Sign in with
Sign up | Sign in
Your question

Cold boot start, stop, start problem

Last response: in Motherboards
August 22, 2011 2:15:55 PM

P8p67 ws Revolution mobo
i5 2500k
8GB g.skill RAM

It seems that many people are having the same problem when powering on their system. The fans spin, then all stops, then the computer starts. Sometimes this cycle of starting and stopping repeats several times.

I just want to know if their is anyone with the P8P67 WS Revoltution board that is not having this issue? I don't know what the solution is and don't know getting a new board would solve it or not.

August 22, 2011 4:11:13 PM

The first place to look, and the simplest, is try booting with (1) stick in either colored DIMM slot but closet to the CPU.

Q - What G.SKILL? I've been seeing problems with bad kits of F3-12800CL9D-8GBXL
August 22, 2011 5:13:05 PM

Thanks for the suggestion. I am using 2x F3-12800CL9D-4GBXL.

Turns out it was a bios issue. I just finished updating the bios and all is good. The cold boot start/stop problem is fixed and so is another problem waking from sleep mode.

If anyone else is having either of these issues and reading this, go for the bios update first.
Related resources
August 22, 2011 7:16:46 PM


Problem is back. I'll check memory.

Any other suggestions?
August 22, 2011 7:34:15 PM

booted with one stick of RAM and same problem: fans start and mobo seems to be powering up, then all stops, then power comes on.

Anything else that I could try?

I am using an OCZ ZX 1000W PSU. Could it be the PSU?
August 22, 2011 7:36:36 PM

First try one {matched} set at a time in the same colored slots.

In the BIOS lets set the RAM Manually:

AI OC Tuner -> Auto
DRAM Frequency -> DDR3-1600 MHz
DRAM Timing 9-9-9-24 & 2 {CAS-TRCD-TRP-TRAS & CMD}
DRAM Voltage -> Auto ; failure 1.55v
VCCIO -> Auto ; failure 1.125v~1.20v

Save & Exit = Yes

Next swap kits and retest.

Next add both sets and test
August 22, 2011 7:40:52 PM

vollman1 said:
booted with one stick of RAM and same problem: fans start and mobo seems to be powering up, then all stops, then power comes on.

Anything else that I could try?

I am using an OCZ ZX 1000W PSU. Could it be the PSU?

IF they're all failing AND you're using the Second DIMM slot right of the CPU then it's something else.

I've posted the following more times than I can count, but it isoltes problems quickly vs HOURS:

Q - What HSF?

Anytime I build the build is tested on a bench first, Tested & Benched, then installed in the Case. If you're not on water by the time you posted here {forum} to now all of this could have been done a several times over.

You can try Lazyman's Breadboard -
* Disconnect ALL Front Panel & USB Headers ; keep only the PWR & Ground <best> short w/wire.
* Disconnect ALL peripherals including the Keyboard/Mouse
* (1) Stick of RAM in the second DIMM slot from CPU
* Unscrew both the MOBO and PCIe screws

Pull the MOBO away from anything conductive and supported by a towel ; short PWR & Ground to start.
August 23, 2011 1:20:31 AM

KK setting RAM manually, but I only have the option of DDR3-1648MHz for frequency next is 1922.

VCCIO is set to auto; not sure where to put the fail numbers in....

If I need to test another kit, I may have to RMA the ones that I have here :( 


The problem seems to not be consistent either; sometimes it happens once (on-off-on then boot, sometimes 3 times (on-off-on-off-on-off-on then boot), and sometimes it does happen at all.
August 23, 2011 2:51:38 PM

Both the voltages are found in the same section as your DRAM Frequency; use the 'DDR3-1648MHz' the base clock varies 98MHz~101MHz all of the time.

Inconsistency is often a symptom of voltage. IF you updated to the latest BIOS then the Phase power defaults are all on which should also help.

Again, you could have a bad stick of RAM. Create a bootable CD/DVD of Memtest, use the ISO/zip file and burn the CD/DVD, boot off the CD/DVD and run the test for at least an hour and see if you get errors ->
August 23, 2011 10:33:21 PM

I'll try the mem test and see what happens. Hope its just the RAM. I'll post the results in the next day or two. Thanks again.
August 26, 2011 1:56:51 AM

Having trouble posting, in case there is more than one update post.

Anyway, I ran MEMTEST for a little over and hour and it made almost two complete passes with no errors. Does this mean the RAM is good?

Where do I go from here? Any help is appreciated.

P.S. not sure it means anything, but I noticed that when I unplug my mouse and keyboard and plug them into different USB ports, I get a "loading drivers" message and the computer seems to go back to the start, stop, start again.

August 26, 2011 2:42:00 AM

It would seem that 'something' on the USB is bad, try a completely different keyboard and mouse. Otherwise, boot with the keyboard disconnected.

Next, open the Device Manager and look for {! or ?} devices listed, also only use the drivers listed at the ASUS site and NEVER use the ones that shipped - 9/10 outdated and generally for a good reason.

Also, look at the Event Manager and in particular for Critical errors. See below.

In building, I prefer PS/2 that are known 'good'.

Event Manager:
August 26, 2011 2:52:23 AM

I am getting 2 errors:

Event filter with query "SELECT * FROM __InstanceModificationEvent WITHIN 60 WHERE TargetInstance ISA "Win32_Processor" AND TargetInstance.LoadPercentage > 99" could not be reactivated in namespace "//./root/CIMV2" because of error 0x80041003. Events cannot be delivered through this filter until the problem is corrected.


Failed extract of third-party root list from auto update cab at: <; with error: A required certificate is not within its validity period when verifying against the current system clock or the timestamp in the signed file.


Found a third critical error which shows recurring from the build date:

The system has rebooted without cleanly shutting down first. This error could be caused if the system stopped responding, crashed, or lost power unexpectedly.
August 26, 2011 3:29:50 AM

Also showing a yellow ! for the device "Marvell 91xx Config ATA Device"
August 26, 2011 1:41:12 PM

Here are the fixes that should correct the stated problems, though the CIMV2 may require more investigation to solve. I assume you're running the latest SP -

1. 0x80041003 {I see this in Vista} - and
2. ->
3. Marvell 91xx Config ATA Device - right click, uninstall, reinstall from ASUS site and reboot. Next, verify both AHCI in the BIOS and Windows match-up. Start = 3 IDE, Start = 0 AHCI:
b. verify ->
c. if you're not using Marvell then IMO Disable in BIOS.

Let me know!
August 27, 2011 3:54:39 PM

Thanks for the advice and links. I have fixed all but one of the errors and I think that I will just leave that one alone for now.

I have more info on the start, stop, start problem: I reset the bios to default settings and the problem seemed to stop happening. Then I ran the automated OCing program and the problem came back again. I am not sure if it is a board defect or a software glitch or even a problem with OCing.

Any thoughts on how to trouble shoot this? My main concern is that I may have a defective mobo and my 30 days are running out.
August 27, 2011 4:39:06 PM

What is the 'one if the errors'?

Simple, uninstall EVERY OC App or BIOS invasive app installed on the PC. Count the results just in Tom's -

The only value to ASUS AI Suite II and alike is the bling followed by the bang. OC a SB on P67/Z68 is too damn easy already.

The App's are defective, if the 'stock' settings run then your MOBO isn't bad, just the Apps and/or OC. A Clear CMOS sets the Defaults, but the Apps set them right back.

Isolate, all defaults, no BIOS OC Apps, and the following:
In the BIOS lets set the RAM Manually:

AI OC Tuner -> Auto
DRAM Frequency -> DDR3-1600 MHz
DRAM Timing 9-9-9-24 & 2 {CAS-TRCD-TRP-TRAS & CMD}
DRAM Voltage -> Auto ; failure 1.55v
VCCIO -> Auto ; failure 1.125v~1.20v ; may require AI OC Tuner -> Manual

Save & Exit = Yes

If that works then the RAM, CPU, MOBO all are fine. Prime95, Prime95 blend, and Memtest 4-passes. When I build it begins on a bench, BIOS updates {good ones}, basic OS and everything is tested at defaults -> pass/bench then OC -> pass/bench THEN the build begins. Otherwise you're trying to solve a problem with a VERY long list of variables. You're especially crazy if you don't test before installing water blocks, etc. 5 minute or 5 days - pick one.
August 27, 2011 6:59:28 PM

Reset the bios to defaults and enabled Hot Plugging for the SATA ports (1 eSATA and 2 SATA hot swap bays) and the problem goes away.

At this point, I really feel that 2 things are causing the problem:

1- the auto oc software

2- SATA ports not being set to hot plug

Does this seem reasonable?
August 27, 2011 7:15:08 PM

To eliminate item (2) SATA; verify that the Start = 0 registry values are indeed set correctly, chances are they are not correct. See -> and either set them manually or allow Fix It to change them.

In either case if everything is AHCI then the registry should look as below.

As far as the OC Software absolutely uninstall it. It's like a 'blanky'.
August 28, 2011 8:38:39 AM

You had asked about which error I hadn't resolved earlier, but I missed it. The error that I didn't "fix" was the 0x80041003 error. After reading that linked thread, it seems that the advice may only prevent the error from being recorded not actually fix it. I also read the last post where the guy's system crashed from trying to fix it. So I just left this one alone. I would like to know what is causing it, but it doesn't appear to be causing any noticeable problems for now.

OC software is uninstalled, as you recommended. But in the bios there is an option to change the profile of the system from energy efficiency to normal to high performance. I think that this is built into the bios and I don't know if it can be removed. I think that it changes bios settings automatically depending on which option is selected. Should I leave this at "normal" setting?

In the iaStorV the start did =3. Changed to start=0.

I have one more item in msahci between the Start and Type items. It reads:

Tag REG_DWORD 0x00000040 (64)

Don't know what this is.

Also, In setting the RAM manually, I cannot find where to enter the failure numbers for:

DRAM Voltage -> Auto ; failure 1.55v
VCCIO -> Auto ; failure 1.125v~1.20v ; may require AI OC Tuner -> Manual

Also, I am getting another consistent error:

The server {E10F6C3A-F1AE-4ADC-AA9D-2FE65525666E} did not register with DCOM within the required timeout.
August 28, 2011 4:08:57 PM

The first link was to the 'WMI Diagnosis Utility' to hopefully diagnose the problem. 'My' way of dealing with Widows errors unless I fix them 'quickly' is to format and re-install Windows from scratch.

Start = 3 is IDE drivers and Start = 0 is AHCI drivers. As far are the other values leave them alone, every registry will have some variance, and honestly 'Tag REG_DWORD 0x00000040 (64)' value to me is not known so IMO leave it alone. edit: Also since this can cause corruption see this image and on the primary drive select both options and schedule a disk repair. If you select both options on the secondary drive there's no harm AND if you schedule on both drives then it'll run the repair one after the other. Note: 1TB takes a l-o-n-g time to SFC. I run these routinely ever 3~4 months. See ->

DRAM Voltage & VCCIO Voltage - refer to page 3-8 and 3-13 of your manual. ONLY change these values IF you are still having memory errors.

The 'DCOM' error, it can happen from a few things my feeling it's your GPU drivers, so uninstall them an see if the that error goes away. If it does then {ass-u-ming} you have the latest installed from nVidia's or AMD's site then download and install the {2ND to the NEWEST} drivers.

Q - Did you install ALL of the latest drivers from ASUS's website?
Q - What GPU?
August 28, 2011 6:42:25 PM

Yeah I installed the newest drivers for NIC, bios, chipset, USB, Audio and Video card.

I have a Power Color Radeon HD 6950 2MB right now and looking to Xfire soon.

I uninstalled the video driver and it seems that the error did not come up when I rebooted. I looked for the second to latest driver, but cannot find it on the AMD website. Only brings up the latest driver. I'll keep working on this one.

The Start, Stop, Start problem (as well as a related problem where the system cannot recover from sleep mode) is almost definitely related to OCing. I was setting manual values and ran into the exact same issues as using the automated software. Resetting to defaults always seems to fix the problems.

So, I should ask is there a setting that I am missing when manually OCing?

The options that I made sure were disabled:

Limit CPUID Maximum
Speedstep Technology
Spread Spectrum

The options that I made sure were enabled:

Internal PLL Overvoltage
Execute Disable Bit
Intel Virtualization Tech

Then, I just upped the multiplier. Nothing else.

Of all of these, I have found that others have disabled the Internal PLL Overvoltage to fix the Start, Stop, Start issue. Not sure what to do about this when OCing....
August 28, 2011 7:14:47 PM

Parroting is a problem - even the same EXACT MOBO + CPU + HSF will render different voltages, temps, vCore, etc. Next your MOBO as I recall has Phase controls so I'll probably have you start from the DEFAULT BIOS set. I will only be able to give you ranges. Example Virtualization is not needed unless you are running your PC as a server of some sort, and worst I don't even know if your MOBO supports VT-d/Vt-x.

Q1 - What target OC on your i5-2500K are you 'trying' to obtain?
Q2 - What HSF? Hows it connected e.g. 4-pin / 3-pin fans, etc.
August 28, 2011 7:24:08 PM

I was hoping to get to maybe 4.9 or 5.0.

I am using the Xigmatek Gaia sd1283 cooler and it has a 4-pin connector to the CPU Fan header on the mobo. But, I did recently buy an Aerocool X-vision fan controller that I plan on using.

I also have a Coolermaster HAF X case with a total of 4 case fans.

I do run XP Mode so I need the virtualization enabled.
August 28, 2011 8:30:39 PM

First 5.0GHz is a hit or miss just from the CPU alone; most i5-2500K won't OC to 5.0GHz. I would never use the XIGMATEK Gaia SD1283 for 5.0GHz. Just because some 'guy' can get a CPU-z Validator screen capture doesn't mean their PC is stable by any stretch of the imagination - run Prime95 or Prime95 blend with that HSF I would imagine the PC in a refrigerator or in the Arctic tundra.

5.0GHz // Default Values + listed changes:
AI OC Tuner -> Manual

DRAM Frequency -> DDR3-1600 MHz
DRAM Timing 9-9-9-24 & 2 {CAS-TRCD-TRP-TRAS & CMD}
DRAM Voltage -> 1.55v {closets menu value}
VCCIO -> 1.125v {closets menu value}

CPU Ratio -> 50
Turbo Mode -> Disabled
Load Line Calibration -> Extreme
Phase Control -> Extreme

VRM {might need to be set manually; 350~400}

CPU Mode Sign -> +
CPU Offset Voltage -> 0.065 {0.025~0.085} ; depends on the CPU's behavior

CPU Spread Spectrum -> Disabled

Save & Exit = Yes

You MUST have a CPU core temp App - WATCH THE TEMPS!!!

Set the Gaia SD1283 FAN(S) on Highest 100% Fan Speeds!!!

1. Prime 95 as a "Stress Test"
link -

2. CPU-z for Settings Summary
link -

3. CoreTemp 1.x

Good Luck!
August 28, 2011 10:55:08 PM

I appreciate all of your time and insight. I will keep tinkering with OCing and see if I can get a stable OC that doesn't cause the original start, stop, start problem.

You really have been a great resource. Thanks for all the help!
August 29, 2011 10:21:12 AM

hahaha I don't think that I will have to worry about reaching 5.0: I was only able to hit 4.2 with multiplier lol

Maybe 4.5-4.7 is more realistic (probably closer to 4.5)....
August 29, 2011 4:14:32 PM

This is one of the best posts for OC your family of MOBOs ->

The limits are two fold: CPU and Thermal Limits. IF you want higher then you'll need to shop for a new HSF, and that's all about budget.
August 29, 2011 7:40:31 PM

Oh yeah I was reading that post that you linked. Really good stuff.

I entered the same settings that this guy did on his i7 2600

All of my settings are the same as the pics and I ended up with 4.5 MHz at 1.392V. It's been stable on P95 for like 4 hours so far. The Voltage seems high though and I don't know how to lower it since the settings are mostly on auto.
August 29, 2011 8:22:24 PM

Simple, lower the voltage the example was for 4.8GHz.

You 'should' at 4.5GHz get stable from {low side :)  }1.25->1.30->1.35v {high side :??:  } My guess is 1.28~1.32v for your MOBO @ 4.5GHz...
August 29, 2011 8:34:05 PM

I feel really dumb, but do you know which setting I should change? I think it is all handled on auto now.

I lowered the LLC from Ultra high to High and voltage dropped to between 1.32V-1.352V.

I am also thinking of changing Phase Control.
August 29, 2011 9:26:20 PM

High(er) Phase = Low vCore + added stability so leave it alone. Randomly, changing stuff can and typically has adverse affects. Leave the LLC on either Ultra High <or> High. Afterwards verify the stability affects with Prime95 + Blend test.
August 29, 2011 9:48:31 PM

glad I asked about the Phase Control. Really wasn't sure about that one. I am P95 blend testing now with 4.5MHz @ the 1.352V and it is stable so far (about an hour running). Temps are 69-76c across all four cores.

I want to do a 24 hour P95 test once I get the voltage as low as I can. Is that a good stability test time-frame? Or is a shorter time frame acceptable?

Where do I go from here?
August 29, 2011 10:40:00 PM

I run the regular Prime95 overnight. If everything is good you should be able to run Prime95 forever.

If you're trying to tweak the vCore then 20~30 minutes is a pretty good tell all.
August 30, 2011 1:01:56 AM

Is a rounding error in one of the cores is a sign of OC instability?

Oh and I found out more information on the original question of the cold boot issue:

If I understand correctly, the issue effects all P67 boards. The reason that the auto OC program was causing the start, stop, start is because it was setting the BCLK over 100. In my case it was setting the BCLK to 103.3 I think and the fix was to manually set this to 100. If Internal PLL Overvoltage effects this, then that is why disabling it also "solved" the issue.

Just thought you might be interested in this info.
August 30, 2011 1:59:45 AM

The BCLK bounces around from load an Intel's SpeedStep 98~105MHz. If you want to reinstall AI Suite then that's crazy. The BCLK isn't per core and instability can be caused typically >105MHz but it's not so much the CPU it's the SATA, USB, etc that are all tied to the SB 'new' put everything on the BCLK approach.
August 30, 2011 2:08:29 AM

No I think that I will leave the AI Suite uninstalled for now :) 
August 30, 2011 4:43:51 PM

Well, I was running at 4.5MHz @ 1.33V for 10 hours and I thought I was stable, but then I got an error in the fourth core.

Back to the drawing board.
August 30, 2011 5:32:48 PM

LLC on Ultra High, keep in mind 10 hours is damn good @ 4.5GHz :) 

Weird as it sounds, often a higher OC works easier. Go figure...
August 30, 2011 11:49:50 PM


setting up the tests with the LLC change now.

Old setting was giving me ~70'c temps at 100% load and ~45'c at idle. With LLC=Ultra High increase of ~5'c on both. Do you have an opinion on whether this is acceptable or not? Seems hot to me and I want to keep my chip running as long as I can without it being burnt out.

I am also going to try messing with the fans to get better air flow I think....

P.S. I am getting a higher core voltage when idling than when under load. Is this normal?
August 31, 2011 12:45:56 AM

My opinion was and still is the HSF. I prefer temps in the 60's, and with a vCore of 1.3x it's certainly obtainable. Again, Prime95...10 hours is damn good @ 4.5GHz

Pull the side off you case and compare temps. If it's >3C~4C then you have an airflow problem. At 4.5GHz the case fans should be a 100% with your setup.
August 31, 2011 2:39:19 AM

I was thinking about getting a Corsair H60 cooler but changed my mind because I didn't know about it potentially leaking. Maybe I should just buy one and install it....

About the core voltage being higher at idle than at 100% load: is this normal?
August 31, 2011 1:46:30 PM

I have now 10 H50's and 1 H60, don't confuse factory sealed {permanently} with cut, spliced and crimped plastic tubing running everywhere. I use the H50/H60 because their simple to clean and maintain, and run them on office PCs.

vCore will vary from Vdroop and Loadline Calibration this is to a degree a byproduct SpeedStep. The vCore doesn't concern me as long as it's 'low' <1.35v~1.37v AND the temps are good 40C~45C Idle and 60C~65C Load.
September 1, 2011 7:03:49 AM

I have to decrease my multiplier or get a better cooler to keep my 100% load temperature below 72'C. Raing the LLC to Ultra-High puts me into ~76'C range.

I am strongly looking at the Corsair H100

You said 10 hours was good; is it considered stable? Or do you want to be able to run longer without errors? Maybe more to the point: how long do you run Prime95 before considering an OC "stable"?
September 1, 2011 5:31:08 PM

10 Hours Prime95 - Gaming stable yes, Production stable no. The best analogy is your car's water temp indicator bring well into the Yellow versus in the Green.

The H100 will and can keep the temps in the 50C~55C with a 4.6GHz~4.7GHz @ 1.32 vCore. Most folks replace the stock 120mm with quieter push/pull to reduce the noise levels at high fan. Most of the reviews for the H100 are on i7 9XX which are inherently hotter 5C~10C+. I've heard there's an H120 in the works, my assumption is the radiator will be thicker like the H70.

Other things that affect temps: Ambient {24C} temps, Thermal Paste {Arctic Cooling MX-4 or Arctic Silver 5}, pulling cool air in or pushing warm case air out, etc.
September 3, 2011 4:01:24 PM

Well, I was able to get 4.5GHz @about 1.33V stable on Prime95 small FFT. I stopped the test at 18hours20min or so. This is with Speedstep disabled.

I kind of want Speedstep, but when I enable it, I cannot run 3D Mark Vantage without a bsod.

Any thoughts?
September 3, 2011 4:45:37 PM

How about as I first posted with Turbo Disabled? You've come a long way from ASUS AI Suite II to now :) 

Turbo Mode -> Disabled
Enhanced SpeedStep Technology -> Enabled

At this point I don't know what you changed or have set in the BIOS?
September 3, 2011 7:17:17 PM

Stable 4.5GHz settings are:

AI Overclock Tuner: Manual
BCLK/PEG Frequency:100
Turbo Ratio: By All Cores
By All Cores: 45
Internal PLL: Enabled
Memory Frequency: DDR3-1600
EPU Power Saving: Disabled

CPU Power Management
CPU Ratio 45
Enhanced Speed Step Tech Enabled
Turbo Mode Enabled
Long Duration Power Limit Auto
Long Duration Maintained Auto
Short Duration Power Limit Auto
Additional Turbo Voltage Auto
Primary Plane Current Limit Auto

Load Line Calibration: High
VRM Frequency: Manual
VRM Fixed Frequency Mode 350
Phase Control:Extreme
Duty Control: Extreme
CPU Current Capability100%
CPU Voltage: Offset
Sign: +
Offset Voltage: 0.025
DRAM Volatge Auto
VCCSA Voltage: Auto
VCCIO Voltage: Auto
CPU PLL Voltage: Auto
PCH Voltage: Auto
DRAM Data Reference Voltage (CHA A) Auto
DRAM CTRL Reference Voltage (CHA A) Auto
DRAM Data Reference Voltage (CHA B) Auto
DRAM CTRL Reference Voltage (CHA B) Auto
CPU Spread Spectrum Disabled

CPU Ratio 45
Intel Adaptive Thermal Monitoring:Enabled
Active Processor Cores All
Limit CPUID Maximum Disabled
Execute Disable Bit Enabled Enabled
Intel Virtualization Technology Enabled
Enhanced Intel Speed Step Tech Enabled
Turbo Mode Enabled
CPU C1E Auto
CPU C3 Report Auto
CPU C6 Report Auto

When the CPU ratio is set manually to 45, the turbo locks in at the same.

If I change the CPU Ratio to Auto and manually set the Turbo Ratio By All Cores to 45, 45, 45, 45 I run stable on Prime95 but I bsod during Mark 3D Vantage, usually during the physics test.
September 3, 2011 7:35:09 PM

Watching Tennis, but I'll look over carefully soon. Nothing at first quick glance caught my eye. As mentioned not ALL SB will even 4.5GHz and I've OC'ed enough to know the subtle differences and the 'walls' you can hit...