Computer Crashing

[Vaillant]

Ok, first off, system specs are in my signature.

So, here goes. I've recently been experienceing BSOD's,and now much more recently BIOS POST problems.
System was overclocked, and I ran stress tests to prove stability.

So our air conditionning unit is broken right now, so I pushed up my case fans to medium instead of low, and immediately my problems started. the very next day, my sound driver went corrupt and the system was BSODing all the time. I re-imaged the PC with a ghost image (image is proven stable, everything works like a charm. And if you doubt me, I work as level 1 tech support in the canadian federal government as a coop term, so I know my stuff. I can firmly say its a hardware problem)

After re-imaging the system, it appeared to run smoothly for 2 weeks, untill once again, BSODs arrise. And after recieving 2 BSODs, the BIOS had trouble POSTing. one, continuous beep. The beep code, unless my source is mistaken, says that there is faulty RAM.

I removed my first stick and moved up the other. In addition, I unplugged all devices except human interface. And lowered the fans back to low.

Given my PSU is a 485watt, and the heat, and all the high-end components in my machine, I cannot be sure wether its a RAM or PSU problem. I believe that if the problem lies with either one of them, its either that my PSU isnt powerfull enough for my rig, or that the RAM stick went bad. In which case, I dont understand why I ran smoothly for another 2 weeks, and why the BIOS POST problem was only intermittant.

Please advise : is my PSU powerfull enough. Can it be faulty ram. Any known issues with this kind of setup ?

Thank you.

[minsonngo]

What is the BSOD you are getting?


Download memtest from memtest.org and run a test on each stick. You can download the ISO version... burn it to a CD as an image... and boot from it to run the test. Leave only one stick in at a time when running the test.

Overclocking can corrupt data on the HDD too. That could be your problem... try running Check Disk... here is how: http://support.microsoft.com/kb/315265

[Vaillant]

Yeah, one of my first steps was gonna be the memtest... as for harddrive data corruption, I unfortunately cant give you that one, cause when I had the BSOD in my main windows, I modified the boot sequence to have it boot off an extra partition... its essentially exactly the same XP as my main one, but a clean copy of my last main image.

That being said, the backup wasnt corrupted because I set the BIOS back to defaults prior to booting it.

Now, unless I find problems with my RAM, which im not convinced I will, im still leaning more toward PSU being too weak, cause I do have alot of hardware in here. Can anyone crunch the numbers and tell me if its sufficient? I have a chart of what the values are APPROX. but that chart is quiet old and I'm no hardware pro just yet :)

I'll begin memtesting tonight or tomorrow. With everything back down to normal specs, the devices unplugged and the 1 stick of RAM thigns are looking pretty stable.

[minsonngo]

Here are some good Power Supply calculators that will help you estimate how much you need: http://www.pcmech.com/forum/showthread.php?t=131195#GenPSUCalc




And... if you go to System Properties... click on the Advanced tab... click on the settings button under the Startup and Recovery section... and then UNcheck the "Automatically restart" option under System failure.

If you haven't done so already... this allows you to see the BSOD you are getting so you can copy the error down without the computer just restarting.

[Mr.Ferrari]

I would say the psu is sufficient. If it wasn't, BSODs would not likely occur, rather the PSU would just shut off or your computer would freeze.

Have you tried clearing the CMOS on the motherboard?

Western digital also has a diag tools for your hdd available via their website. Hitachi also has a good diag called "Drive Fitness test" available for use on most hdd brands.

[Vaillant]

Hmm, well I havent experienced ANY data loss or problems of that order in the 4 months of this computer operating (almost non-stop, litterally)

I just flashed the BIOS to the newest version, 2004, because when I tried to flash it last night iwth a USB stick (this time around I plugged my floppy drive in), it got corrupted, and I saved my azz thx to the crashfree bios. ran off an old version to see if system was stable as I stated earlyer.

Right now, under new bios, so that rules out the CMOS clearing problem, I have re-inserted the other RAM stick, and running a blend test with ORTHOS. My friend told me that overclocking Can lead to bios corruption on some rare cases, so thats why I was trying to flash it in the first place...

no errors so far. I will let it run all night. it looks quite stable.

In other news... my sound is doing some wierd stuff... listening to mp3 files, on winamp OR windows media player, it sounds as if a CD was skipping. happens intermittantly every 30 seconds or less often.. just thought I'd add this in. It wasnt doing this before, started doing this yesterday after my BIOS POST beeps.

***Edit

Just did the PSU calculator on eXtreme Power Supply Calculator :

according to my setup, and the calculator of course, it gives me a requirement under max load of 491watts.
my current PSU is 485.
This reinforces my belief that the PSU is inadequate, given that my problems all started when I upped my chasis fan speeds...

[Mr.Ferrari]

You honestly would be fine even with a 450w psu. You can try to replace the psu if you really want, but your symptoms don't point to a psu problem. Although I wouldn't totally lock it out of my mind.

Did you lock your PCI and PCI-E buses while overclocking?

If its not a Ram issue, it looks to me like a mobo issue, if your using onboard sound especially and having these problems. Also try to reinstall your motherboard and chipset drivers if you have not.

If you get another BSOD, remember to right down the error code.

[Vaillant]

Oh, I wrote them all down :)
My buses are locked at all times, overclocked or not. It's just smarter. And I'm starting to think that it may be mobo as well... 1 hour running now under orthos mix test and no sign of trouble.

For the fun of it, I'll give ya the stop codes now (didnt write down the parameters though)

STOP 0x8E
STOP 0x4E PFN_LIST_CORRUPT
STOP 0x19 BAD_POOL_HEADER

these stops mostly lead to saying theres a RAM problem... orthos seems to think otherwise. My plan of action for now is to let orthos all night
then, the following 2 nights, run memtest on each stick.

Right now, I cant definately say its a mobo problem, my sound driver's corrupted before... maybe, Maybe the driver is having some trouble right now, given that I didnt test if this install of my image was perfect. You never know.

Now the reasons I tend to think PSU problems, just so you know, are the following :

firstly, the errors Always happen during bootup, in one step or the other. that means my system is under 100% load ( as is expected while the machine is loading up )

cause the BIOS beep is obviously at a crucial point where more juice might be needed.
further more, all 3 STOP codes occured while entering windows, right before the desktop. again, a stage that requires alot more umph than the usual load.

There's nothing to say that my PSU isnt under its designed rating as well. Furthermore, theres no way to guarantee the power is absolutely clean from that PSU.

I recognize that its a bit too soon to know anything of consequence, that I have more testing to do, dont worry about that im not jumping ship. I just want to let you guys know what im thinking about and why, so that we can work the solution out faster. After all, the best source of info in the tech help Is from the client :)

[Mr.Ferrari]

Orthos with LargeFTTs can concentrate more on the memory.

Also try windows memtest, it is more harder on the ram than memtest in DOS.

I understand what your saying, but the powersupply, if its not able to provide the juice needed, will usually shut off or freeze the computer. OCP, over current protection, will trip in this instance. Though the motherboard and psu could be in conflict. I've seen it happen in the past with Enermax (DFI) and Antec (Asus) particularly.

Since your getting errors dealing with components on or in the motherboard, I had just thought it could be it.

EDIT: Just for your knowledge, the Extreme PSU calculator crunches numbers based on the maximum possible power draw of each and every single device running at its maximum potential. :D

[Vaillant]

Thanks, I was aware of that :)

8 hours of running ORTHOS in standard blend test, successful. Will do windows memtest now.

***Edit

I don't have time to make a long search, but I cant find memtest for windows. Link please?

[Panama Red]

http://www.memtest.org/

[Vaillant]

Yeah I found the DOS version, im asking where the windows-runnable one is. Unless I misunderstood what Mr.Ferrari meant...

[Panama Red]

You don't run memtest in Windows. You run in from bootable media, either a floppy or a CD, depending on your hardware.

[Cricket]

Are you running the computer at stock speeds now or is it still overclocked?

:) Cricket

[Vaillant]

Panama Red - I know memtest is traditionally run under boot media. Mr. Ferari says theres one that runs on windows that stresses the RAM even more. What I'm asking is wether or not that statement is accurate, and if so, the link to that app.

Cricket - Running on everything stock right now. That being said, I tried putting things back to stock specs to fix the problem in the first place, before I started unplugging everything, removing RAM stick and lowering fans. It hadnt done a difference back then.

I'm an hour and a half into an orthos large ttf test, so RAM looks really good. I'll run memtest on it tonight, just for the hell of it. Memtest is still the safest and most guaranteed way to say RAM is good.

[Panama Red]

MrF is probably referring to MS's Windows Memory Diagnostic.

http://oca.microsoft.com/en/windiag.asp

edit: It too runs from disk.

[Vaillant]

Okay lol :P I also found it fishy they had a RAM diagnostic tool that could run under windows.. and somehow be better than the bootable ones

[Mr.Ferrari]

This is the one I was talking about. http://hcidesign.com/memtest/

[Panama Red]

This is the one I was talking about. http://hcidesign.com/memtest/

Looks like the free version will run in Windows but will not check all of your ram. In order to check all of the ram you need the Deluxe floppy or CD package and once again, run from a boot media. At least that's how I read the manual.

[Mr.Ferrari]

Oh yeah very true. Many others, including me, have found errors while overclocking with the basic windows memtest while memtest 86+ came clean.

Just trying to offer any last ditch efforts. :o

[Vaillant]

Ok so back to my initial plan. I'm 4 and a half hours into a large FFT test with orthos, no errors.
I'll run memtest on the system tonight. Seeing how everythings proving to be stable RAM-related, I'll just run it with both sticks in... cause I mean, surviging 12 hours of stress tests without errors, the ram MUST be good, I'll only run memtest as a last check for this thought.

So, with RAM proven good, and by the same time the proc (even if that wasnt really doubted), it leads me back to the forefront of questions.

What I'll do tonight, after letting the memtest run for an hour or 2, I'll try and duplicate the problems. I'll set the overclock back in place, replug all my USB devices. Then, if all proves to still be running, turn up those fans.

I'm more and more convinced it may be my PSU.... cause as soon as I dropped the OC, and fans, and USB devs., I could boot into windows again. So in other words, as soon as I lightened the load on the PSU..

(ok yes it MIGHT be a USB conflict but I dont really see how or why, cause the systems been running just fine for the last 4 months with those USB devs. and the problem only started after I put that lil extra strain on the PSU with the fans.

[Panama Red]

How do you have the fans connected? Are they all on one psu branch or are they connected to the same branch as the hdd? If on the hdd branch, the extra load can cause a voltage drop. Do you have the mobo monitoring software installed so you can watch the temps and voltage numbers as you make changes?

[Vaillant]

Indeed. All my fans run off one of the rails (well may be mistaken cause im not PSU expert) but anyway all 4 fans are chained and attatched to one line of connectors, straight out from the PSU, having both HDD's on their own line, opticals on another

And I have everest ultimate, having it display the temps and other important things onto my G15 LCD display, and of course can go check all the voltages as well thx to everest.

[Vaillant]

12 hours Orthos blend test
13.5 hours Orthos large FFT test
6.5 hours Memtest

all of the tests came back clean. No errors. That puts to rest any possible doubt on my RAM or proc. Now we gotta consider other things.. like the PSU :P

[Mr.Ferrari]

Bear with me here, but have the BSODs been occuring at all? Or do they still happen when you switch back to your old settings?

[Vaillant]

Well, thats exactly the point I'm up to now, Mr. Ferrari..
Once I get home tonight, I will put my computer through the same steps I did previously (the only diff is that it wont happen over a few months I'll do it all now lol)
I'll overclock it back to where it was (also plug USB devices back in), run orthos again for 20 minutes (that was my testing time when I determined where I should cap my OC in the first place)
if still bootable and stable, then push the fans up

At this point, I would Expect it to crash again.. Considering my initial hypothesis is correct. I may just crank out my mulltimeter as well to test the voltages... cause the onboard sensors cant always be trusted -.- just to check if the sensors are Accurate..

I just noticed something else !

looking at my voltages right now, at stock speeds my voltages are all pretty dead-on the mark... but when overclocked, my voltages would drop.. a bit more than what is considered "tolerable" here's an approximate chart to prove my point :

VOLTAGE EXPECTED | STOCK SPEEDS | OVERCLOCKED
+12v | 12.04v | 10.8 (or lower)
+5v | 5.02v | 4.5 (or lower)
+3.3v | +3.3v | (I forgot)
(cpu vcore) 1.25 | 1.2v | (set to 1.2526) 1.2 as well (sensor may be falsed)

but I mean bottom line the voltages were dropping a bit much for comfort... another sign that makes me believe the load on the PSU was too high for its capacity

[minsonngo]

It does look like the PSU based on the voltages you posted. So it runs fine at stock speeds right?

[Panama Red]

It does look like the PSU based on the voltages you posted.

Sure does.;) 10.8 is right at the -10% low end of the 12v rail and 4.5 is also fully 10% below 5v. "I need more power, Scotty!":eek:

[Vaillant]

Yeah! I mean, I had never really taken note of the power drops given the system was stable. But like I've said many times now, right after I did the fans, boom problem city!

So, just for good measure, I Will try and see if I can replicate the problems by boosting my system back to what it was this staretd... I'm not about to go buy a 300$ PSU if im not absolutely sure (I plan on getting the Enermax 850w Galaxy modular PSU)

(yes thats alot more power than I need, but I'll need the umph for crossfire later on etc, and modular will help clean up my case)

I'll let ya know how it goes at the end of the night. :D

[Mr.Ferrari]

I would personally pick the Silverstone ST85F or DA850 over the galaxy. It does worry me that the PSU droops so heavily after only a ~.05v increase.

[Vaillant]

Ok, home...

I'm running under my overclock again, I plugged all devices back in, fans are at Max, and I'm running TWO instances of orthos blend test!

Everything, even the sound, seems to have re-established itself. I simply dont understand anymore.

Either the power supply is really flaky, cause all the voltages are correct according to everest, and backed up by my digital multimeter... or maybe the BIOS had been corrupted by my overclock and flashing it sunday ngiht fixed all my problems.

I could have sworn my voltages were what I put in the chart above before... But scince I never paid THAT much attention to it, im no longer sure about much of anything. I'm at a total loss to understand the problem lol

On the bright side, version 2004 of my BIOS made my CPU fan controller pro-active.... the fan, set to "optimal" mode, actually does speed up when the core gets hot, to keep it cool... contrarily to staying mostly the same RPM and seeing my full load temp go to 64-65, its sticking at 50 now..

Please advise..

[Mr.Ferrari]

There should be a set of q-fan controls under the hardware monitoring section in the bios.

[Vaillant]

Yes, I know that very well Mr. Ferrari
the only q-fan I ever enabled (also cause its the only one attatched to the mobo) is the CPU fan
they fixed a bug in the BIOS for sure, the behaviour of the fan controller totally changed.

Edit : update

Now 1hr40 minutes into orthos test. closed one cause it wasnt processing and clogging the RAM for nothing.
Everything's running as smoothly as if nothing ever happened. I guess maybe it Had been BIOS corruption. it's the only thing I can logically say right now.

If you guys have Any other ideas, please tell me.

[Panama Red]

If you guys have Any other ideas, please tell me.

The oft quoted last resort - Gremlins!:D

[Vaillant]

lol :P

Well, thanks everyone for your insight, and for being my sounding board. For now this topic is coming to a close. I'm gonna run one last Orthos test, with my overclock on, for at least 12 hours to really guarantee its stable again.

Might consider a more powerfull PSU given the calculations that psu calculator came back with.

Once again, thanks. :D

[Vaillant]

One final update :

as predicted, the system is incrdibly stable now. The bios reflash fixed everything, and the updated fan controller code makes that my new load temp is 50c instead of 65 :D

Ran orthos blend test for 20 hours, no problem.

Thanks again everyone, keep up the good fight.

[minsonngo]

Nice. Glad you go it working.

[Panama Red]

Thanx for the final update, Valiant.