Break the problem down into parts. A suspect that causes
strange failures is the power supply 'system'. Yes, a
'system' that includes one component - the power supply. We
first take numbers. There is no better power supply 'system'
test (other than with expensive equipment) than the 3.5 digit
multimeter - a tool so ubiquitous as to be sold in Home Depot,
Sears, Lowes, and Radio Shack. A previously posted procedure
defines how to confirm power supply integrity in but two
minutes. But you need only concentrate on one particular
condition. First confirm voltages on orange, yellow, red, and
gray wires first when computer is just idling, then again when
game (and other programs) are accessing all hardware
simultaneously. If the voltages 'under load' drop out of the
upper 3/4 of those limits (in chart in previous post), then we
have a guilty component.
The procedures and some pictures that demonstrate the
equipment:
"Computer doesnt start at all" in alt.comp.hardware on 10
Jan 2004 at
http://tinyurl.com/2t69q
"I think my power supply is dead" in alt.comp.hardware on 5
Feb 2004 at
http://www.tinyurl.com/2musa
http://techrepublic.com.com/5102-10586-5566528.html
www.ochardware.com/articles/psuvolt/psuvolt.html
Once the power supply 'system' is confirmed, then move on
to other suspects. Responsible computer manufacturers provide
comprehensive hardware diagnostics either on hard drive or via
their web site. Comprehensive diagnostics execute without
Windows. You don't want (yet) testing with Windows loaded.
Testing without the OS is what diagnostics are for. Only test
hardware.
If your manufacturer is not so responsible, then seek
diagnostics from third parties or component manufacturers.
Start with the memory diagnostic such as Memtst86 or Docmem.
Perform the memory test at room temperature. Then repeat test
with memory heated by a hair dryer on high. One need only
review manufacturer data sheets to know this is 'pig heaven'
normal temperature to memory. But defective (intermittent)
memory tends to become obvious when heated at temperatures
well above what a hairdryer can output.
Also repeat the heat and diagnostic test on chips that
interface with memory. Heat and diagnostic testing also
applies to other components - disk drive, video controller,
etc. Download those diagnostics as well.
In an NT OS, the most likely suspects after power supply are
memory, video controller, sound card, CPU, and CPU support
chips. NT limits what hardware can actually crash the system.
Once you have verified hardware, then move on to usual
software suspects.
Meanwhile, I assume you have already reviewed the standard
error date in Device Manager and in the system (event) logs.
I assume your problem is with any program that, for example,
does massive data processing such as video games.