Locking up computer - CPU Vcore experiencing massive fluctuation(33%)

[MarkC0]

Hey guys,
I'm currently attempting to fix a locking issue with my newly built computer.

First and foremost I will describe the problem:
- Once every 12-72 hours my computer will lock up, displaying a frozen frame of whatever was last on the monitor.
- The lockups occur randomly; sometimes when nothing is running at all but the OS.
- All functionality is gone.
- Case fans, CPU fan, GPU fan, and mobo are all still powered.
- Computer is still generating heat (overall; haven't been able to tell part by part)
- Anything in a USB slot can still receive power. However anything new plugged into USB cannot.
- I've unplugged the DVI from the video card and plugged it back in a few minutes later, only to see the same frozen frame of my desktop (in this case, Breaking Bad in full screen).
- The only resolution is to hold down the power and restart, then all is fine for 1/2 a day to 3 days.
- On top of this, my direct replacement warranty expires in 4 days.

My specs are as follows:
ASUS P8Z68 V-Pro (MOBO)
Intel i7-2600k (CPU)
Zotac GTX 580 (VGA)
NZXT HALE90 1000W (PSU)
G.Skill 8GB 1600mhz Sniper (RAM)
ASUS PCE N-13 (LAN)
Intel 510 120gb (SSD/BOOT)
Samsung 1TB HE103SJ (HDD)
Cooler Master Hyper 212 (COOL)


Potential culprit 1
- I'm getting an odd voltage reading using my NZXT 1000W PSU. Massive fluctuations. Using Speedfan and HW monitor, the "Vcore" voltages are flying about everywhere. My ASUS board features an "EZ overclock" button in this BIOS so I thought this might be it, but set back to stock the problem still persists.

Overclocked to 4.5ghz:
http://img339.imageshack.us/img339/6217/voltagessketch.png
http://img406.imageshack.us/img406/4834/vcore2.png

Stock at 3.5ghz
http://img263.imageshack.us/img263/4613/speefanvcorenooc.png
http://img171.imageshack.us/img171/4250/vcorenooc.png

I should comment, in the ASUS UEFI (BIOS) window, it doesn't show the voltages fluctuating by more than a couple %. I'm not sure how that fits into the picture here. I do however feel this is the likely culprit to my freezing issue.



Potential culprit 2
The Intel SSD, since SSDs in general simply aren't that reliable (even Intel has close to 10% RMA in first year). It does figure out pretty nicely. I can't physically do anything because I'm not longer connected to a drive containing an OS, letting alone drivers for anything connected. The last image could be stored in RAM, hence why I can still see the last frame when unplug my DVI and plugging it back in. Not sure how it would explain the USB fiasco though.


Potential culprit 3
My video card (Zotac NVIDIA GTX 580). Even though I unplugged the DVI and plugged it back in, only to receive a signal, it could be that the card had that stored in it's RAM. There is nothing that points me in this direction however through testing, is it does above with the CPU/PSU.


Potential culprit 4
My RAM, simply because many people report that locking up is commonly a RAM problem. However I do not quite understand how to use MemTest x64. Is there a simpler program I can use for the same result, preferably through the OS? That or maybe a link on how I can get it to boot off a thumb drive. Thanks in advance.


So there we have it, I know it's a lot to read but I figured I'd be thorough as to not waste anyone's time with back and forths needed to attain info. I am open to suggestions on how to solve this problem as I really am not that savvy when it comes to building or repairing computers. I'm stumped and in dire need to solve this within 4 days, so I can return any faulty parts for direct exchange.

Thank you so much for your time if you have made it this far.

I appreciate any and all help!

 

[essenbe]

Statistically, Intel is the most reliable SSD. To eliminate that possibility, you could clone your OS to a mechanical hard drive and see if that resolves the problem. There is a ram test built into Windows 7, however it is not as reliable as memtest86. I would recommend you to use memtest86. If you want to eliminate the video card, remove it and use the on board video. But, that will relieve some of the stress on the PSU. If the PSU is not putting out enough voltage to run everything, relieving the stress could make it work OK and make you think it was the video card. Perhaps you could borrow a PSU from someone just to test that.

EDIT: given your situation, I don't believe you can test all of the components properly in 4 days.

 

[essenbe]

In looking at your images, Speedfan is not reading your PSU correctly. It appears to be reading your +12V at 6.8V. If that were true, your computer would not boot.

 

[MarkC0]

I checked out the voltages in BIOS and guess what? RAM was set to 1.65V. It is rated at 1.5V as per the label on the RAM sticks. Would this be enough to have caused the issue? I also changed the Vcore from offset to manual (using recommended voltage of 1.31V when OC'd to 4.5ghz) and the whole Vcore voltage jumping around problem went away. It seems it's meant to do that, adapting power based on processor needs I suppose. Do you know if either setting (offset or manual) would be okay for now if I'm just trying to troubleshoot this issue? I'm also not sure what you mean by clone my OS. Do you mean just install it on a different drive or...? Thanks for the reply btw, really appreciate it.

 

[essenbe]

A clone is a 1-1 transfer of the OS. It can be done with several free programs, or you could image your ssd and restore it to a mechanical hard drive, or as you say just do a clean install for a few days.

The voltage is supposed to fluctuate some. Intel speedstep wil downclock the CPU to 1.6 and .09 -1.0V at idol no matter what your OC is. Unless you set the bios to manual. With setting it at manual the CPU will still downclock to 1.6 at idol, but the voltage will not drop, it will stay at what you set it at. That will cause you to use more power and create more heat. If it were me, I would set everything at optimized defaults then only change what is necessary. Set the dram voltage at 1.5 and set your sata controllers at ahci I assume, whatever you had them set to. You have to get the system stable at default settings before you try to OC.

 

[fseal]

Ack, I've had HORRIBLE experiences with the "Auto Overclock" features Turn it all off and leave it off.

Set everything back to normal with the auto overclocking option(s) off and give it a try, then you can manually overclock as you wish, but the auto stuff is just death...

[Edit] I should say that I meant the motherboard specific auto overclocking features (if that is what you were talking about). The standard intel chipset auto V/Clock features have been perfectly stable for me. Some motheboards try to add to that and that's what is full of fail.

As essenbe said, put everything to optimised defaults first and try that...[/edit]

P.S. If your HD light comes on and stays lit during the lockup, you can try turning off "Native Command Queueing" on the SATA interface (in the hardware manager) for the one port that the SSD is connected too. It's a feature for spinning drives that in some systems (my old one but not my new one) can cause the HD drivers to lock up when using an SSD. In fact when I disabled it, not only did the lockups go away but I got more speed out of the SSD as well.

 

[MarkC0]

Thanks for the help so far guys. A few updates and questions.

1) The computer locked up again today. It was overclocked and experienced the same problem as usual, however the RAM was set to proper voltage and timing. Event viewer shows nothing relevant. This time however something odd happened. After I reset the UEFI back to "optimal default" setting my computer would not post. It reset itself (even the PSU turned off) 3-4 times within 20 seconds, then posted. This happened twice before, when I first built the computer and a few days ago. It looked like the LED light was on the graphics card when it would not post this time, right before it would restart. Last time however it was on the RAM (voltages were set too high back then though). Does this recent restart suggest a PSU problem or a VGA problem (if any)?

2) I'm a little confused as to how to reset my components to true stock settings. The only option I can find in the UEFI is "reset to optimal defaults" which doesn't seem right. Maybe I'm missing something.

Here are the options I have to work with using the P8Z68V-Pro:
http://www.youtube.com/watch?v=3vBN-cBPs98
8:06 shows the UEFI option I have with the board
13:45 shows the OS options I have with the board

Is there a way I can as well access normal BIOS or would this suffice?

3) As of now I have done the following changes since the last lock up:
- Reset back to "optimal defaults" in the UEFI.
- Set RAM back to 1.5V after reset
- Tested my RAM twice in MemTest and passed both times.
- Disabled the TPU and EPU switches on my mobo (TPU is auto-overclock, EPU is for power saving).
- Disabled Turbo Mode
- Set Duty Control to "Extreme"
- Disabled "C3 and C6 reporting to OS" in the "CPU" tab in the Advanced section of the UEFI
Are there any other changes I should be considering before putting it through the next "test run"?

4) Do SSDs get hot on the outside casing? I did not check for heat on there. If it cools down after a lock up would that make it a suspect?

- EDIT: I should note: Two lock ups ago, I was checking my temperatures in various sensing programs and the GPU was stagnant at 59 degrees celcius despite having no load. No matter what I would run (or lack there of), it wouldn't budge more than hitting 60 for a few seconds here and there. 30 minutes later the computer locked up while playing Oblivion.

 

[essenbe]

Here is what I suggest. Set the bios to oprtimized defaults, set the dram voltage, timings and frequency at whatever the sticks are. Storage controllers should be set to whatever they were on install. With an SSD I would assume it was ahci. Save and exit. If it boots into windows, monitor CPU and GPU temps and run it just like that. The idol temps of the GPU seems high to me, but not to the point of dangerous. Some GPU's run hotter than others. Run it like that for a while and see if you ger any lockups. I noticed you said it was overclocked. Do not overclock it. leave it running at defaults until we get it stable. Before you can think about overclocking, the computer must be stable at stock. Overclocking leads to instability, so you have to have a perfectly stable platform at stock before you try to overclock.

 

[MarkC0]

Okay thanks. I have reset to optimal defaults and set the RAM. I was told to disable the following items however:
Here is what I have disabled so far after resetting to optimal defaults:
- TPU switch on mobo (for auto-overclocking)
- EPU switch on mobo (for power saving)
- Turbo mode (UEFI)
- CPU C3 and C6 reporting to OS (UEFI; I'm in offset mode +)
- CPU Speed spectrum
- Change duty Control/Line-In Calibration up to Extreme/Ultra High respectively as well
- Intel Speedstep

What do you think?

 

[essenbe]

I'm not familar with Asus BIoS settings, but Intel Speedstep is a good thing, the C states have to do with sleep. turbo mode is what you bought the K chip for, even the non K chips use that. I don't have the same settings for load line calibration, but as I understand it, on Asus boards that is the setting you use for high overclocks.

Who told you to set those settings? I may be wrong in what I am saying, but I'm sure someone with an Asus board will respond soon.

 

[GeneO]

Go back to square one- don;t overclock and set the computer up as such making sure you are using a proper memory voltage for the memory - running 1.5v memory at 1.65v will surely not work. Then if you want to overclock, read up on it and as mentioned - don't use auto overclocking. Most voltages are OK left on their auto setting, but some are not - especially memory voltage and Vcore. I was using ta 1600MHz profile from my memory which specified 1.6v, had the memory voltage set to auto, and my MB set it to 1.7v! For Vcore use a voltage offset - that will increment it above or below the VID for the chip and is the best way to manage Vcore IMO. You also need to do overclock incrementally - up it a bit, then run tests for quite some time, up it again if the test succeeds otherwise you need to do some tweaking, etc. It is a slow process, especially if you are new to it.

 

[essenbe]

GeneO, he was trying to get it stable at stock first. He was told to make those settings in his previous post to run at stock. They don't look right to me. I don't have an Asus board and some of the settings are a little different on mine. Look at his settings in his last post and see if that looks right for a default setting with no OC. Thanks

 

[GeneO]

Those settings are all fine, though it should not be necessary to turn them off if running at stock speeds and voltages. It is often wise to turn at least some of those those off if you are overclocking.

As you pointed out, what is important is he test the system at stock multiplier, stock BLCK (processor frequency), stock Vcore, a valid memory profile and memory voltage. The best and quickest way to test for this level of instability it is to run prime95 with it set to check for roundoff errors with it set to use 75% of the memory.

Cheers

 

[MarkC0]

Alright, thanks. I'm trying to run TRULY stock as to get rid of any ASUS overclocking as a culprit. I do think optimized defaults includes an auto-overclock (Turbo mode) as it states it's turbo target is 3800mhz. Who knows what voltages and such they use, could this not be causing the locking up?

Here is an updated list of what I have changed so far after resetting to "optimal defaults":
- Disabled TPU switch on mobo (for auto-overclocking; was off stock)
- Disabled EPU switch on mobo (for power saving; was off stock)
- Disabled Turbo mode (UEFI)
- Changed duty Control/Line-In Calibration up to Extreme/Ultra High respectively as well
- Changed RAM voltage/timings

There is also something called "Ai overclock tuner" which I left at "Auto". The other options were "manual" and "XMP", both of which controlled the BLCK/CPU ratio and the latter which also controlled the RAM profile selection. Now I will take a guess that with turbo mode turned off, this doesn't exactly function anymore anyways. Should I leave this as AUTO or manually set it to 100% ratio?

 

[MarkC0]

Had another lock up even with the overclocking completely disabled as described above.

This time something to note: After I rebooted I got a graphical glitch twice that I have gotten a few times before. After it boots past UEFI/BIOS a pixelated image occurs. The pixels were orange, white, green, and one other color, randomly dispersed almost like a messy game of tetris. Interestingly enough, This error occured when booting off the SSD, and also when booting off the thumb stick (was trying to run MemTest)

 

[essenbe]

The pixelation sounds more like the GPU or the drivers.

 

[MarkC0]

That is my fault for not mentioning it happened before (forgot until it happened again twice today).

I am using Intel HD graphics now (testing graphics card). I am not using Virtu as I don't need multiple GPU support, however I can hear my video card is still running (fan atleast). Is this normal?
- If this still locks up I will be cloning my OS onto an HDD (testing the SSD).
- If that locks up, I will be trying this HDD on a SATA II port (test SATA III).
- If that doesn't work, I will be taking the computer back to the retailer for them to deal with . Is that a good plan?

 

[essenbe]

MarkCO, I realize it is a PITA, but if you think it could be the graphics card, to test it I think I would take it out of the machine.

 

[MarkC0]

You are right, I will get on that. Thanks!

 

[MarkC0]

I returned the video card and have a Radeon 6770 to test in the meantime. The pixelation I spoke of on boot is STILL HAPPENING, but only after I exit the UEFI; it never happens when I boot straight past into the OS. This may as well have been the case with the GTX 580 and this pixelation on boot may be a separate problem altogether. Are errors after exiting UEFI on ASUS boards common? (Happens when booting to different drives as well)