Asus P6T Deluxe and i7 920 co/c1 - 4.2 ghz overclock problem

[Revrip1]

Hello,
I have a small issue/question about my recent overclock. First here are my computer spec's.

Asus P6T Deluxe Motherboard
Intel i7- 920 co/c1
Corsair 1600mhz R.A.M. (12 GB)
Windows 7 Ultimate x64
Liquid Cooled

I have it running at 4.2ghz and it will run Prime 95 for 12 hours and then prime fails. The computer doesn't BSOD, just Prime 95 fails at 12 hours in. Any suggestions would be greatly appreciated. I am kind of new to overclocking other than what I've read. Here are my bios settings.

Ai Overclock Tuner - (Manual)
CPU Ratio Setting - (21.0)
Intel SpeedStep Tech -(Disabled)
BLCK Frequency - (200)
PCIE Frequency - (100)
DRAM Frequency - (DDR3-1603 MHz)
UCLK - (3408 MHz)
QPI - (7218 MT/s)

DRAM Timing Control - 9-9-9-24-2N (All Set to Auto)

CPU Voltage - (1.48125)
CPU PLL Voltage - (1.92)
QPI/DRAM Core Voltage - (1.48125)
IOH Voltage - (1.10)
IOH PCIE Voltage - (1.50)
ICH Voltage - (1.10)
ICH PCIE Voltage - (Auto)
DRAM Bus Voltage - (1.64)
DRAM DATA REF Voltage on CHA - (Auto)
DRAM CTRL REF Voltage on CHA - (Auto)
DRAM DATA REF Voltage on CHB - (Auto)
DRAM CTRL REF Voltage on CHB - (Auto)
DRAM DATA REF Voltage on CHC - (Auto)
DRAM CTRL REF Voltage on CHC - (Auto)

Load-Line Calibration - (Auto)
CPU Differential Amplitude - (Auto)
CPU Clock Skew - (Auto)
CPU Spread Spectrum - (Auto)
IOH Clock Skew - (Auto)
PCIE Spread Spectrum - (Auto)

Advanced - CPU Configuration

CPU Ratio Setting - (21.0)
C1E Support - (Disabled)
Hardware Prefetcher - (Enabled)
Adjacent Cache Line Prefetch - (Enabled)
MPS and ACPI MADT ordering - (Modern ordering)
Intel Virtualization Tech - (Enabled)
CPU TM Function - (Disabled)
Execute Disabled Bit - (Enabled)
Intel HT Technology - (Enabled)
Active Processor Cores - (All)
A20M - (Disabled)
Intel SpeedStep Tech - (Disabled)
Intel C-State Tech -(Disabled)

Also, my temps are pretty good when running Prime 95. Average of around 78c.
Thanks, Revrip1

 

[elitesix]

Sounds like borderline instability - what is happening to you was common as I overclocking my 920 and just took lots of trial and error. Could be anything. CPU. Ram. Motherboard. PSU. Upping the appropriate voltage usually can fix it though, unless you're just hitting the OC limit.

I've definitely resolved the problems you describe by finding out it was the cpu, or ram, or psu, or motherboard chip. So it really is a tossup. Your cpu voltage is already pretty high though so I doubt it's that.

I'd also add linx stress testing (I use the OCCT frontend for it) since it usually detects 12 hour prime95 errors in typically around 2-3 hours of testing. Gl!

 

[Revrip1]

Hello,
Thanks for the reply! I have tried a little more Vcore Voltage 1.4875 and Prime 95 fails in 5 hours. So I think I am at my voltage limit. I thought about raising CPU PLL Voltage to 1.97 volts, as I saw someone else had it that high on a 4.0 overclock with basically the setup as mine. And also raising my DRAM Bus Voltage to 1.66. But I am kind of scared to go that high on the DRAM Voltage, when my RAM only calls for 1.6 volts @ 1600 Mhz?

 

[elitesix]

Hello,
Thanks for the reply! I have tried a little more Vcore Voltage 1.4875 and Prime 95 fails in 5 hours. So I think I am at my voltage limit. I thought about raising CPU PLL Voltage to 1.97 volts, as I saw someone else had it that high on a 4.0 overclock with basically the setup as mine. And also raising my DRAM Bus Voltage to 1.66. But I am kind of scared to go that high on the DRAM Voltage, when my RAM only calls for 1.6 volts @ 1600 Mhz?

If you want to test the RAM, you could keep the voltage the same while significantly loosening the timings or lowering the Mhz and see what happens.

 

[Revrip1]

Thanks, I will give that a try. I will get back to you and let you know the results. Thanks again for your reply!

 

[turbobooster196]

I have also the deluxe motherboard, these are my stable settings at 4.2 on air and HT on. 100% stable

cpu ratio 21
speedstep disabled
bclk 200
uclk 3208
dram on 1603mhz
timmings 9-9-9-26 1.64vcore
qpi link data auto
cpu vcore 1.325v
cpu pll 1.84
qpi/dram vcore 1.325
ioh voltage auto
ich voltage auto
llc auto
cpu differential amplitude 700mv
cpu clock skew auto
ioh clock skew auto

of course all power options are of, and I set the ram timings a bit more lower, so instead of 7-7-7 a set it at 9-9-9

and for the rest the difference between you and my beside the cpu is that I set cpu differential amplitude to 700mv/uclk at 3208 and qpi at aut

 

[Revrip1]

Thanks for the reply,
I will mess around a bit with some of the settings you have used and see if that works.

 

[turbobooster196]

top, don't forget, I have a D0 so need less vcore then you.
goodluck.

 

[Evilsizer]

also if you need help and prime95 or orthos or intel burn test fails, if it gives you a size (i know prime95 and orthos does) please post it. the size it fails at tells you in a general area where to poke at. as it could fall inside the size of the L2 meaning more voltage related to the cpu or another cpu voltage. if it falls into the range of being in ram, that could mean more IMC voltage or ram voltage/different timings/speed.

 

[Revrip1]

Thanks again! Been really busy, but will mess with it tonight.

 

[SF101]

had the same issue with the same mobo and a 930.

4.0 is fine but 4.2 fails without ability to fix issue i tried for weeks to get it stable and had nothing but random lock ups.

i settled on 200 bclock - 20 multi no more issues and far less voltage.

sometimes you have to take that also into consideration after a point of so many more volts its not worth the extra heat and volts to get that extra 200mhz esp @ multiplayer mhz instead of bclock.

sometimes you just need to put those equations on the table and be happy with settling where the chip likes to run.

what this means? 1.4875 is far to many voltages for +200mhz when you should be able to get 4.0 @ 3.0-3.2v

 

[elitesix]

had the same issue with the same mobo and a 930.

4.0 is fine but 4.2 fails without ability to fix issue i tried for weeks to get it stable and had nothing but random lock ups.

i settled on 200 bclock - 20 multi no more issues and far less voltage.

sometimes you have to take that also into consideration after a point of so many more volts its not worth the extra heat and volts to get that extra 200mhz esp @ multiplayer mhz instead of bclock.

sometimes you just need to put those equations on the table and be happy with settling where the chip likes to run.

what this means? 1.4875 is far to many voltages for +200mhz when you should be able to get 4.0 @ 3.0-3.2v

Re: Core Voltage Suggestions: Supposedly intel documents state that 1.55v is the max voltage for i7 920s. However, I haven't read anyone whose has ran their original i7 at higher than 1.45v for long periods of time. Though I haven't read stories of i7s dying at higher voltages suddenly either. All things considered, I still think it's sound advice to try to lower your core voltage if possible.

Re: 20x multiplier: I've overclocked a variety of setups over the years, but only in my current 920 setup do I get instability with even multipliers, and stability with odd multipliers, even at the same overall mhz. Never seen a preference for even/odd multipliers before (maybe the motherboard is varying something else based on these settings though I've verified these multipliers do not touch dram or pcie speed, so I am at a loss to explain it), so be aware of this.

 

[Revrip1]

Thanks for the reply! I will try and lower my vcore and raise my qpi a bit like your and see what happens. Thanks again. I will let you know.

 

[elitesix]

Also, SF101 is right about the relationship between the mhz wall and voltage. You start to see the voltage required for stability to increase rapidly as the mhz wall is hit and then no extra voltage helps. Your wall might be around 4.1 or so based on what you've told us.

For example, in my own setup, my mhz wall is around 4.25 Ghz with HT off, and my voltage starts scaling up rapidly:
3.9 Ghz: 1.29v or .04v over stock for 1.23 Ghz gain
4.03 Ghz: 1.41v or .16v over stock for 1.33 Ghz gain
4.2: 1.445v or .185v over stock for 1.53 Ghz gain

You also see this effect in this 920 oc investigation: http://www.tomshardware.com/reviews/overclock-core-i7,2268-4.html

I've tried lots of voltage to get 4.3; I can boot, I can keep temps under 85C, but no go, can't get it stable at 4.3. YMMV or you yet may have luck with playing with the secondary voltages.

 

[Elfie]

I'd suggest disabling the following instead of having them set to Auto:
CPU Spread Spectrum
PCIE Spread Spectrum

 

[Revrip1]

Hey Thanks for your replies! I finally got it to run stable!!!
I had to lower BLCK to 198 and disabled CPU Spread Spectrum and PCIE Spread Spectrum.
I also, increased PLL from 1.92v to 1.94v. And DRAM Bus Voltage from 1.64v to 1.66v.
And kept CPU voltage and QPI/DRAM the same.
I tried to lower both the CPU and DRAM Bus Voltages with no success.
It just ran Prime 95 for 24 hours with no errors!!!

Thanks again for all of your support and suggestions. I would like to hear what you think of my final settings though?

 

[elitesix]

I would like to hear what you think of my final settings though?

Your final OC 198x21 @ 1.48v w/ HT enabled is very similar to what my i7 920 OC c0/c1 is turning out to be. I haven't yet tweaked mine to its full potential though, unlike you (currently testing 4.03 @ 1.456v w/ HT enabled on air cooling).

These c0/c1s sure do love the juice!

 

[Revrip1]

Yeah, your right about the juice. WOW!!! Lots of it to run stable. But I guess as long as it's running cool (78c) under full load and (42c) while at idle everything should be O.k.?
At least I hope so. If not oh well it's not that much money for a new chip or ram memory modules.
Thanks for the reply and good luck on yours! Let me know how it turns out.

 

[[W2k]Shadow]

tempreture looks ok, i reach the same under water, if i have 30°c in the room and prime with furmark is running.

Tj is at minimum 100c, so you are away from it.

 

[Revrip1]

Thanks, That's good to hear!