computer problems

douchebagatron

<div class="IPBDescription">please help! anyone!</div>I've been having problems with my computer hanging. completely freezing at random times, whether i'm using it or not. I was told it might be hard drive failure, so i got a SMART log viewer (hdtune) and examined my hard drive, and found it failed to pass several tests, so I replaced it. worked great for a day or 2, and now it happens again on a brand new hard drive that does not fail these tests.

I've reseated everything in my case except my cpu since it was rather difficult to put on. I cannot recreate these freezes, they just happen randomly. all my temperatures seem to be completely fine at the time of the freeze.

everything in my computer is under a year old. What else could it be?

does anyone have any ideas? or any programs they use to check verify that different components are working?

I have:
core i7 920 processor
radeon hd5770 video card
Gigabyte motherboard
Crucial brand ram (3 gigs)

The problem started when I moved across town, though my computer was in the seat of my car and secured the whole time.

please help. any suggestion, comment or idea is welcome.

Chris0132

Well if it has all the symptoms of a hard drive, you change the hard drive, it starts working, then fails again, either your computer is burning out hard drives really fast, or there may be something wrong with the connectors used to interface the drive with the rest of the system.

If the drive-motherboard connection is broken that may be giving you the errors, and unplugging/reinserting the drive could temporarily fix it if it's a faulty connection.

Otherwise, check all the other components, memtest the ram and try switching out the video card for another one or just disable it (think you can run the computer without a graphics card) and see if it still gives you the problem.

The only way to be sure with computers is to remove every component and test it individually in another system that you know works properly, when that system exhibits the same symptoms, you know which component is broken.

If you strapped the computer to the car then it would be jolted around a lot, cars bounce around a lot when you drive, you just don't notice it much because your body is designed to stabilise your head, but it would probably damage a computer if you drove around with it strapped to the car. Really you should disassemble it and transport the components separately.

Talesin

Disassemble the machine, clean the parts and put it back together. I've seen all sorts of odd results from vibration through shipping.. RAM that unseats itself just a little is the worst, as it results in erratic errors. One of the PSUs managed to get its voltage switch slightly slid over to 240... ended up frying itself. Pulling it apart and making sure EVERYTHING is seated right is the easiest fix for this. Generally I don't bother taking out the CPU/cooler, but that's me being lazy and not wanting to re-apply thermal grease/paste.


Then there's the vibration-damaged. If you aren't using the proper standoffs over the ENTIRE motherboard, and tied down properly, the whole thing can flex and slightly (or more than slightly) break the traces inside the motherboard PCB. Increased resistance, flaky operation, and worse. Can happen with standoffs put in properly if you have too heavy a CPU coooler, or transported it in any direction other than laying flat, motherboard side down (to prevent shearing from the cooler's weight, if you didn't pull the cooler off in the first place).


Or it could just be your PSU deciding to go bad. Grab a motherboard monitoring suite and graph the voltages, logging the values. If you see a major drop just before a reboot, then there's your problem. Could have happened during the move, vibration damage or just bad luck. PSU is always the place I start with a diagnostic on erratic behaviour, if taking it apart and putting it back together again doesn't work.

Thaldarin

<!--quoteo(post=1782345:date=Jul 24 2010, 04:12 AM:name=Talesin)--><div class='quotetop'>QUOTE (Talesin @ Jul 24 2010, 04:12 AM) <a href="index.php?act=findpost&pid=1782345"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Disassemble the machine, clean the parts and put it back together.<!--QuoteEnd--></div><!--QuoteEEnd-->


<!--quoteo(post=1782329:date=Jul 24 2010, 02:35 AM:name=6john)--><div class='quotetop'>QUOTE (6john @ Jul 24 2010, 02:35 AM) <a href="index.php?act=findpost&pid=1782329"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->I've reseated everything in my case except my cpu since it was rather difficult to put on.<!--QuoteEnd--></div><!--QuoteEEnd-->

Talesin

There's a significant difference between disassembling and cleaning, and simply reseating. Also, to some, reseating is just pushing the components down to make sure they're not poking up... which can still leave them cocked enough to either not be making full, solid contact, or worse yet even contacting two (or more) contact pads. Fully removing and reinserting the component also will tend to scrape off a bit of any corrosion on the contacts, leading to a more solid connection... which is easy to do a more complete job on components with a Pink Pearl eraser, but much harder to do on the contacts inside the slots... usually the friction is enough.

Also, disassembling <b><i>and cleaning</b></i> is important. Dust is conductive, and even a little bit in the wrong place can ruin your whole day; I had a DIMM I could have sworn was going bad erratically, until I used compressed air to puff the slot clean. No problems after that.

douchebagatron

I've done the whole dissassemble thing, everything is less than 4 months old so there wasn't much dust on it but i cleaned it off anyway. I've also run a memtest (3 times) and haven't had any problems. I've attempted to run the OCCT test (friend recommended it) on the cpu a few times and it seems to always freeze up when I do. This pins the problem on the processor, correct? If that's the case I can replace it as it's still under warranty.

To be sure, what would be a good program to monitor my voltages as you were recommending?

and is there any way to test the motherboard specifically, instead of the processor/motherboard combination? I'd hate to go through the trouble of replacing the cpu and find the problem is still there due to the motherboard.

spellman23

I assume you've also done memtest and the rest?

Thaldarin

<!--quoteo(post=1782642:date=Jul 25 2010, 12:03 AM:name=spellman23)--><div class='quotetop'>QUOTE (spellman23 @ Jul 25 2010, 12:03 AM) <a href="index.php?act=findpost&pid=1782642"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->I assume you've also done memtest and the rest?<!--QuoteEnd--></div><!--QuoteEEnd-->

<!--quoteo(post=1782516:date=Jul 24 2010, 07:43 PM:name=6john)--><div class='quotetop'>QUOTE (6john @ Jul 24 2010, 07:43 PM) <a href="index.php?act=findpost&pid=1782516"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->I've also run a memtest (3 times) and haven't had any problems.<!--QuoteEnd--></div><!--QuoteEEnd-->

Kouji_San

try running <a href="http://numbers.computation.free.fr/Constants/PiProgram/pifast.html" target="_blank">pifast</a>... and see if it crashes, ifso your processor is borked or overclocked/haswrong bios settings


This one has a batchfile to run, so yadon't have to set it up yourself:
<a href="http://pifast.hexus.net/instructions.htm" target="_blank">http://pifast.hexus.net/instructions.htm</a>

douchebagatron

thanks kouji. i can run pifast fine without any problems. i have an i7 which is supposed to manage clocking on its own, underclocking when not in use and overclocking up to a point when necessary. it seems to only overclock by up to 6.3% (from 2.66ghz to 2.833ghz) but i'm not sure if this is a very good benchmarking tool for this processor since it has 4 cores and this program appears to run only single threaded, only being able to make use of a single core.

I've tried running the OCCT test a few more times and I haven't been able to reproduce the freezing on it, but that is because the tests stop themselves after a minute saying the processor is too hot when it goes over 80C, but from what I understand the i7s are supposed to be fine up to 95c, and restart the computer at 90c.

here's some more info on my temperatures. I idle around 55-60c, pifast got me up to 80-85c. i'm going to run prime95 and see what that does.

spellman23

Idle 55C? That seems bit high, but I don't know if that's normal for a i7

douchebagatron

i just ran prime95, sending all my cores up to 100% and got all my core temps to around 95c, sometimes even touching 100c but never going over. comp didn't freeze at all. any new ideas?
edit: even with the side panel off and an extra fan blowing on my comp i still reach these temps

a side note: since i moved i've also had problems with my wireless going in and out, and sometimes working but very lossy. I figured it was coincidental and caused by the antenna, but could it be possible that the pci card is causing it?

a side side note: my comp froze 6 times yesterday, and hasn't froze a single time today.

Talesin

I'm running an i7, and that's quite high, tempwise, for idle. Mine idles at around 40-45C. Hell, I don't ever see over 68C under full load with Prime95, with a <i>stock</i> heatsink/fan. Time to take off the cooler and re-grease, sounds like. Or get a better cooler. Or make sure it's on all the way... I've seen that before, where it hadn't *COMPLETELY* seated, so there was a millimeter or so of airgap.. toasted the CPU pretty damned quick.

Kouji_San

And keep an eye on airflow in the case, so the front intake cooler airflow isn't blocked by cables and stuff as well.

Obraxis

That's a VERY high temp for i7 chip. You should be MAX about 65-70oC at load, between 30-50oC idle.

I wonder if your heatsink/fan needs looked at or the thermal paste needs re-applied.

douchebagatron

reapplied thermal paste, idling at 45c and getting 85c on prime95, big improvement. we'll see if this solves the problem.

douchebagatron

update: not fixed. froze again.

Obraxis

Could be PSU. They're cheap enough to buy a spare, doesn't even have to be a decent brand just to test, and if it doesn't work, send it back or keep as a spare.

Kouji_San

Indeed, PSU are weird. They do not give you any error signs, but they sure make your pc do funky stuff if they are actually broken. Mine did the last two months of it's life (it died last week, got a new one and all probs are gone...)

douchebagatron

due to the fact that i've moved into a new house, could it be caused by power fluxuations in the house? or would the PSU fix any of those?

Kouji_San

Current PSU have built in protection against that, but nothing is bulletproof...

lolfighter

The PSU irons out minor power fluctuations through a capacitor buffer, but major fluctuations could cause undesired behaviour. Unless you are aware of major power fluctuations, not your best bet for a solution if you ask me.

Cereal_KillR

You need to use a surge protector. Even if power fluctuations are not the reason for your problems, surge protectors are one of those cheap things you will eventually regret not having bought earlier if you ever happen to fry your PC.

Thaldarin

You can get cheap devices to monitor the voltages on PSU's to see if they're varying.

douchebagatron

I've been monitoring my voltages through Speedfan and have some numbers for everyone.

Vcore1: .93V
Vcore2: 1.52V
+3.3V: 3.34V
+5V: 5.05V
+12V: 1.22V
-12V: -1.17V
-5V: -8.12V
+5V: 3.6V
Vbat: 3.25V

These all look fine to me except for the +5 and -5 that are registering 3.6 and -8.12 respectively.

I'm also monitoring these over time, they generally stay pretty steady. i'll see what happens during a freeze.

The only problem with monitoring these is that the voltages are polled every 12 seconds through speedfan, so i'm running multiple instances of speedfan to get a better idea of what the voltages do at the moment of freezing. If anyone has a better method of watching my voltages, please tell me.

Thaldarin

Multimeter and an article like this may help test: <a href="http://articles.techrepublic.com.com/5100-10878_11-1056686.html" target="_blank">http://articles.techrepublic.com.com/5100-...11-1056686.html</a>

Spooge

Not all power supplies are created equal. What is the brand and model of your PSU?

douchebagatron

<a href="http://www.newegg.com/Product/Product.aspx?Item=N82E16817153023" target="_blank">Thermaltake TR2 W0070RUC 430W ATX12V</a>

also, i just got back from best buy where they tested my psu to make sure it output all the right voltages, and it did. no problems at all.

they suggested it could be the motherboard because the motherboard is what specifies certain voltages, which would explain why it showed as +3 and -8 instead of +5 and -5 on my comp. anyone have an motherboard diagnostic utilities to share with me?


I have a <a href="http://www.newegg.com/Product/Product.aspx?Item=N82E16813128423" target="_blank">GIGABYTE GA-X58A-UD3R</a> motherboard.

Talesin

<!--quoteo(post=1783273:date=Jul 25 2010, 08:09 PM:name=6john)--><div class='quotetop'>QUOTE (6john @ Jul 25 2010, 08:09 PM) <a href="index.php?act=findpost&pid=1783273"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->...
+12V: 1.22V
-12V: -1.17V
-5V: -8.12V
+5V: 3.6V
...<!--QuoteEnd--></div><!--QuoteEEnd-->
The 12V rail looks right to you, at 1.2 and -1.1V? The 5V rail only compounds it. It's your PSU. It is toast. It is not regulating voltages properly; most likely one of the transformers is getting ready to die. Replace it now; waiting until it pops could power spike your system and fry some of the more delicate bits when it goes, if it isn't already damaging them with the fluctuations at the moment.

And don't trust Best Buy; they plug the PSU into a $5 PSU tester and make sure that all the lights come on. It's just testing to make sure one of the rails hasn't <b>completely</b> failed yet. They don't actually LOAD test it, as those testers are ridiculously expensive, even if they're just massive tunable resistor banks and heat sinks... and under load is where components start to fail fastest. I'd trust the results you're getting from your computer.
A better way to test (minus the load of your video card) or at least with a faster polling rate is to go into your BIOS. Usually there's a health monitoring page there that updates about once a second.

If you're hardcore about getting exact results and/or ruling out the motherboard as a culprit, grab a multimeter and a pair of pin probes. Check a spare molex connector, or slip the probes into the backside of a connected one, if they're all in use. Make sure to check each individual lead if possible, if it's a split-rail PSU.

Obraxis

Winnar is me \o/

Grab a new PSU, that one is toast.

Spooge

There's always a temptation for builders to spend extra on the CPU, GPU, and RAM and cut back on "lesser" components like the PSU. The lessen from this should be that the PSU is at least as critical as the other components if not more.
Brand is important also. Cheap companies use cheap components. Look for Antec, Corsair, or OCZ for midrange PCs or PC Power & Cooling and Silverstone if you're building a higher end rig. I'd suggest something in the 550W-650W range depending on how many drives you have.