Bizarre hard-freeze issue with a new build

[Razakel]

I've just built a new system and am experiencing a weird hard-freezing issue. The system will randomly lock up completely for no obvious reason (does so whilst idle or loaded).

With a GTX 460 card in, the display goes completely blank and the monitor goes into standby. When I press reset, the motherboard sounds the "no VGA" beeps. Normally it comes back after a cold reboot, but on one occassion I had to leave it an hour or so before it would come back.

When I tried with a friend's GT 260, the freezes still occur, but the screen remains on, so that rules out the graphics card. The system comes back fine with that card in after pressing reset.

Motherboard: Asus P8P67 (B3 revision)
CPU: Intel i7-2600
RAM: G.Skill RipJawsX 2x4GB 1600MHz (F3-12800CL9D-8GBXL - not listed in Asus' s QVL, but G.Skill state it's compatible)
PSU: Coolermaster Silent Pro Modular 700W

It doesn't seem to be heat-related (CPU core temp. ~60 degrees whilst gaming, motherboard temp ~34 degrees and graphics card ~60 degrees). However, I am just wondering if it may be related to the thermal paste. The CPU temperature hit 90 degrees whilst running Prime95, which does worry me slightly, despite that being within the stated limits.

I'm running it now with one stick of RAM removed (same issue occurred with the other stick removed) to see if that makes any difference, although memtest86+ and Windows Memory Diagnostic both came up clear (edit: no luck with the other stick).

To be honest, I'm completely stumped. Does anyone have any suggestions?

 

[Mark56]

You should be able to find some Critical error in the event viewer under applications which may give a clue or two.

I would also make quite certain that your RAM is running at the correct voltage. The settings will be in your Bios under Peripherals.

Although Prime95 does push the CPU up to 100% that temp of 90ºC is quite high but under normal use if it isn't overheating it should not be causing the freeze up. Arctic Silver is one of the better pastes you can use.

 

[Razakel]

No clues in Event Viewer other than the "unplanned shutdown" event. Everything is running at stock.

I've tried reseating the memory in the two unused memory slots, stable for three hours now (fingers crossed). Unfortunately that would mean it's the memory controller that's knackered, therefore the motherboard... which I've just had bloody replaced after it wouldn't even POST...

EDIT: Nope. Died again. I have absolutely no clue now.

 

[mailpup]

Can you substitute another power supply for testing purposes?

 

[Mark56]

Have you used the manufactures diagnostics on the hard drive.

Identify the make of your hard drive and then use one of the links below to get the manufacturers diagnostic for ISO CD. Burn the image file to a CD, boot the PC with the disc in the drive and run the diagnostics. You first need to set the CD drive to 1st in the boot order in the Bios setup.

If you do not have an image burner use this free software to make the CD.

http://www.isoimageburner.com/


ExcelStor: http://www.excelstor.com/eng/support.php?sub_id=3

Hitachi/IBM: http://www.hitachigst.com/support/downloads/

Samsung: http://www.samsung.com/global/business/hdd/support/downloads/support_in_es.html

Toshiba Fujitsu: http://sdd.toshiba.com/main.aspx?Path=SoftwareUtilities#diagnostic

Seagate, Maxtor & Quantum:http://www.seagate.com/www/en-us/support/downloads

Western Digital:http://support.wdc.com/product/download.asp?lang=en

 

[Razakel]

I'm trying another power supply now.

As for the hard drive diagnostic, the SMART data all looks fine, and the Quick Test passed OK. I'm running the extended test now (needs about two hours). However, I am curious as to why a SATA drive failing would cause the system to lock up completely - isn't it meant to be hot-swappable? Plus the fact that it sounds the "no VGA" beeps after pressing reset means I doubt it's disk-related.

Now, if it is down to the power supply, how can I actually prove it? Open it and stick an ammeter on the 12v rail? Voltages look fine.

 

[Mark56]

Under the circumstances testing the hard drive is just a process or elimination.

Reading back to your first post, when you tried the other card (GT260) you still got a freeze but the screen remained on, with your other card the display went completely blank. Did you load the correct drivers for the GT260 card for testing purposes? The two cards are showing different behaviour so I would not rule out the card.

Have you thought of trying your card in another PC? If that goes OK, then you can rule out the card.

I am curious as to why a SATA drive failing would cause the system to lock up completely - isn't it meant to be hot-swappable?

It does happen but it is more often down to memory or graphics card. Drives can only be hot swapped if using a drive bay as used on servers, hot swap the drive on your PC and it will crash.

 

[Razakel]

Under the circumstances testing the hard drive is just a process or elimination.

Reading back to your first post, when you tried the other card (GT260) you still got a freeze but the screen remained on, with your other card the display went completely blank. Did you load the correct drivers for the GT260 card for testing purposes? The two cards are showing different behaviour so I would not rule out the card.

Have you thought of trying your card in another PC? If that goes OK, then you can rule out the card.

nVidia provide the same driver for both cards. I'm running it with an 8600 in now and my old 425W PSU (edit: crashed again).

I'm not sure I have a system with a PSU of high enough wattage to power the 460. I'm a little reluctant to run it in any of my or my friends' systems, where the PSUs are all <500W.

It does happen but it is more often down to memory or graphics card. Drives can only be hot swapped if using a drive bay as used on servers, hot swap the drive on your PC and it will crash.

Yeah, it's just the blank screen and then the beeps that makes me doubt it's the drive.

 

[Mark56]

Can't you fit your card in the PC the 260 came out of, as that card has a higher wattage the PSU in the PC it came from should handle your card without any problem.

Using another known good PC to try your card in has to be the easiest way to test it.

 

[Razakel]

Can't you fit your card in the PC the 260 came out of, as that card has a higher wattage the PSU in the PC it came from should handle your card without any problem.

Using another known good PC to try your card in has to be the easiest way to test it.

A GT260 takes more power than a GTX460? That's quite surprising, I thought it'd be the opposite.

I'll give the 460 a try in another PC, but, TBH, since the same problem's appeared with two known good graphics cards and one known good PSU, I'm inclined to think it's CPU or motherboard-related.

This is turning into a complete nightmare.

 

[Mark56]

Yes it is a tricky one to diagnose. The GTX 460 is 160W and the GTX 260 is 180W according to Nvidia site.

The thing that concerns me is the PC behaved differently with your card and the known good one. Your assumption may well be correct but you have not yet proved that your card is 100%. Obviously it is your decision but I feel it is best to prove that your card is OK before looking for something else.

 

[Razakel]

Yes it is a tricky one to diagnose. The GTX 460 is 160W and the GTX 260 is 180W according to Nvidia site.

The thing that concerns me is the PC behaved differently with your card and the known good one. Your assumption may well be correct but you have not yet proved that your card is 100%. Obviously it is your decision but I feel it is best to prove that your card is OK before looking for something else.

The other card is actually a GT 260, not a GTX 260. For some reason it's not listed on nVidia's site.

Well, it behaved identically with the two known good graphics cards, it's only with the 460 that the screen goes off. It doing the same thing with three different cards doesn't mean that the card is 100%, but it does mean that the problem is with CPU/mobo/RAM.

So how can I narrow it down here? All I can really think to do is order a new CPU/mobo/RAM, test them and then return them under the cooling-off period once I've found which one's causing the problem.

 

[Mark56]

I thought maybe you had missed the X on the GT260 as I could not find it either, any site with a reference to it displayed its name as GTX in the specs.

Anyway, going back to the test you did on your RAM in your first post I would say it is unlikely to be the RAM unless by some amazing coincidence both sticks have developed faults at the same time. So the conclusion has to be the mobo or CPU, the mobo is the most likely, CPU's very rarely go wrong unless repeatedly overheated and you have run Prime95 which would most likely have reported a fault if there was a problem.

Best of luck.

 

[Razakel]

I thought maybe you had missed the X on the GT260 as I could not find it either, any site with a reference to it displayed its name as GTX in the specs.

Anyway, going back to the test you did on your RAM in your first post I would say it is unlikely to be the RAM unless by some amazing coincidence both sticks have developed faults at the same time. So the conclusion has to be the mobo or CPU, the mobo is the most likely, CPU's very rarely go wrong unless repeatedly overheated and you have run Prime95 which would most likely have reported a fault if there was a problem.

Best of luck.

It's bloody sod's law - I've just had that board replaced because the retailer sent me a totally knackered one that wouldn't even POST, and I had to wait for the Sandy Bridge recall...

Only problem is that I can't actually prove it's the motherboard, which might make returning it a pain.

 

[Mark56]

What can I say, the odds are very high that it is the mobo, they may well have the test equipment, but if not, they will just have to take your word for it. Personally, as they have already supplied one faulty board, I would insist on a refund and go somwhere else.