How can you tell if a computer problem is a hardware or software problem?

November 4, 2006 4:54 PM

On my computer, sometimes I can't load Windows. Sometimes I can and it freezes. Can intermittent problems like this be due to the power supply?

I'm having problems starting Windows on my computer and don't know if it's a hardware problem or software. When I power on, the processor runs through the self check properly. It goes to the loading windows screen. At this point, sometimes the entire computer restarts. Sometimes it goes to the logon screen and freezes. Sometimes I can get into Windows, but when I try to perform tasks like running Internet Explorer, the computer freezes.

I was wondering if the intermittent behavior of the problems implies a hardware problem. I was also wondering if anyone else has had this problem and found a solution. If not, is there a way to find out what the problem is without taking the computer to a repair shop?

I've tried searching on Google and Metafilter for items like "intermittent freezeing," but I think my search is too broad.

If this question is inappropriate, is there a better way to phrase this, or any pertinent information that I should add?

Thanks!

computer is 5 years old

cpu: amd athlon xp 2000 +/266
hd: 80gb Western Digital 7200rpm
video card: vga leadtek geforce3 ti 200 64mb
motherboard: epox 8khal
power supply: allied 300 watt power supply

posted by secret.osha to computers & internet (15 comments total)

You might have better luck Googling intermittent freezing. ;)

In any event, this sounds like a classic overheating problem to me. Every time I've had a PC that couldn't go more than a couple of minutes without spontaneously rebooting, it's been a problem with the CPU fan (which cost about $15-$30 to replace, and which is definitely a do-it-yourself job with about half an hour of research).
posted by Doofus Magoo at 5:21 PM on November 4, 2006

Yes, bad power supplies can cause that symptom. Unfortunately, however, it's not the only possible cause. It could be RAM, the motherboard, or a failed fan as well. It sounds hardware-ish, but you can't completely rule out software either.

First thing I'd check: make sure all your fans are spinning. If you can borrow a beefier power supply from another computer, try that too. Try swapping the positions of your RAM sticks as well; if the symptoms change, try running with just one and see if that helps. (some boards won't run with just one memory stick. If it won't boot up with only one, don't panic.)

Beyond that, I probably wouldn't spend much time or effort trying to fix the machine. You can get brand-new PCs for like $300-$400 that will be substantially faster. Your computer isn't that slow, and it's perfectly fine for routine use, but if it's giving you trouble, it's probably most sensible to just replace it. Diagnostic labor is expensive. You'd be better off sinking that money into new hardware, instead of tech time.
posted by Malor at 5:24 PM on November 4, 2006

I'm having very similar problems with my own computer, and it's not even two years old yet. I even Asked MeFi a while back and the consensus is that it was the hard drive nearing death, but nothing I've tested indicates that is the problem.

Personally, I believe my computer is overheating, as it loads up and works fine for 15 minutes or so, then locks and will not load again for about an hour or so (enough time to cool down?). Then it'll work again for 15 minutes, freeze, etc. Here's the catch though: it only did this in extremes during the hot summer months when it's warmer in my home. Now that we're headed for winter and things are cooler, it doesn't freeze quite so much. In fact, as of right now it's been turned on and working fine for over a day.

So I'm still living with the problem, but I haven't been able to fix it. I sent the computer back to Gateway under warranty for repair and they only spent 2 minutes looking at it (seriously; two minutes! The "Repair Report" that came back with the PC listed the "check-in" time for repair and the "send back to shipping for return" times as being only two minutes apart) and then sent it back even more broken than it already was (it came home with the front face plate ripped off, leaving the dangling power switch and two dangling wires with status LED lights on the end). Then the DHL delivery guy left the PC box outside my apartment door in the rain, but that's another complaint altogether.

Long story not-so-short, I haven't fixed the problem. But maybe my frustrations can help you eliminate a few possibilities with your own PC problem.
posted by Servo5678 at 5:25 PM on November 4, 2006

When you Google, search in Groups, rather than Web.
posted by weapons-grade pandemonium at 5:36 PM on November 4, 2006

If you suspect that your issues are temperature related, pop your case open, boot up the system and try working like that for a while. If you've got a small room fan you can point at the case so much the better.

It sounds silly but if the problem clears up under those conditions, you know what you're dealing with. To add to what Malor said about your checking that your fans are spinning, make sure they're not covered in thick dust...you might just be surprised how filthy they can get (and how slowly they'll spin as a result).

If it comes down to it, a replacement or additional fan is cheap and relatively easy to install in most cases.
posted by JaredSeth at 6:08 PM on November 4, 2006

Have you been upgrading it, or is that list the original configuration?

A 300W supply seems a bit wimpy for what you've got listed there.
posted by Steven C. Den Beste at 6:10 PM on November 4, 2006

Yep, sounds like overheating. I remember build two identical computers and was puzzled why one of them kept crashing, freezing, etc. I had put one of them on a table while the other was sitting on top of barely a year old plush carpet. Once I moved the crashing one off the carpet, it worked fine.

Read what the bluescreen says. If the "following file" or reason ("PAGE_FAULT_IN_NONPAGED_AREA" in this example). If they vary, your computer is probably overheating.
posted by easyasy3k at 6:21 PM on November 4, 2006

Power supply sounds like it might be an issue here. I also had some similar problems a few weeks ago, and discovered that the problem was I had stuffed up installing Windows XP (some files were missing). Might be something else to consider.
posted by Effigy2000 at 7:53 PM on November 4, 2006

The easiest test to see if it's HW/SW is to use a CD to boot a different operating system(Damn Small Linux is a good choice) and see if the machine still locks up.

Memtest86 is a good diagnostic tool for finding failing memory. -

If you have a can of compressed air/coolant handy, you can spray down one chip, reboot, wait for another crash, then repeat. The one you spray that lets the system stay up siginifigantly longer is probably the bad one.
posted by Orb2069 at 7:59 PM on November 4, 2006

BTW, the comment that your power supply is underpowered was misplaced. A good 300w power supply would have NO problem with that hardware.

That said, it may not be a good 300w power supply. I've never heard of Allied, which tends to make me think it's a bottom-of-the-barrel supply.

The cheapos will often work fine at first, and gradually (or suddenly) start failing. Of all the components in your computer, it's one of the most important. But, because it's not usually immediately obvious that one sucks, it's a spot where many manufacturers cut corners.

I know four brands, just off the top of my head, that are at least decent. Antec used to be very good, but I was reading about problems with them earlier this year. Probably okay again by now. Enermax is a premium brand. Sparkle/Fortron is a cheap brand that's claimed to actually be fairly good, despite the low price.

If you want the ne plus ultra of power supplies, get one from PC Power and Cooling. They're usually pretty loud, but are incredibly robust. For your hardware, it would be silly to spend $150 on a power supply, but for those who have more expensive equipment, PCPAC can be a very good investment.
posted by Malor at 10:31 PM on November 4, 2006

Something else you can try is running something other than Windows and seeing if that dies a similar death. Memtest86+ is good for this; it's a completely standalone test (no operating system at all) and, if it doesn't freeze and crash, will be testing your RAM as it goes.

If you suspect it's a heat-related issue, try running Memtest86+ with your PC sitting in the warm air stream from a fan heater, then in the same air stream with the heat off, and see if it crashes faster in warm air than cool.

In my experience, the most common cause of overheating in an older machine is dust and filth clogging fans and coating heatsink surfaces with a woolly blanket, which you can fix by taking off the cover, taking the machine outdoors, and blowing in it a lot. Overheating from dust can affect CPU's, video cards and/or power supplies.

In newer machines, the culprit is often poor heat transfer between the CPU and its heatsink, caused by either a complete lack of heatsink grease, overzealous application of same, or the use of a shit-grade self-adhesive heatsink pad with a little wrinkle in it somewhere. You can tell if this is likely by measuring the heatsink temperature and seeing whether it's within say 10° of the CPU temperature as reported by the BIOS.

If your CPU is way hotter than its heatsink, take the heatsink off, carefully clean all the old thermal crud off both the heatsink and the CPU, and put it back together with the thinnest smear of a good heatsink grease. Get a reputable tech to do it if you fear screwing it up.

The same considerations apply to the chipset and video card heatsinks. Overheating in either of these can freeze a system just as effectively as an overheating CPU.

But it might all be down to a dodgy Windows install; see this previous AskMe thread.
posted by flabdablet at 12:00 AM on November 5, 2006

Have you tried to boot it into safe mode (without anything)?
If it boots time after time into safe then one or more of your drivers may be hosed. In my experance if it is at room temp at start up, heat related things take at least 10-20 min.
Safe mode function (or lack of) is an important clue as to "is it a hardware or software problem?".
posted by blink_left at 12:40 AM on November 5, 2006

Malor, actually Allied makes high quality, low priced power supplies. They are pretty solid.
posted by doomtop at 6:35 AM on November 5, 2006

I also think it's probably the power supply that's causing your problems, but overheating is equally likely. Those old AMDs run hot, and that GeForce card does too.

First: Open the case and make sure the fans are spinning and that your heatsinks aren't coated with an insulating layer of dust. If that fixes it, you're good. If your mobo supports it, you can get chip temp monitoring programs. You want the temp to be below 85F.
Second:Because it's easy to do, run memtest86, mentioned above. If you get errors from the test, replace the memory.
Third: Pick up a 450W "AMD Certified" power supply.
posted by Mr. Gunn at 9:47 AM on November 5, 2006

It seems like the problem was the power supply. I installed a new one and it's running fine now. Thanks for the help, everyone. I was surprised by the speed and quality of every single one of the replies.
posted by secret.osha at 8:53 PM on November 5, 2006