New build randomly turns off

[CeeDub]

So I built my pc with the below components. Everything seemed to work just great. Halfway through installing the os, the pc turns off, then turns back on saying that overclocking failed (and it seems to randomly turn off). I am not overclocking. In fact I have the default settings in bios. I've checked all the cables and everything and I can't find anything wrong in Bios. My feeling is incompatability with the ram, but I'm not sure. Can anyone help?

DVD - LITE-ON 20X DVD±R DVD Burner

HDD - Seagate Barracuda 7200.11 ST3500320AS 500GB 7200 RPM SATA 3.0Gb/s Hard Drive - OEM $119.99

Video Card - ATI X1950GT

PSU - CORSAIR CMPSU-550VX ATX12V V2.2 550W Power Supply - Retail $99.99

Memory- Crucial 2GB DDR2 800 Ballistix Kit

Mobo - ASUS P5K-E/WIFI-AP LGA 775 Intel P35 ATX Intel Motherboard - Retail $139.99

Intel Core 2 Quad Q6600 Kentsfield 2.4GHz LGA 775 Quad-Core Processor Model BX80562Q6600 - Retail $279.99

[liambl]

Usually you will get that message after CPU overheat. Turn on the PC and let it run for a few mins in the System Health page of your BIOS. If it goes over 50C, you have a heatsink problem, reseat it and apply new paste, making sure all four locking tabs are securely attached.

[CeeDub]

I've been running my new machine for about a week now, and I'm stilling getting the same problem. I've checked the BIOS, but the temperatures all seem pretty low. Does anyone have any ideas to what could be wrong?

[doubledragon5]

do like liambi said check to be sure the heatsink is properly mounted.. When building intel builds securing the heatsink properly can be a pain.. You may want to try this, and you must remove the mb from the case.. http://forum.pcmech.com/showthread.php?t=132409

[CeeDub]

I've double checked the heatsink and everything looks good and solid. I've run the computer non-stop, and it turns off twice a week. So if I followed that procedure, I wouldn't know if I was making any process. Should I take everything apart anyway and try again?

[doubledragon5]

Are you able to get to the bios to check the temps? Are the pins on the heatsink secured to the motherboard. Even though you have a good psu you may want to try a spare if you have one.. Did you check to see if the ram is compatible with that board?

[CeeDub]

The bios checks the temps fine. The pins are secured (double and triple checked). I don't have a spare psu, but I'll look into finding one to check it out. As for the ram.... It's not specifically listed in the mobo manual, but It was suggested on newegg, so I assumed it was good. Could that where I went wrong?

[doubledragon5]

Could be, some ram if not certified by the manufacture can make a system become unstable. But I would suspect the psu first.. Corsair is excellent choice, but doesnt mean you won't get a faulty one from time to time..

[CeeDub]

Well, I've been playing around with my setup. Here's what I've done. I went into BIOS and set the cpu and case fans a little higher and I haven't had any problems for a few weeks. Today I went back into BIOS and set the fans at a lower setting, and the system shut down.
I'm experimenting now. My fans are at the lower setting. I'm monitoring the temps carefully. At a full load (I wrote a quick while(true) loop and ran it four times) none of the core go above 68 C. And everything seems to be working fine. Now I don't suspect the cpu anymore, but now I suspect the ram. The cpu is at full load and working fine, but the ram isn't being accessed at all. What do you think? Is there any way I can test this theory? Any program that excessively accesses the ram enough to do a thorough experiment?

[shadowpr]

Run memtest.

What are you using to check the temps of the cpu under load? What are you using to put it under load?

[CeeDub]

I'm running linux right now, so just the basic lm_sensors. I realize they probably aren't extremely accurate, but the point is that the temperature topped out at 68 C and the computer hasn't shut down like usual.

To put it under load, I'm running 5 movies at the same time, and I wrote a quick c program:
while(true)
which means that the cpu's will just spin. I ran this program four times in the background so that each a program will basically be running on each core. The load on the cores are indeed 99-100%, however, I am not sure if this is truly putting the cpu under load.

[CeeDub]

I just ran the memtest with success. Computer didn't shut down.

It's very strange. I can't seem to reproduce the problem, but It does happen at the worst times. Like when I'm doing a million things at once putting a huge load on the cpu. I guess I'll just have to live with the fan a little louder than I would like it to be. Thanks for your help.

[Katreat]

The fact that speeding up the fan fixes the problem makes me think you are running into heat issues. There is a program out that monitors your temperatures and provides you with a record of what they are doing. My problem is I can not remember what it is called :(. But at this proint I would try to fine something like that so when you re-boot you can look back and see exactly what is happeniing before shut down occures. Perhaps someone esle on these boards will remember what it is called.........

Kat

[Cricket]

SpeedFan (http://www.almico.com/speedfan.php)

:) Cricket

[not important]

I have to agree with Katreat. If you increased the fan speeds and the problem ceased..it sounds like overheating. Is your case adequately vented? Are the fans directed properly..front intake/ rear exhaust?

[amdalex]

Boot up with windows and get Prime95. Do a torture test with large in place ffts, monitor the temps with speedfan. See how high they get and tell us.

[shadowpr]

He already stated he was running linux. Not sure of the equivalent of prime95, or orthos for linux. Don't think they have linux versions either.

[amdalex]

Oh, I thought he had a dual boot going on.

[CeeDub]

I found an alternative to prime95 - mprime. I'm running the torture test. Here is the outpu:

Beginning a continuous self-test to check your computer.
Please read stress.txt. Hit ^C to end this test.
Test 1, 4000 Lucas-Lehmer iterations of M19922945 using 1024K FFT length.

The load on the cores is pretty small, averaging around 5% on each. Is this normal? The temperatures on the cores immediately climbed to the mid 60's C. Unless someone tells me differently, I'll let this run for a few hours and report back.

[CeeDub]

I have to agree with Katreat. If you increased the fan speeds and the problem ceased..it sounds like overheating. Is your case adequately vented? Are the fans directed properly..front intake/ rear exhaust?

My case is adequately vented, fans are directed properly. Overheating does seem to be the culprit, but I'm concerned that I can't re-create the problem with overloading the cpu. I basically want to ensure that nothing else is going on.

[CeeDub]

Ooops. I double checked the cpu load, and mprime is creating 100% load on a core (and it jumps around to different cores). Should I run mprime 4 times, so each core is being fully utilized?

[shadowpr]

If you only have 100% load on one core, then see if you can run four of them. Set the affinity, so each one gets a different core, if possible.

[CeeDub]

I ran mprime 4 times, so each core is getting the full load. Temperatures skyrocketed to 71 to 73 C and stayed there. The computer didn't crash like usual. I wonder what has made it crash before, but it won't crash now. I guess I'll never know what is making it crash.