Solved Multiple cold boot BSOD's, PSU related?

[a15995]

Hello!

I'm starting a new thread here, don't want to spam others.

Everyday when I turn my computer on, I get a series of BSOD's which causes my system to become unbootable and/or corrupting my raid disk array with data.

After an hour spent repairing the array, repairing the boot disk etc. the system becomes completely stabile. I mean rock stabile, I can throw anything at it and not being able to cause a crash.

This behaviour seems odd to me and analyzing the crashdumps points in all directions leading me to some kind of hardware issue. I know my PSU is somewhat too small and probably doesn't even deliver 500 watts anymore but I just want to be certain. I know the boot sequence draws a lot of power...

Has anyone else experienced this cold boot issue and solved it with a larger PSU?

I'm REALLY getting tired of starting my day with fixing the pc and at some point I guess I can't fix it easily anymore. Data is of course backed up but anyway...

I was thinking about reducing the boot sequence to only using 1 CPU but what are the chances this will do the trick?

For the record, I tried many things. At first I updated all my drivers etc. and tried a verifier.exe cycle which only sent me into an endless loop of BSOD's I couldn't get out of (no F8/safe mode). Then I thought it might be because of my pagefile being on the raid array. Then I removed the "lock files in memory" option of the gpedit. I have run numerous memory tests without errors and prime95 on the CPU. Videocard also checks out...

All these things have helped but I can't get rid of the cold boot BSOD's and hope someone has tried something similar and can throw in some advice.

Greatly appreciated...

Thanks,
Søren

 

[Vir Gnarus]

Can you provide us a JCGriff Report with crashdumps attached? Thanks.

As for Driver Verifier causing a BSOD loop, that's actually a good thing. It means it was detecting a bad driver that was loaded at Windows startup and was BSODing the system as a result. Again, send us all your crashdumps and the JCGriff BSOD Report and that should give us a good bit of detail to start off with.

 

[a15995]

Files zipped and attached. I don't think the dumps will give much info though...
/Søren

 

[a15995]

One processor...

Only one crash this morning running on only one processor at startup. Might be on to something here, but general as well as crashdump input still welcome...

/Søren

 

[a15995]

Update:

Have tried running memtestx86+ V4.20 at cold boot which on the first two boots produced massive memory errors (100K+) and freezing. Third boot produced no errors and I booted into Windows with no BSOD's so far. This way I avoid a hard crash and possible hard disk corruption.

My thesis is that on cold boot some or all of the memory gets too little voltage so when memtest writes to a part of the memory the data gets corrupted.

My questions is then: is this a typical sign of having to replace the PSU? I mean, when booting a pre-os application the demand on the PSU can't be that high right? Maybe it's just a single rail being defective - or will I never be able to deliver enough voltage to all four RAM slots?

It's definitely a RAM issue but again, the behaviour is odd since it's not persistent...

ANY experiences with this behaviour appreciated!

Thanks,
Søren

 

[Vir Gnarus]

What brand and model is the PSU? This can be typical behavior of failed PSUs, especially of cheaper generic brands and lower quality models. Motherboard can also be attributed to this.

Since it's evident we aren't dealing with software failure but hardware, and that hardware happens to be potentially the dreaded Trio of Trouble (PSU, Mobo, CPU), then your best bet is to simply start swapping hardware with replacements, starting with the PSU. There are no hardware tests for the motherboard and PSU that exist aside from using a multimeter and a motherboard diagnostic kit.

 

[a15995]

NorthQ...

What brand and model is the PSU? This can be typical behavior of failed PSUs, especially of cheaper generic brands and lower quality models. Motherboard can also be attributed to this.

It's a NorthQ NQ-4775-500BU2 (NorthQ : Products : NorthQ 4000 Silent PSU)

It's an "old" one, probably from 2008, so just as probably not delivering 500 watts anymore if ever. It should be an ok brand but about a year ago it started sounding like a dentist drill (which was because the moisture in the room was to high, solved by moving the pc elsewhere).

I know for a fact that I'm running on the border with this PSU since my system at full load requires more than 500 watts - closer to 600, but recent behaviour does NOT happen at full load, rather the opposite...

Anyway, if this sounds like the problem here, I'll be looking for a new PSU with standard dimension 150*86*140mm(W*H*L) - there's no room for a longer one, so it's kind of hard to find one with enough wattage. Any suggestions?

Thanks again,
Søren

 

[Vir Gnarus]

That is a big problem for sure. Not only would constant use age the components inside and whittle down the wattage output, but also any time you bring the PC above 80% load you are exerting extreme pressure onto the PSU's internals and they will drastically suffer as a result. While it can be normal and safe to have spikes that go beyond 80%, what you're stating is you've had it for long periods of time where it was above such load, in which case that will put a big damper on things.

Yes, it is getting more and more evident the PSU here has been abused beyond its capacity through aging and load. Time to get her replaced.

As for a replacement, I've often recommended this article as to what to look for and some recommendations. Never good to skimp on price for a PSU when your PC's life is on the line. As for whether any of those are standard dimensions, I personally know some of the SeaSonics are. Again, best to check each and determine their size from there. As for what wattage to look for, it's always best to go around 100W above your anticipated load in order to prevent it from reaching 80% load as well as give it leeway as it starts to degrade with age. You can use some wattage calcs out there like this one that you can use to get a baseline to work with.

 

[a15995]

Thanks Vir Gnarus!

That's a very good article indeed. In particular I noticed this: "A recent fact reported by Anandtech in their review research is that the majority of damaged RAM returned to memory manufacturers is destroyed or damaged by fluctuations in voltage from the PSU unit."

My system is closest to "The Power Desktop Computer Class" with no option for SLI. I have found a 650 watts 80 Bronze PSU with the correct dimensions but with the 80% load in mind that would be cutting it a bit close (80%=520W, calculator returned 483W)...

Thanks again for your input, I think I'm all set here.

/Søren

 

[Vir Gnarus]

All right mate. If willing, come back here after you install a new PSU and if your problems are resolved (hopefully the PSU didn't fry anything else) then mark this as solved for future reference. Thanks a mil and hope this plays out well for ya!