Computer shuts off randomly, only while playing trem.

[c4]

Hi all,

I built a new computer recently, installing Ubuntu with Gnome3 on it.  I have a 2.8GHz AMD X6 Hexacore Processor and a GTX560.  According to power consumption calculators, I should most certainly not be using all of my 600W power supply, but trem still shuts off randomly during larger games on more graphics intensive maps, such as Cruz, Nano, Arachnid, and Nexus.  Looking at my system monitor while in a big game, I've noticed that trem uses about 100% of two cores of my processor.  I'm using generally around 1.7GB of my 8 gigs of RAM.  After flash updating my BIOS as per superpie's instructions, I continue to have this problem.  I have even disabled 4 cores on the computer, running just on Cores 0 and 1, thus further proving that this cannot be a power supply issue.  The computer doesn't even shutoff during netradiant compiles using 100% of 4 cores to compile, only in trem.  I've had no troubles running Unreal Tournament 3 and BattleField Bad Company 3 on it.  The problem does not occur in 1.1, only in GPP.  Any suggestions from anyone?  Thanks in advance for your help.

[David]

Update the graphics drivers, it's possible they are buggy.

Is it overheating?  My old GPU got full of dust and started crapping out, some have heat sensors to switch off before it gets any damage / fire.

But trem is single threaded, it shouldn't be able to eat two cores :s.

[Celestial_Rage]

Sounds like it's overheating. Check your sensors output to see the temperatures of your system components.

[KillerWhale]

Sounds like it's overheating. Check your sensors output to see the temperatures of your system components.

Don't rely entirely on sensors. I once had sensors tell me my entire computer was running at 700C. Clearly, since I still have a computer, it was not.

[Tremulant]

I've noticed that trem uses about 100% of two cores of my processor.

It only ever maxes one core here, smp support is disabled at compile time for my binary.

Have you tried running prime95/mprime and stress testing the cpu for a while, do you actually have a six core production chip or a lesser chip unlocked to six cores? I have an unlocked dual core running quad here, and did experience switch-offs due to overheating when the stock heatsink couldn't cope with dust+warm weather, this isn't helped by the fact that unlocking apparently disables the on-chip temperature sensor, so accuracy is considerably reduced.

[c4]

Pretty sure this processor, which I have is not unlocked

http://www.newegg.com/Product/Product.aspx?Item=N82E16819103851

[Tremulant]

Pretty sure this processor, which I have is not unlocked

http://www.newegg.com/Product/Product.aspx?Item=N82E16819103851

Only pretty sure? You presumably bought a six core chip and stuck it in without needing to use unleashing, i'd be absolutely certain, if i were you.
How sure are you that trem's maxing out multiple cores? I take it you've been stress testing?

[Aelita]

Don't rely entirely on sensors. I once had sensors tell me my entire computer was running at 700C. Clearly, since I still have a computer, it was not.

qft.

As for your issue, try running sshd on your machine and remotely logging in when it locks up. If you can establish a connection, check /var/log/Xorg.0.log for anything relevant. If not, then it's likely a hardware problem. Do any of the caps/scroll/numlock keys work? Are their respective lights flashing (kernel panic)?

Are you using nouveau or the binary NV drivers? Nouveau is pretty buggy but so are the binary drivers (currently dealing with a bug where X decides to lock up and flash the cursor on both of my monitors while moving the cursor between them.. )

[Tremulant]

try running sshd on your machine and remotely logging in when it locks up.

Computer shuts off randomly, only while playing trem.

of course i can't be sure that he means what he says, but the way i read it there's going to be no opportunity for sshing in.

[c4]

It doesnt just lock up.  It shuts entirely off.  My indicator lights on the front of the box turn off as do the fans and everything else.  I have to flip my PSU off for atleast 5 seconds, and then turn it back on before I can get the machine to turn on again.

[Aelita]

Yeah, I can't read 

If you have to switch your PSU entirely off, that really sounds like it's faulty/failing. I would suggest getting an RMA from Newegg for it.

good luck!

[David]

Or, clean the dust out of the fans.  Quicker and easier, and probably going to solve everything.

[Thorn]

I have to flip my PSU off for atleast 5 seconds, and then turn it back on before I can get the machine to turn on again.

Crowbar/safety circuit on PSU kicking in due to failure. What brand?

Also try the OCCT PSU test, it will run linpack+gpu at the same time to stress all components.

...

what happened to brinstr?