What's causing my new computer to crash?

September 30, 2009 3:19 PM

My new computer freezes or reboots when I'm pushing it (e.g. encoding video). Can you help me figure out which part is bad before I start having to ship things back to newegg and pay restocking fees?

CPU: Core i5
MB: MSI P55-CD5
RAM: Crucial 4GB (2x2) DDR3 1333
PSU: 500W OCZ ModXStream Pro
VIDEO: MSI Geforce 9600 GT
HD: Samsung 1TB 7200RPM

It's usually fine for low intensity use (web browsing, office, etc.), but if I push it (say video encoding), it typically crashes within about 20 minutes. In Windows 7, sometimes it reboots, sometimes it just freezes. I had one BSOD. I think it also reboot once in Linux (gentoo), while running genkernel (unless that was some other problem)

My guess is it's either the RAM or the Motherboard. I have memtest86+ running right now, but so far it's been running for an hour and done 2 complete passes with no errors. On the other hand, one of the newegg reviews says:

This RAM does not play nicely with 1156 motherboards with the P55 chipset. Two other reviewers have mentioned rebooting issues with Gigabyte boards, and my Intel DP55WB had the same problem.

Did some testing and found that during PC-Check, the RAM would always lock up/reboot once it got to the block rotation test, but would pass MemTest.

PC-Check's price is not listed on their website, which leads me to believe it's out of my budget for the limited use to which I'll put it. As such, I cannot try to replicate these results myself.

Also, I've changed the Command Rate to 2T, but the problem still persists.

If it's not the RAM, it may be the motherboard. Is there any good way to test this other than trying new RAM and seeing if the problem persists?

The other thought I had was that the PSU isn't strong enough. I found a PSU calculator online which suggests it should be fine. Besides what I listed above, there's an old IDE hard drive (which is only connected right now to transfer files) and an internal CD-RW drive. There's also a USB DVD drive and a USB printer, but those both have their own power.

Does the hivemind have any suggests on the best and cheapest way to figure out what I should return? Also, do you have any suggestions on what I should replace that part with?

Thanks!

posted by chndrcks to computers & internet (19 comments total)

How is your heat situation?
posted by sonic meat machine at 3:29 PM on September 30

My first guess would be overheating as well. Are all the fans spinning?
posted by bjrn at 3:32 PM on September 30

Just for the record have you flashed the bios to the most recent version and downloaded updated drivers for everything? I was having a lot of similar problems and flashing the bios solved everything.
posted by sully75 at 3:40 PM on September 30

Woops, I forgot to put that in. Whenever I catch one of these crashes, I go into the bios and check the temps. The CPU temperature is usually around in the upper 40s C, and the system temperature is listed in the upper 30sC. From my quick googling, those didn't seem dangerously high. All the fans are running. The CPU heatsink was kind of a pain to get on (which is odd, because I think they designed it to be easier), but if it weren't connected properly, I'd think the temperatures would be higher.
posted by chndrcks at 3:41 PM on September 30

BIOS is the latest version (1.2)
posted by chndrcks at 3:41 PM on September 30

try encoding video with the side of the computer open and a large fan blowing at it or some other obvious way of totally ruling out overheating.
posted by imaswinger at 3:49 PM on September 30

Also should check to make sure the heatsink and processor mounting is secure. And BIOS temps may not be actual temps, for a variety of reasons.

Nonetheless, the processor should throttle itself before it gets so hot as to freak out.

I would run memtest continuously, overnight, on each stick of memory separately. Note the time that it starts running, so that if it reboots itself, you will be able to tell by the times it reports on the screen. You are running memtest off a floppy or a CD, right, and not in Windows?

Feel the air coming out of the power supply while its running hard. It shouldn't be too hot. If it is, it might be bad. It can be underpowered- not because you are using 500 watts, but because one of the individual voltages might be overloaded, or flaky. The only way to size a PS is to add up the wattages for the individual voltages separately.
posted by gjc at 4:04 PM on September 30

I had a laptop that would do this for heat reasons. Fan would go full bore, then blammo. Try to minimize variables--do you have any RAM to swap in? Maybe another computer? The external fan idea sounds promising too.
posted by RikiTikiTavi at 4:22 PM on September 30

Checking the temp in the BIOS after a crash doesn't tell you anything because the chips cool really fast once the load is off them. Try picking up a copy of Speedfan and leaving it running on screen while you're stressing your computer, to see what the temp is doing.
posted by Chocolate Pickle at 5:53 PM on September 30

Could be your GPU or your Motherboard chipset overheating. Do you have any auxiliary cooling on those devices? Also what imaswinger says, open up the case and have a fan blow on the MB/CPU. This will quickly tell you if it's a heat related issue.
posted by white_devil at 8:07 PM on September 30

Thanks for all the suggestions so far. I think I may have found the culprit, but I need your help interpreting.

I ran my encoding while watching speedfan as Chocolate Pickle suggested. The system locked up pretty quickly, but fortunately, it froze instead of restarting, so I was able to write down all the numbers.

GPU: 41C
Temp1: 48C
Temp2: 57C
Temp3: 37C
Core0: 48C
Core1: 42C
Core2: 45C
Core3: 46C

I'm not sure exactly what the Temp1-3 are measuring. When I go to configure it says "Chip F71882F"

Power Numbers
VCC3V: 3.39V
Core: 1.22V
VRAM: 1.47V
VChipset: 1.41V
+5V: 5.80V
+12V: 9.94V
VCC1.5V: 0.74V
VSB3V: 3.39V
Vbat: 3.28V

When I saw these, I thought for sure that the +12V had to be the problem, but I've just pulled up Speedfan again and it's listing +12V at 9.94V and the system's fine. Is it not a problem? Or is it just a problem when I'm stressing the system?
posted by chndrcks at 8:45 PM on September 30

The ATX spec says voltages have +/- 5% tolerance. So, 9.94V on the 12V rail (which powers your CPU) is failing. So is anything above 5.25V on the 5V rail.

Try a new power supply. Buy it by the pound...all else being equal, the heavier PSUs will have higher quality internal components. Good luck.
posted by edverb at 9:08 PM on September 30

At the risk of being too involved in my own question...

I'm not sure speedfan was accurately reporting the voltages. The MSI control center was reporting it around 12 when speedfan said 9 (which is what it reports even when the system is working fine). I'm worried about the temp2. I can't figure out what that's reporting, but if it's on the motherboard, approaching 60C may be the problem.
posted by chndrcks at 9:36 PM on September 30

Maybe not enough airflow on the chipset heatsink?

(This doesn't help, but I've really been liking Intel branded boards lately.)
posted by gjc at 10:22 PM on September 30

You said that this is a new computer. I think it's time to consider returning it under warranty.
posted by Chocolate Pickle at 10:41 PM on September 30

Is your BIOS set to all defaults? It's tedious, but setting it as conservatively as you can and turning up settings one by one is sometimes the only way.

Prime95 is a good way of stressing your CPU. If it survives 30 mins of the tests that don't use much RAM then you can probably eliminate the CPU as the fault.

As gjc says, check your northbridge temps (usually a heatsinked chip near the PCI-mumble slots), sometimes the airflow gets really screwed up depending on case and graphics card.
posted by samj at 1:39 AM on October 1

Why on earth would you pay a restocking fee if you've been sold a faulty computer? There's clearly a hardware problem so send it back and get a replacement.
posted by salmacis at 2:47 AM on October 1

I suspect the computer was built from parts bought separately. It's more fun that way, but a bit riskier.
posted by samj at 3:24 AM on October 1

Turns out it was the RAM, or at least the RAM/MB combination - others on newegg reported the same problem. I tried comparable RAM from Kingston and haven't had a problem since.

PRIME95 was really key. Even though memtest didn't detect any errors, prime95 tests that stressed RAM would fail almost instantly. It also helped rule out heat (tests that were high in CPU stress but low in memory usage still generated plenty of heat without failing).

I think the video encoding was partly a red herring. Some of the problems have been fixed (randomly restarting during encoding), but I think other problems (audio garbling) are actually the result of buggy software, which I hadn't noticed before because I changed how I did it (basically, the audio encoding fails if I try to simultaneously encode multiple files, which I didn't use to try on my old computer).
posted by chndrcks at 2:38 PM on November 22