Sign in with
Sign up | Sign in
Your question

System shuts down under 100% CPU load.

Last response: in Systems
Share
February 25, 2013 2:48:18 PM

I ran into a troubling issue yesterday which has never crept up before.

I have a year and a half old homebuilt box,

The specs of the box are...

Biostar MCP6P3 Ver. 6.x Nvidia Nforce 430 / GeForce 6150 mainboard.
AMD Athlon II X4 640 quad core CPU.
Corsair Vengance 2x4GB DDR3
2x1 TB WD SATA 3 HDD. 800 GB Windows 7, 1.2TB CentOS 6.3 dual boot. (This will eventually change, Windows 7 will end up as a VM under CentOS...)
PATA DVDRW drive. If I recall correctly Samsung.

No add in cards of any sort, no RAID etc...

Yeseterday I attempted to use this box, which gets little use, to speed up the process of encoding DVD to mp4 conversions so that I can stash my DVD media away and stream my collection over my network... I ran into a snag...

Using Handbrake to encode the files, I get a warning from Norton Internet Security about high resource usage, and then the box shuts down...

Now I would expect if this was a processor, mainboard, or RAM issue it would BSOD, but no, not even. It just shuts down. QUICKLY.

I grabbed Prime95 to test, and same thing. So I booted to CentOS, and found the same thing. And it wasn't like it just abruptly yanked power. It is quite obvious that a shutdown command was initiated by SOMETHING...

Is it possible that maybe my thermal thresholds on the CPU are set too low in the BIOS? Hitting that 100% heats the chip up and throws the shutdown command?

Other than this, the box has been running smooth as butter since build day. I am not overclocking it, for the most part this box exists so that my wife can play Facebook games, work spreadsheets etc...
February 25, 2013 2:52:17 PM

Monitor the temps while it's under load and see how hot it's getting. Could be as simple as re seating the proc with some new thermal grease.
m
0
l
February 25, 2013 4:17:41 PM

Thank goodness for the bags of extra alcohol wipes and heat sink paste tubes Dell ships with replacement motherboards! (no matter how many sockets you have populated, they ship 1 each for each socket...).

I must admit, this is somewhat concerning. This is the first time I have had something go bad with any of my builds in so short a time frame... The box I usually use is a custom built a long time ago, Athlon XP 3000+ 2GB / 500GB HDD box from back in 2002 or 2003... It runs so well I don't want to mess with it...
m
0
l
Related resources
February 25, 2013 4:52:32 PM

pit_1209 said:
Check your temps with realtemp, applying thermal paste is somewhat easy but there are tons of way to do it, check this.
http://www.tomshardware.com/forum/333968-28-thermal-pas...

That´s a way of many, check youtube for more.


I am very familiar with applying heat sink compound. This particular heat sink is different from ones I had done previously. This had heat sink goo already applied to the heat sink itself...

m
0
l
February 25, 2013 4:58:18 PM

yeah that´s common just wipe that out and apply some good one like the mx-4 if you have it, the pre applied paste is not made of good quality (personal experience)
m
0
l
February 25, 2013 6:14:54 PM

I was thinking Arctic Silver, but hey, whatever conducts well, and makes a solid conductive layer between the heat sink and the die... (I have the Arctic Silver literally in hand).
m
0
l
February 25, 2013 9:46:05 PM

dbhosttexas said:
I was thinking Arctic Silver, but hey, whatever conducts well, and makes a solid conductive layer between the heat sink and the die... (I have the Arctic Silver literally in hand).


You´re good to go then but remember that artic silver needs a curing time so maybe you won´t see any drop in temps for now.
m
0
l
February 26, 2013 12:45:17 PM

Not convinced it is temp at this point. Decided to rip out the Radeon card since the main board is Nvidia chipset, figured drivers might be doing something stupid... Re-enabled the onboard Geforce card, and have video artifacts galore... Haven't had the time to kick it off yet, but planning on flogging the system with Memtest86. At this point, I strongly suspect the RAM, possibly the main board. The shutting down cleanly instead of throwing a BSOD is what has me stumped...
m
0
l
February 26, 2013 3:25:23 PM

dbhosttexas said:
Not convinced it is temp at this point. Decided to rip out the Radeon card since the main board is Nvidia chipset, figured drivers might be doing something stupid... Re-enabled the onboard Geforce card, and have video artifacts galore... Haven't had the time to kick it off yet, but planning on flogging the system with Memtest86. At this point, I strongly suspect the RAM, possibly the main board. The shutting down cleanly instead of throwing a BSOD is what has me stumped...


Ok but did you discard temps problem with coretemps? if it´s not then it could be your Psu also, makes more sense to me.
m
0
l
February 26, 2013 4:38:42 PM

pit_1209 said:
Ok but did you discard temps problem with coretemps? if it´s not then it could be your Psu also, makes more sense to me.


Haven't been able to keep it stable long enough to grab and install realtemp.

Hadn't considered a power issue... It's possible.
m
0
l
February 26, 2013 9:01:19 PM

Try to use another psu for a test to see if that´s the problem and try to see your temps on the Bios.
m
0
l
February 26, 2013 11:14:54 PM

Try swapping a power supply from another machine and seeing if it does it anymore. Processors are indirectly powered by the 12V PSU bus via their own personal PSU that converts 12V to 1.5V, so in theory if your PSU is tired and your CPU starts to draw some serious amps, the PSU may crap out, the CPU will crash and that alone is enough to cause BIOS to cut the juice. Old boards (pre 2000s, Pentium 3) can be started with a faulty (or missing) CPU, but a Pentium 4 board will refuse to turn the fans and anything else on without the processor in working order. So make it a point to check the PSU.
m
0
l
February 27, 2013 3:39:57 AM

I might have this licked. And it may just have been thermal, and not for obvious reasons...

I got home tonight and tossed in the Memtest86 disk, and ran through a few passes when it shut down again... So I went and popped the side of the case, and found among other things, a quite obnoxiously hot heat sink, crammed full of dust, actually the whole stinking thing was full of dust...

Took the box to my shop, and have it a good blast of filtered and dried compressed air (the same hose sprays reduced lacquers for fine furniture finishes...), The heat sink, power supply, fan bearings etc... were caked, it took me better part of 10 minutes of brushing, blowing and cleaning to get the crap out of it...

I sort of forgot, this box was in the living room when we had to replace a bunch of sheet rock due to storm damage... Apparently the contractor just dusted off the outside of the case, not covered it in the first place...

Been running for 2 hours now... Back to encoding my DVD collection using all 4 cores!
m
0
l
February 27, 2013 11:09:50 AM

Figures. Stuff like that happens sometimes. We had to tear down a wall at our house due to termites. We had EVERYTHING within a hundred feet of the damaged wall under several old blankets as we dissected the wall to amputate the termite stricken 2x4s. But Sheetrock powder goes EVERYWHERE, so we had a satellite box and a few laptops that were all gummed up. Glad to see that all is well again, and at least nothing died from heat.
m
0
l
February 27, 2013 12:24:14 PM

Yeah, it's not only stable, but it's fast again...
m
0
l
February 27, 2013 3:58:10 PM

Glad the thermal stress did not KO your CPU... My Core I5 was $210.10 from the cheapest place I could find when I built my tower a year ago.
m
0
l
!