SOLVED (Re: Python hits the spot)

David K. Trudgett dkt at registriesltd.com.au
Tue Jun 25 01:30:47 CEST 2002


On Monday 2002-06-24 at 22:27:55 +0200, Chris Liechti wrote:

> >> It was really the laptop processor. I tried the calculation on a
> >> stationary machine (Pentium II 450 MHz and SuSE 8.0), and it works.
> >> 
> 
> you have to admit that it was a nasty failure... is bad hardware your first 
> thought when your program fails?  

Should be an object lesson in that: software engineers need to keep an
eye on hardware issues, including potential failures. It is actually
more common than one would think that 100% CPU for extended periods
can cause hardware failure. Reason: most PCs are engineered and tested
under light load conditions such as those that 95% of users subject
their machines. Many servers, however, run at or near 100% CPU for
extended periods of time, with near continuous disk activity and so
on. Therefore, server class machines are engineered and tested to a
higher standard, and are less prone to such failures (although all
hardware fails eventually).

I remember that I could never do serious image processing on an "old"
Intel Pentium 166 MMX machine because after five minutes it would
become overheated with buzzers sounding to remind me of the fact! It
had the regulation heat sink and CPU fan but still overheated. Go
figure.

Classic symptom of heat failure: machine reboots, stops responding, or
specific components (such as a video card) fail after X minutes from
power on, or X minutes of high CPU usage.

Memory faults can be even harder to spot because they often cause
bizarre symptoms that seem unrelated to memory issues. I once spent
several hours diagnosing and fixing a random "CDROM" fault, including
swapping in a replacement drive and data cable, only to find that the
problem went away after replacing a faulty SIMM card.


David Trudgett







More information about the Python-list mailing list