Strange hardware behaviour, can you explain it?

bobbytables

Background - July '11, built an i5 2500k rig on a ASRock P67 Pro3 (B3) with 8GB (2x4) G.Skill DDR3 PC12800 & a Crucial M4 SATA6 SSD 64GB. All other components in the machine were not replaced, but working fine or so I thought. Rig worked perfectly for 3 months and then all of a sudden started to hang during OS boot. Would complete POST, hand over to OS (running Ubuntu), but after the point where keyboard and mouse usually get power cycled (during init) disk activity would stop. Tried tonnes of stuff (did the hokey pokey with RAM sticks), and also considered it had to do with the new UEFI/BIOS, but finally the word on the street and my own hunch suggested that it was just a bad motherboard. So motherboard gets replaced....

October '11, same as above except now with an Asus P8P67 (Rev 3.1). After initial build, machine did the exact same thing (hung during boot). Feckit bobby said, and whipped out a stick of RAM and it booted. It kept booting fine with just 1 stick of DDR 3. Put the other stick in, it sometimes worked & would boot like there was nothing wrong. I couldn't detect a pattern. Eventually I enabled XMP in the UEFI/BIOS and now it started working perfectly for another 2 months.

Then and here's the current problem: When I first power on the machine. The power light comes on, the fans whirl up and about 1 second later, the power dies. Lights go out, and the fans stop. About 3 seconds later, it attempts to turn on again by itself, and dies again 1 second later. Then usually on the 3rd attempt it powers up and stays up working fine. That's what happens when I boot the machine cold when it has been off either overnight or for several hours.

If I shutdown the machine and come back E.g. 1hr later and attempt to cold boot again, it just powers up first time without any problem.

Some days when I attempt to power it on, it goes through the few failed attempts to power on, so I go to the wall power socket. I kill the power to the socket for about 60 seconds, and then turn it back on. Then when I go to power on the machine after doing that, it works 99% of the time.

Which component do you think is at fault?

I mean I'm posting this from the machine right now. Once it comes up, the temps are fine, and the machine works flawlessly. Its just what happens right at the start when I hit the power. I can always get the machine finally going within 2 or 3 attempts at most, but it's a strange problem I'd like to get to the bottom of. Any ideas?

Unregistered.

Test with a different PSU for a while.

bobbytables

Hmmm, right now a spare PSU is probably the only component I don't have.

Current Power Supply is a OCZ GameXStream 600W.

Other components are:

• nVidia 8800 GTS 512MB (yes yes, I know, but I don't game anymore)
• 2 x SATA3 Hard Drives, 1 x SATA6 SSD
• PCI Wireless Card

No optical drive, I hate optical storage.

I wonder is that too much on a 600W power supply?, but I imagine if it was, it would have never worked or have I just pushed it to its last legs.

Serephucus

USB drive. :P

My family's PC has a very similar problem. It would essentially cycle-boot. POST, and about a second of Windows logo, then stutdown and repeat. It kept doing this indefinitely however (stopped it after ten minutes).

Turns out it was a USB drive that was doing it. Wasn't even in the boot list or anything, but for some reason the computer just didn't like it in there during boot up.

Probably not the same thing though. Tried single stick of RAM again? If both of the previous times something went wrong you fecked with the RAM and it worked, I'd think it's at fault again.

Genghiz Cohen

RAM I think.
I 'upgraded' the RAM in my dads netbook, running Ubuntu.
Wouldn't POST. Fan spun up for about 2 seconds then nothing.
Turned out the RAM was dodgey. (And I hadn't installed it properly)

bobbytables

There is no USB storage devices connected to any USB ports when this happens. Also I'm not running Windows (haven't in years), so you can rule out any Windows issues here.

With the RAM, both sticks have been installed for several months now. From what was described above it sounded like the machine never POSTed, but from my OP this is not the case. My gut is that it's the power supply, but I don't want to go to the expense of a new one until I can be sure. Testing with another one makes sense but a 600w is the max I've got around here.

So if anyone can offer me more reasons as to why you think it's not the power supply please let me know otherwise just vote Power Supply.

Thanks all

deconduo

Cold boot problems are caused by the PSU 90% of the time. Your setup will pull about 250W on full load and nowhere near that much when just booting, so overloading isn't the problem. It could be that the PSU is just old and on its last legs anyway. Best thing to do is to ask to borrow a friend's one just to test it out.

Thor

I would suggest highly that it's the psu!

If you can't borrow a psu, Try disconnecting every power cable on it and reconnect them, Also try using a different molex/sata power if there is any unused ones, Also make sure that any 8/4 pin cable is in correctly.

It more than likely is the power supply as it's the common dominator in the build, The ram wouldn't cause such issues as it's not that random(forgive the pun) when it comes to cold starts.

Tea_Bag

if its rebooting sometimes before it even posts, the RAM isn't at fault, though I wouldn't rule it out entirely.

some people have also been experiencing problems with EUFI/Ubuntu/SSD combinations.

I still voted PSU. I'd get a loan to rule it out before touching anything else.

bobbytables

Tea_Bag wrote: »

some people have also been experiencing problems with EUFI/Ubuntu/SSD combinations.

I still voted PSU. I'd get a loan to rule it out before touching anything else.

Yes a few months back the ASRock board was the first I'd had which included the UEFI system, which introduced its fair share of issues. However I think I got past most of that.

Like most engineers, I much prefer absolute failure over random failure, because it makes diagnosis that much easier. Once the hardware boot manager hands over to the OS successfully, then as far as I'm concerned we can move on from UEFI issues affecting boot. There was plenty of crap alright about Intel/Other SATA6 connections on some of these new 1155 boards, so I did some port hopping to test that hypothesis.

Right now, either POST doesn't even start or else it completes successfully. Actually today when I powered on the machine at around noon, it just powered up first time, no issues and the machine wouldn't have been on since last night before that. So again, it's almost completely random to me at this stage , which is so frustrating.

Anyway, what I'll try and do is find a PSU that I can install for up to a week (just to be sure) and see if the failure rate drops to zero. Then I'll be more confident in the diagnosis before more cash is spent.

Thanks for the help folks

twinsen

I have the same issue like OP. All I have to do to start it, is just unplug any usb devices. Once done starts fine .

poggy

May not be in the right area but used to have problems with DDR2 when the bios default timings were not the same as the ram was rated at may be worth a look the timings for the RAM you posted are 7-8-7-24-2N.

Some times manufactures "loosen" the timings to get the RAM to meet the required operating frequency.

bobbytables

poggy wrote: »

May not be in the right area but used to have problems with DDR2 when the bios default timings were not the same as the ram was rated at may be worth a look the timings for the RAM you posted are 7-8-7-24-2N.

Some times manufactures "loosen" the timings to get the RAM to meet the required operating frequency.

Cheers, that actually was a problem before with stock settings. Once I changed them, I was back in business. Good advice though

I didn't try plugging everything (all USB devices) out before booting. I can assure you there are no USB storage devices connected, but keyboard/mouse are the only other USB devices hooked up. To be honest if it was a case that the mobo had issues with a keyboard and mouse like that, then I'd be going down the road of firmware upgrades, etc. That sort of thing is just not acceptable. I'd understand if I had USB hubs maxed out with USB powered kit, but that's not the case here.

bobbytables

FYI, well done to all of you that thought it was the power supply. It was!

Finally got around to replacing the PSU the other day and the machine is working flawlessly ever since.

Thank you for your help