The Design of Software (CLOSED)

A public forum for discussing the design of software, from the user interface to the code architecture. Now closed.

The "Design of Software" discussion group has been merged with the main Joel on Software discussion group.

The archives will remain online indefinitely.

software shutting down my PC

Before I ship a release I run a command line version of my numerical solver on a big batch of problem files. If there are any crashes or asserts I know I have got a new bug in the solver.

Several times recently this command line solver has shut the machine down! I sat and watched it. It chugged away for about 20 minutes and then suddenly my XP box shut down - ffftt - machine off. No blue screen or OS shutdown. Nothing in the Windows event log.

It seems to happen on a particular large problem file, but only sometimes.

Anyone else seen something like this? Any ideas what would cause it?
anon to protect the guilty
Wednesday, January 03, 2007
 
 
I would ask this on the Sysinternals forums. It's a Windows internals & troubleshooting hangout.
*myName Send private email
Wednesday, January 03, 2007
 
 
You specifically mention a "particularly large file" when this happens and that tends to set off a red flag in my mind concerning memory issues.  Perhaps you have a hardware failure in the memory that is starting to surface?  Have you ran this on a different box to see if the problem persists?  This is what I would do if I found myself faced with your situation.  Memory failure is often overlooked and though I would be the first to chime in on Microsoft Windows being buggy along with faulty 3rd party drivers and poor application code; truth is many people overlook that faulty memory is a very real issue and more common place than publicly given credit.
Chris Send private email
Wednesday, January 03, 2007
 
 
Yea, I'd agree it's either a hardware flaw or overheating.
Steve
Wednesday, January 03, 2007
 
 
That is EXACTLY the behavior of my PC when I'm playing an intense video game and the CPU fan has gotten covered in dust--in other words, overheating.  My BIOS shuts the PC down when I go over a theshold temperature.

Download SpeedFan (http://www.almico.com/sfdownload.php) and watch it while your app is chugging away.
PA Send private email
Wednesday, January 03, 2007
 
 
I agree.

A lot of hardware is marginal in this way, designed assuming there will be frequent intervals of idling.  Improved cooling can help.

This could be exactly what you're seeing.
Old Guy
Thursday, January 04, 2007
 
 
Thanks for the helpful pointers.
anon to protect the guilty
Thursday, January 04, 2007
 
 
And if the computer is running WinXP, there is a setting to change it from automatically rebooting on errors and showing you the blue screen.

MS must have gotten tired of reading about all the BSOD, so they hide it by default and "helpfully" reboot the computer.
Michael
Thursday, January 04, 2007
 
 
I had a similar problem years ago.  It turned out that I had a bad RAM chip.  There are single-floppy based memory diagnostic programs available for free that boot the machine and perform hardware diagnostics.  I would suspect this first before a bad disk, then a bad disk, then (who knows.)

This is the one I used:

http://www.memtest86.com/
Meganonymous Rex Send private email
Thursday, January 04, 2007
 
 
The cpu shuts down suddenly like that usually because of a cpu overheating problem. Intel chips do this to protect themselves from flaming out (after throttling down their speeds to cool down the processor)

1. Watch your cpu usage ... you can use task manager (type taskmgr in the run box) for this ... if its cycling at greater than 90% utilization kill the process. Now you know you have a bug in the program.
2. You should also get some temperature monitoring software for you computer ... I use MBM (Motherboard Monitor) but it doesn't work well for prebuilt computers (Dell, Sony etc)
3. Clean out your cpu heatsink and fan (and check your power supply too) ... Dust covering the fans in both items can a. cause your cpu temperatures to be much higher than they should b. cause your computer to malfunction/shutdown/die. If the computer is prebuilt this is usually a must.
anonymous_coward
Friday, January 05, 2007
 
 
This PC runs 24x7 without any problems, except when doing these compute intensive tasks. It seems to crap out after about 10-20 minutes of 90%+ cpu activity. So it sounds like CPU overheating.

>if its cycling at greater than 90% utilization kill the process. Now you know you have a bug in the program.

It is a number crunching program. I *want* it to run as close t0 100% cpu as possible. Thats not a bug.
anon to protect the guilty
Friday, January 05, 2007
 
 
What kind of cpu is it and what kind of computer (hand built or HP/Dell/Sony)?
anonymous_coward
Friday, January 05, 2007
 
 
>What kind of cpu is it and what kind of computer (hand built or HP/Dell/Sony)?

Its a 'mesh' PC running XP. It is one of a batch bought by my old workplace. It runs 24x7 and its probably full of dust!
anon to protect the guilty
Saturday, January 06, 2007
 
 
Asuming hardware first is a good bet, on newer computers and newer OSes that is more likely than the other thing that raised a flag in my mind.

When you spoke of comand line and large files(long run time). I started wondering about two things first, the long run time gives more time for a small incremental error to surface, such as a memory leak.

On the other hand the only time I've had trouble with a program rebooting a computer was back in the day with a almost 32bit dos c++ when I dereferenced a null pointer.

Like I said probably not helpful, but I felt I should mention it.
Michael G Send private email
Monday, January 22, 2007
 
 

This topic is archived. No further replies will be accepted.

Other recent topics Other recent topics
 
Powered by FogBugz