voltage than necessary. Setting the phases to extreme also helps keep the idle voltages down.

For example, In my results above with LLC and Offset in Auto, a 46 multiplier was hitting over 1.4v.
Trying out different combinations of LLC and offset got me to where I need to be.

Using this method allows the CPU to downclock in idle vs a manual Vcore where it runs at that voltage always.
What does the phase setting do? Is that the digi+ load line calibration?
 

Eldata

Limp Gawd
Joined
May 10, 2011
Messages
366
I agree. You should be able to reach 42-43 with just changing the multiplier.
Caution though, the MoBo increases the voltage automatically.

I'm stable at a turbo ratio (TR) of 44 with all else relevant at defaults (P8P67 Deluxe and 2600K). Anyone else notice much lower idle Vcore @1600 MHz (offset mode) with TR set by per core as opposed to set by all cores?

Also, the behavior when Windows starts or resumes from sleep is entirely different. When TR is set by all cores, the turbo frequency is set and lasts for several seconds at startup or wake as though Windows is trying to optimize Vcore for all cores.

Anyone with sleep resumption issues and with TR set by all cores may want to try TR set by per per core instead. Or, vice versa. Again, this is for a P8P67 Deluxe and Vcore offset mode.
 
Last edited:
Joined
Nov 15, 2006
Messages
655
I'm stable at a turbo ratio (TR) of 44 with all else relevant at defaults (P8P67 Deluxe and 2600K). Anyone else notice much lower idle Vcore @1600 MHz (offset mode) with TR set by per core as opposed to set by all cores?

Also, the behavior when Windows starts or resumes from sleep is entirely different. When TR is set by all cores, the turbo frequency is set and lasts for several seconds at startup or wake as though Windows is trying to optimize Vcore for all cores.

Anyone with sleep resumption issues and with TR set by all cores may want to try TR set by per per core instead. Or, vice versa. Again, this is for a P8P67 Deluxe and Vcore offset mode.

So there is a big difference if set per core instead of all cores? Is stability affected at all? I currently have my system set to all cores, but am willing to change if there is a benefit, could you kindly explain more about this? I would appreciate it. Thanks.
 

Eldata

Limp Gawd
Joined
May 10, 2011
Messages
366
So there is a big difference if set per core instead of all cores? Is stability affected at all? I currently have my system set to all cores, but am willing to change if there is a benefit, could you kindly explain more about this? I would appreciate it. Thanks.

My observation only applies to Vcore default settings. I started off using TR by all cores as well since the only reason for using per core seemed to be if one wanted cores with different ratios. Found it odd that per core was used and each core was set to 48 here;

http://hardforum.com/showpost.php?p=1036716421&postcount=2

All the MB manual indicates is that by all cores allows adjustment in the O/S and all cores will be set to the same TR. Decided to try by per core several months later just out of curiosity. That's when I noticed that the default idle Vcore voltage was much lower, we're talking always over 1V before vs .8 to .9V after.

More significant was the way Windows behaved at startup or at sleep resumption. Apparently, the MB manual's "Can adjust in O/S" means that Windows will do any auto adjustments. Never noticed any effect on stability or maximum OC.
 
Joined
Nov 15, 2006
Messages
655
hmmm, interesting. I have my ratio set to 45x with everything on default. I may try per core and see if it makes any difference. I do get over 1.00v as well, usually 1.016v at idle. As long as stability isn't affected, it could be a quick and easy change in the bios to net a few watts worth of savings. Thanks for your input on this.
 

Sotiri

Limp Gawd
Joined
Feb 5, 2011
Messages
178
What does the phase setting do? Is that the digi+ load line calibration?

I meant the Phase Control setting. Extreme means full phase mode.
I found this setting drops the normal operating voltages.
For example, when I was playing around with x45, with Phase Control set to Optimized my voltage range was 1.048 to 1.392.
Just changing Phase Control to Extreme (Full Phase Mode) droped the range to 1.040 to 1.384.
 

speedlever

Limp Gawd
Joined
Sep 29, 2006
Messages
447
For the record, my P8Z68-V Pro/2500k looks like this:

43x - all cores, using CPU-Z: idle as low as 0.752v
P95 small FFTs: Core V jumps to 1.280v and stays there
LinX: Core V jumps to 1.304v and stays there
IBT: Core V jumps to 1.304v and stays there

43x - per core, using CPU-Z: idle 0.720v - 1.320v
P95 small FFTs: 1.280v
LinX - 1.304v
IBT: 1.304v
 
Last edited:

CPL0

Limp Gawd
Joined
Mar 22, 2010
Messages
292
Decided to try by per core several months later just out of curiosity. That's when I noticed that the default idle Vcore voltage was much lower, we're talking always over 1V before vs .8 to .9V after.

That's probably just the way the BIOS handles C3/C6. With the same maximum Turbo Ratio for each core C3/C6 is not required, with a different stepping between cores it is, that is if you wish for it to work the classic way.

If you use "set by all cores" and C3/C6 to enabled you should also see less than 1.0V vcore at idle, how much by will depend on how package c-states are set.
 
Last edited:

Sotiri

Limp Gawd
Joined
Feb 5, 2011
Messages
178
I just picked up http://www.newegg.com/Product/Product.aspx?Item=N82E16813131702 recently for a friend and I installed the AI software but I don't have access to the AI Auto OC.. is there a reason why? I want to push her 2600 to 4.8! In the EUFI I did the oc option and it's only 4.3ghz too...

43-44 is all the Auto OC gave me too. If you want 4.8 you'll have to do it manually.
I suggest you start with LLC High, Phase control Extreme, Current capability 130% and probably an offset voltage of 0.05. Be careful though, you'll probably be pushing close to 1.5V Vcore and a few sites are suggesting that anything over 1.4V begins to degrade the CPU.
 

speedlever

Limp Gawd
Joined
Sep 29, 2006
Messages
447
42x103=43.26 is what the EZ fast tuner did for me. When I ran the AISuite extreme auto-OC, it shut down the computer after several iterations. I cranked the computer back up and found it had gone to 47x. I'll have to play with that sometime and see what I can do with it.

OTOH, I am having a problem with a 45xOC using auto values. It won't pass the LinX test. I can tweak the BIOS and get a 45x to do 20 runs without a problem, but I thought it was supposed to be able to do 45x on auto settings. It has to be a voltage issue, but I thought auto would cover it at that mild OC.

Check out this thread and tell me where I've screwed up. Other than auto, I just plugged in the hard values for my ram and I think that's it on the AI tweaker page and the Advanced/cpu config page. I made a few changes to the other pages... enabled hot swap, etc. but that shouldn't affect the auto issue.
 

Sotiri

Limp Gawd
Joined
Feb 5, 2011
Messages
178
Auto settings will probably only get you 42-43. If you want an easy 45 OC just set LLC to Medium. This will give you more voltage than you need though. You can do better by setting LLC to High and starting with an Offset Voltage of maybe 0.02. Phase Control set to Extreme will also help bring down normal voltage. What I mean by normal voltage is the spikes during normal use, not during stress testing. For example, I see voltages up to 1.384 during normal use, but during Prime95 it's pretty steady at 1.368.
 

speedlever

Limp Gawd
Joined
Sep 29, 2006
Messages
447
Thanks. If you note in the thread I linked, my tweaked 45xOC uses a -0.015v offset. I tried -0.020 but it failed during LinX testing. As a point of reference, I know that some like a + offset and some like a - offset. I figured I'd start with a minus and see how it would shake out, slowly decreasing the - offset in 0.005 increments until it was stable. I felt lucky to find I only needed one incremental change to reach a stable OC with a -0.015 offset and didn't need to go positive.

Based on info in the first post of this thread, I thought that auto settings should handle OCs up to 47x. It did mention changing the VRM to 350, which I should try with the other settings in auto, just to see if that will work.
 

Sotiri

Limp Gawd
Joined
Feb 5, 2011
Messages
178
Also try setting CPU Current Capability to 130%. I found I could pass prime95 with lower voltage using this setting. It seems to be specifically suited for load testing.
 

speedlever

Limp Gawd
Joined
Sep 29, 2006
Messages
447
I'll try that.

Ya know... I'm sorta punching buttons in the dark here as I have little knowledge as to what the various parameters actually do. C1E, C3 or C6 for example. Or spread spectrum. And yet certain things being enabled or disabled appear to have an effect on performance and/or temps.

I wish there were guides somewhere that explained what the various options actually do. The guides I've seen basically provide settings to try without explaining what the feature actually does.

Some of the guides from Asus personnel say to use CPU Current Capability at 140% for OC (no X factor specified)... which makes it appear that any OC should be run that way.
 

Sotiri

Limp Gawd
Joined
Feb 5, 2011
Messages
178
I wish there was better documentation as well. I spent a great deal of time logging the difference each setting made.
 

oRuin

n00b
Joined
Dec 4, 2011
Messages
7
Thought i'd add my overclock. Used to use an offset of -0.100 for a while with LLC on extreme. Switched things around alittle and now have LLC on Regular and an offset of +0.020. I'm not sure which is best to use tbh. Seems 100% stable over the last week. Not bad for 1.3v.

 

speedlever

Limp Gawd
Joined
Sep 29, 2006
Messages
447
That's pretty nice. What HSF are you using?

Just for the heck of it, run your stress test again but use Real Temp 3.69.1 instead of Core Temp and see how your temps look. I've found CT to be quite a bit lower in reporting the temps than RT. I don't believe CT is accurate.
 

oRuin

n00b
Joined
Dec 4, 2011
Messages
7
Using the Corsair H100 Liquid Cooler.
Running the same test again but with RealTemp. I'll post back later.
 

speedlever

Limp Gawd
Joined
Sep 29, 2006
Messages
447
Ah... liquid cooled. Very nice.

I found a good 10*C difference (higher temps) with RT than with CT.

I think my case inhibits my cooling capability.
 

oRuin

n00b
Joined
Dec 4, 2011
Messages
7
Ran Prime95 for just over an hour early this morning, I had to stop it because I needed to get some sleep and would rather not run it through the night. Temps seemed to be about the same overall.
Im using IC Diamond 24-carat thermal compound, using the 'pea size' method. Pretty hard to apply that stuff, even after heating the tube up in some warm water.

 

speedlever

Limp Gawd
Joined
Sep 29, 2006
Messages
447
Maybe that liquid cooled HSF you run is the great equalizer. In my air-cooled HSF, I saw at least a 10*C diff between CT and RT.

I've not tried that TIM.. only used the TIM that came with the CM Hyper 212+ HSF.
 
Top