ASUS P8P67 Pro/Evo/Deluxe OC discussion thread

[coolhandluke41]

Pretty stable for me here as well.

As of now, I'm sticking to Medium LLC with +120 offset for a load Vcore of 1.408 using Linx, and 1.386 using Prime95.
Idle, voltage drops as low as 1.1V

Using an Ultra high LLC, then the idle voltage drop actually goes somewhat lower than 1.1V. Here I need +.055 offset.
Had a couple of ramdom BSOD with these settings so I went with the medium LLC instead.

More testing to go....

BTW, this is using bios 1253 and windows 7 SP1. Haven't tried 1204 yet...

1253... considered as flaky bios

[coolhandluke41]

If you guys are trying to get power savings with offset mode, then a low-volt 24/7 setup can be a good alternative.
Many of these chips should do 4-4.2GHz with low-volt, easily. Anything over 4Ghz would be overkill for most desktop application, anyways.

I've used manual-vCore and Load Line Calibration = Ultra High for @4.2GHz 1.19v/1.17v (Idle/load) and tested it with 30-min Prime.
I used it for several days without any problem, and then increased the vCore by one notch, just to keep it on the sure side, and call it stable, until it crashes on me.

The best part, it idles with 1.19-1.20v with little heat and power-usage, and all fans are running quietly at ~60%, etc ..
I've under-volted almost everything else too, including the PCH=1.0v to reduce leakage. Here it it's:

I was just going back through some of the posts ...
First of all you don't need LLC @Ultra HIgh for 4.2 at all
30-min Prime is a joke to call it " stable" no offense (min 1~6 hours in my opinion)
Here is the thing that caught my eye:
CPUPLL-- this voltage is for the internal clock generator for CPU (PLL that gives you 3.4GHZ for 26K and 5.2+ overclocked) maximum 1.89/1.8 stock if you want to save power and keep temps down ,you could turn it down to 1.71v

VCCIO aka QPI/VTT--processor power for I/O (voltage for integrated mem controller and PCI-E controller) maximum is 1.05+/- 3% no need in changing this 1.08 is absolutely safe and you can go as high as 1.2 for 24/7 it will allow you to OC you RAM but increase your temps

VCCSA aka System Agent Voltage--Maximum 0.971v ,minimum 0.879v
Stock 0.925v System Agent should Not be touched ,it is supposed to be fixed, it powers a lot of things that the VCC does not power like Power Control
Most of this is based on recommendations on overclocking the K series which i found very helpful and written by Sin
http://www.overclock.net/intel-gener...uide-p67a.html

here is the update on the offset ( i thing i found my 24/7 )
@ 4.7 Medium LLC @110% +.06 (minimum load 1.024v / max 1.368v)
stable under P95 blend (1hour+) and Linx (20 passes ),no BSOD so far ,will give few more days @"light load" and probably run P95 overnight
@ 4.7 High LLC@120% +.045 (min as low as 1.000v / Max 1.400 )

will report back if something pop out
P.S I guess i found choice #4
crank that bad boy up ,it's K series for crying out loud !

[ZenEffect]

i like 1253 better than 1204. i flashed back to 1253 using ezflash

[Sam_oslo]

I was just going back through some of the posts ...
First of all you don't need LLC @Ultra HIgh for 4.2 at all
30-min Prime is a joke to call it " stable" no offense (min 1~6 hours in my opinion)
Here is the thing that caught my eye:
CPUPLL-- this voltage is for the internal clock generator for CPU (PLL that gives you 3.4GHZ for 26K and 5.2+ overclocked) maximum 1.89/1.8 stock if you want to save power and keep temps down ,you could turn it down to 1.71v

VCCIO aka QPI/VTT--processor power for I/O (voltage for integrated mem controller and PCI-E controller) maximum is 1.05+/- 3% no need in changing this 1.08 is absolutely safe and you can go as high as 1.2 for 24/7 it will allow you to OC you RAM but increase your temps

VCCSA aka System Agent Voltage--Maximum 0.971v ,minimum 0.879v
Stock 0.925v System Agent should Not be touched ,it is supposed to be fixed, it powers a lot of things that the VCC does not power like Power Control
Most of this is based on recommendations on overclocking the K series which i found very helpful and written by Sin
http://www.overclock.net/intel-gener...uide-p67a.html

here is the update on the offset ( i thing i found my 24/7 )
@ 4.7 Medium LLC @110% +.06 (minimum load 1.024v / max 1.368v)
stable under P95 blend (1hour+) and Linx (20 passes ),no BSOD so far ,will give few more days @"light load" and probably run P95 overnight
@ 4.7 High LLC@120% +.045 (min as low as 1.000v / Max 1.400 )

will report back if something pop out
P.S I guess i found choice #4
crank that bad boy up ,it's K series for crying out loud !

Well, 30-min documented Prime is better that 30-hours stable-claims. We haven't seen any documentation about your "stable"-claims yet. I don't care how you test your setup, it's up to you. People test in different ways, but if you are going to call it a "joke", then you better do it right yourself.

I'm still experimenting with different setup, and those 30-min Prime-runs are just for getting a "feeling" how it's reacting to changes. None of them are final yet, and there is no need to have a 8-hour prime testing for each little change. This is a new platform and I would rather do my own experiments than following those rules and suggestions from the link you have provided. Now lets see if you can do it right yourself.

[Darkstormz]

Well, 30-min documented Prime is better that 30-hours stable-claims. We haven't seen any documentation about your "stable"-claims yet. I don't care how you test your setup, it's up to you. People test in different ways, but if you are going to call it a "joke", then you better do it right yourself.

I'm still experimenting with different setup, and those 30-min Prime-runs are just for getting a "feeling" how it's reacting to changes. None of them are final yet, and there is no need to have a 8-hour prime testing each little change. Now lets see if you can do it right yourself.

I have to agree with you. 30-min is just an indication. If you settle for 24/7 settings then you can start running 8-hour prime just to be on the safe side

[|ron]

Uhm... "System Agent should not be touched".. why?
I'm benching at 5,3ghz, with a voltage between 1,2 and 1,25 to the S.A. without problems... and all the clockers are doing so without broking a cpu per day

[coolhandluke41]

Well, 30-min documented Prime is better that 30-hours stable-claims. We haven't seen any documentation about your "stable"-claims yet. I don't care how you test your setup, it's up to you. People test in different ways, but if you are going to call it a "joke", then you better do it right yourself.

I'm still experimenting with different setup, and those 30-min Prime-runs are just for getting a "feeling" how it's reacting to changes. None of them are final yet, and there is no need to have a 8-hour prime testing for each little change. This is a new platform and I would rather do my own experiments than following those rules and suggestions from the link you have provided. Now lets see if you can do it right yourself.

sure buddy ,here ;
http://www.overclock.net/intel-cpus/...z-club-64.html
here is linx;


Uploaded with ImageShack.us

[coolhandluke41]

Uhm... "System Agent should not be touched".. why?
I'm benching at 5,3ghz, with a voltage between 1,2 and 1,25 to the S.A. without problems... and all the clockers are doing so without broking a cpu per day

benching and running 24/7 are two different things ,sure you can "touch" all you want

[|ron]

Oh ok, I didn't get you were talking about 24/7, sorry

[Lu(ky]

FYI: Gigabyte Gives Out Free Tool to Check 6-Series SATA II Issues works on any motherboards link ---------> H E R E

[Splave]

benching and running 24/7 are two different things ,sure you can "touch" all you want

just fyi, Ive taken both vtt and VCCSA north of 1.3v and no fire

[|ron]

Ehi splave, any improvement over the weekend?

[Sam_oslo]

Experimenting with different setup, and sharing the results, is one of the best side of such a forum and community. Specially for a new platform.

If one site suggests "System Agent should Not be touched", it may be right, or wrong, but we don't know before trying ourself. I personally would like to experiment myself, and see documented results form others, before ruling it as a fact.

I personally am interested, and have been experimenting with under-volting both VCCSA=0.9v and VCCIO=1.0v for low-volt 24/7 setup. I'm still testing and not ready to conclude yet, but it seams to be doing good anyways.

[Splave]

Ehi splave, any improvement over the weekend?

Increasing VCCSA while keep VCCIO lower helped get 5200 out of the 1 2600k but just blink command prompt if I try to boot x53. I have a 2500k I played with as well, and its wprime1024+32 pifast and aida bench stable....but fails superpi1m within first 1-10 loops

So that makes 3 chips that max out on this board at x52 multi and wont bench past 5200mhz hmmm.... bad luck or bad board?

[|ron]

Increasing VCCSA while keep VCCIO lower helped get 5200 out of the 1 2600k but just blink command prompt if I try to boot x53. I have a 2500k I played with as well, and its wprime1024+32 pifast and aida bench stable....but fails superpi1m within first 1-10 loops

So that makes 3 chips that max out on this board at x52 multi and wont bench past 5200mhz hmmm.... bad luck or bad board?

uhm... maybe it's not the cpus or the mobo taken singularly, but the combination between them that doesn't "work".
I would try another mobo in your case... it's really strange that "no boot" issue with 53x multi with 3 cpus.
I ordered 30lt of LN2, we'll see on sunday if we could squeeze something more out of our 2600k

[Splave]

Going to try psc ram this afternoon and see if it makes a difference then I will buy more chipz I guess

[G_Sup]

FYI: Gigabyte Gives Out Free Tool to Check 6-Series SATA II Issues works on any motherboards link ---------> H E R E

All this thing does is tell you which SATA ports are populated. Seems pretty useless to me.

[Gr1mR34p3r]

LLC Regular P95 Blend stable:
46x100 +0.030V (1.288-1.304V)
48x100 +0.090V (1.344-1.360V)

LLC High P95 Blend stable:
46x100 -0.040 (1.288-1.304V)
48x100 +0.010 (1.344-1.368V)

LLC Regular vs. High
46x100 LLC Regular +0.030V / 0.976-1.136V idle / 1.288-1.304V load / min 8.924W in idle (voltage from CPU-Z, W from HWiNFO32)
46x100 LLC High -0.040V / 0.920-1.064V idle / 1.288-1.304V load / min 9.013W in idle

Manual vs. offset voltage
46x100 LLC Ultra High 1.290V / 1.280-1.296V idle / 1.264-1.280V load / min 9.343W in idle - 65-71°C after 10min P95 Blend
46x100 LLC High -0.040V / 0.920-1.064V idle / 1.288-1.304V load / min 9.013W in idle - 68-75°C after 10min P95 Blend

Conclusion, Load = better on manual (less voltage needed, lower temp.), Idle = better on offset, higher the LLC, the better
I'll stick to offset because it runs 90-95% of the day in idle/light load.

These are still on default:
VRM Frequency
Phase Control
Duty Control
CPU Current Capability, didn't need to raise it from 100%, completely stable in idle

to test the "light" load stability,test linx 20 pass on 1 thread.
in my tests i could do 25 pass linx avx on 4/8 threads but was failing linx 1 thread on the second pass so i had to up offset to pass linx 25 runs on 1 thread.
You can do it with prime too.

By my experience, Prime95 Blend (I use 4 threads) should let you know where you are with idle stability, it BSOD's 0x124 instantly even on +0.01V more what was IntelBurnTest max 20x stable.
Should be almost fully stable with 30min Blend. For example, I got BSOD immediately after boot with -0.050V, was 8 hours stable with -0.040V.

Currently on 46x100 LLC High -0.040V, will see how stable idle is. On -0.050V it wasn't stable overnight, so 24 hours in idle should be enough to call it stable.
All in all I'm pretty happy with last 2 days progress, at least those WTF's with random BSOD's are gone.
My main concern remain temperatures with 48x100, cannot even test Prime for long time, will try to remount in next days.

[timchaos]

I've been having an issue with my USB mouse that I believe may be related to the computer going into sleep mode.

For anyone else having odd mouse-related issues, I can confirm they only occur after sleep. I disabled sleep and rebooted with no further issues. I found a post elsewhere that also said you can disable USB sleep in the power profiles for the same result, but I haven't confirmed that yet.

[Darkstormz]

Using T-Monitor i discovered that my cores will go from x45 -> x16 for about 1/10 of a second, then back to x45. It happends about every 7-8 seconds running LinX... any solution?

Temps are ok, 47c-50c-50c-48c

[crossg]

Booted to 54x after someone told me how to get past.1.5V.(thanks) This is rock solid stabe for at least 7 seconds. I keep taking screen shots with CPUz's on the same tab. Ram is 4 GB 1600 Gskill Ripaws @ 6-8-6-24. Need faster ram for under 7 sec I think.

[Splave]

^ awesome man knew it wanted more!

[Sam_oslo]

A little 30-min Prime-run @4.5Ghz to test low CPU-PLL.
I've tested this with same vCore before, but after playing with voltages in 4-4.2GHz low-volt action, decided to see how it can do with low CPU PLL=1.71v. It seams to be doing good for now.
I've also decided to reduced VCCIO=1.0v, and PCH=1.0v, because not running any high RAM-speed, and to reduce leakage in chipst. Nothing final yet, but so fa so good.

2600k @4.5GHz 1.27v 30-min Prime small FFT

[Sam_oslo]

You guys are not getting a lot of Event ID 1001 Source Dhcp ... in your event Viewer?

I've got 55 of these in a short time . That Bluetooth-thingy keeps trying to obtain an IP-address from Dhcp server every few minutes, and makes a new entry in log . I had to disable it from network adapter list to get rid of that error.

[crossg]

^ awesome man knew it wanted more!

Yeah thanks again for the help.

Got 55X boot and 1M SPi but it bsod'd (0124) 4 min into 32M SPi

I did try and increase the PCH and it walled right away. (blinking cursor)
VCCIO = 1.1000
VCCSA = Auto
CPU PLL = 1.8375.
PCH Voltage = Auto