Computer havok, help?

[bloodstar23]

So my computer has decided to pretty much host an all out war against me.
I've tried every troubleshooting method, debugging, scanning, everything I
could think of and nothing is working. The computer just randomly decides it
wants to shut down whenever I'm trying to run like two or more active
graphics programs, such as trying to run a game and a flash at the same time,
multiple flash files, multiple games, even multiple GIFS will cause this, and
it doesn't blue screen, it just shuts down. Upon turning it back on I have to
try multiple times as it will start up, then shut down again several times
before actually booting up. I figured this HAD to be a GPU issue, I have two
graphics cards so of course that's it, but upon removing and individually
testing each under the same conditions, it still crashed just the same. Yes,
the crashing is bad enough I can actually quite easily force it to crash.
This time, however, upon re-installing both cards and rebooting, my nTune
suddenly doesn't work, and after launching the Nvidia monitor (having to use
cmd to do so because I couldn't boot up nTune), and putting it under the
exact same conditions that crashed it before, it refused to crash. I opened
over 20 flashes simultaenously and it wouldn't go above 74% CPU usage. (I was
trying to force a crash and monitor what happened at the moment of crashing)
It's like the computer is fighting against me, it won't crash when I'm trying
to see why. The temperatures of the system are stable and at a healthy level,
they've never gone much above it seeing as I have the side panel OPENED and a
FAN pointed directly at it to make sure it does not overheat. I just don't
know what's wrong with this thing or how to fix it, it's demonic I swear.

I'm running dual NForce 8600 GT graphics cards with an SLI connection
3.000 Ghz dual core pentium processor
3 gigs 533.333 MHz DDR2 RAM
FSB: 800.000 (QDR) Mhz
PCI-E: 2500.000 MHz
GPU core (3D): 540.000 MHz
GPU Memory: 700.000 MHz

It could just be coinscidence, but the problem seemed to start after I
updated my graphics drivers. Honestly it wouldn't at all surprise me if this
was an issue with drivers, because it seems that the point of new drivers is
to completely sabotage computers, they ALWAYS break something. If you know
what this problem is, or know a way I can roll back my drivers, please let me
know. This computer is driving me completely insane at this point.

Note: No matter how much I run, Memory always remains at 18%, what's that
about?

 

[DDW]

If you know
what this problem is, or know a way I can roll back my drivers, please let me
know. This computer is driving me completely insane at this point.

Go to Control Panel (Classic version)

Device Manager

Display adapters

Right-click the adapter

Properties

Driver

Roll Back

DDW

 

[Steve Thackery]

The computer just randomly decides it

wants to shut down

There's the clue. A shutdown, rather than a bluescreen, is almost always a
hardware problem.

To be specific, it is pretty well always one of two things. The most likely
by far is your PSU being under-rated (or gradually failing). Its overheat
or over-current sensors are shutting it down.

The cure is to fit a new, bigger and better one. (You could try blasting
the dust out of it, but I don't think it'll work - I think it's simply too
underpowered for your rig).

The second possibility is your processor overheating. That doesn't normally
cause a complete shutdown like that, but it *can* do. The cure is to remove
the heatsink, clean off and renew the thermal paste, and blast the heatsink
and fan clean of dust. After refitting, check the fan works.

But I wouldn't bother. I'll bet the remains of my bank balance it's the
power supply.

SteveT

 

[Richard Urban]

What size power supply do you have installed? It could make a huge
difference.

 

[bloodstar23]

Well the drivers I have now came straight from the driver website, and the
power supply reads:
AC Input: 115V/230V, 10/5A, 60/50Hz

DC Output: +3.3V==30.0A(ORG), +5V== 28.0A(RED), +12V1==14.0A(YEL)
+12V2==15.0A(YLW/BLK), +5Vsb==2.0(PURP), -12V==0.3A(BLUE)
(+3.3V & +5V = 140W Max)

It's a model number PSF500V Switching power supply.

I can't really make sense of that, I'm more a software guy than a hardware
guy. By the way, what's PSU? Never heard if it before, is that power supply
unit?

 

[bloodstar23]

Okay yeah I get that PSU is power supply unit now lol. Anyway I'm going with
Steve on this one, with the way it's been acting, it does make sense it could
be the power supply which I'm thinking by what it says is a 400W power
supply. Is there a way to calculate just how much power I need? I'm a bit low

 

[Richard Urban]

Here is one:

http://www.adecy.com/psu/

 

[w_tom]

Okay yeah I get that PSU ispower supplyunit now lol. Anyway I'm going with
Steve on this one, with the way it's been acting, it does make sense it could
be thepower supplywhich I'm thinking by what it says is a 400Wpowersupply.

It could be a power supply - and anything else. Replies are classic
wild speculation. A recommended solution to wild speculation is to
keep buying and replacing parts until something appears to work. Also
called shotgunning. Typically a most expensive solution.

Do you want speculation or an answer that is definitive the first
time? For example, is that power supply definitively good or
definitively bad. Nobody has or can answer definitively given the
provided information. Definitive means numbers in less than two
minutes from a multimeter. In your case, the best information is to
get the computer doing as much as you can make it - then take voltages
from any purple, red, yellow, and orange wire. Those numbers must
exceed 3.23, 4.87, and 11.7 VDC.

The meter is available in most any store that also sells hammers -
for about the same price. This is the least expensive, the fastest,
and the only definitive way to establish what is and is not defective.

Your power supply could have been defective all along. A defective
power supply can boot and operate a computer. As it ages, the
defective power supply eventually caused computer crashes.

400 watts is more than enough power. Those current numbers (not
watts) are important. But that means you know how much current is
required from manufacturer specs. Since that information is typically
not available from every component manufacturer, then use the above
experiment to learn if each power supply actually does provide
sufficient current.

BTW, if replacing a power supply, again, a defective supply can
still boot a computer. IOW also use the meter to confirm a new power
supply actually does provide sufficient current. IOW find a failure
before it even happens. The meter is a powerful tool (and even sold
in Kmart or Wal-Mart).

Bottom line - the only useful answer is one that either says so
without speculation OR provides a method to get that answer without
speculation. An answer without any doubt means numbers and an
inexpensive meter.

 

[Steve Thackery]

It could be a power supply - and anything else. Replies are classic

wild speculation.

I agree, but the speculation isn't "wild". It is based on years of
experience. Of course, as you say, without the right diagnostic equipment
it is impossible to diagnose the problem definitively.

However, that doesn't alter my argument: that symptoms like this are almost
always due to the power supply over-current or over-temperature sensors
shutting it down. I made it clear that this is as close as we can get to a
diagnosis, by using the term "most likely".

I don't fully agree with you about using a multimeter. You can measure the
voltages, of course, and if they are out of spec you need to change the
power supply. But really you need to know how much current is being drawn
from each line - that is the crucial measurement, and it isn't really
feasible without intercepting each line. Then you compare the measured
current with the rating for that line on the PSU label.

A low voltage on the power lines is NOT necessarily an indication that they
are being overloaded. It may mean that the PSU has drifted out of spec (so
it will still need changing). Similarly, a voltage within spec is NOT
necessarily an indication that the current draw is within limits. Some PSUs
will continue to provide the right voltage right up until the over-current
sensor trips out.

In summary, if you measure the voltages and they are low, you should change
the PSU. However, if they are within spec, you might still have to change
the PSU.

400 watts is more than enough power. Those current numbers (not
watts) are important.

The two sentences disagree. A 400W power supply may not be powerful enough
if one of the lines doesn't have sufficient current rating for the job.
This is especially true of low cost supplies. But as a general rule, the
higher the power rating, the higher the current rating on each of the lines.

SteveT

 

[Snidley W.]

In summary, if you measure the voltages and they are low, you should change
the PSU. However, if they are within spec, you might still have to change
the PSU.

Then what's the point in measuring? HMMM?

 

[Steve Thackery]

Then what's the point in measuring? HMMM?

Because if they are low, you definitely need to change the PSU.

If they aren't low, you must continue with further diagnosis.

Or don't bother. PSUs are cheap. Just change the damn thing.

SteveT

 

[Snidley W.]

Get a multimeter and test the Red and Black (+5V) and Yellow and Black
(+12V) under various loads.

GEESH. Why didn't _I_ think of that!? I'll just reach back on my
workbench and pull a multimeter outta the pile of 'em I have.

Like most everyone reading these groups have </sarcasm off>.

 

[Steve Thackery]

Get a multimeter and test the Red and Black (+5V) and Yellow and Black

(+12V) under various loads. If they drop below 5V and 12V, replace the
power supply. What are the specs on your power supply? What are their
combined specs? You have a pair of 8600 in SLI?

And as I've said elsewhere, that is NOT a reliable indication of whether the
lines are being overloaded.

SteveT

 

[westom1]

And as I've said elsewhere, that is NOT a reliable indication of whether the
lines are being overloaded.

Swapping a power supplly is not a reliable indiciation of whether
the lines are overloaded. Those who do this stuff professionally know
that. There is no other method of identifying an overloaded supply
without much more expensive equipment.

An overloaded supply can still boot and run a computer. And a
perfectly good supply can appear defective in an otherwise perfectly
good computer. Swapping power supplies may create additional computer
damage. It does not report anything useful. It also takes
significantly longer. Longer because the meter reports on the supply
and other possible problems - in much less time.

Due to no numbers from a meter, then no useful reply exists. And
with numbers, other information may also be obtained from the better
informed responders. No meter numbers means only the naive can
reply. Only useful reply occurs when those meter numbers are
provided. And if an power supply is defective, the meter must again
be used to confirm the new supply is both good and sufficient. Just
another reason why the electrically knowledegable technician uses a
meter.

The meter will easily identify an overloaded power supply when the
computer is doing many things simultaneously. Professionals routinely
knew before PCs existed even 40 years ago.