On Fri, 23 Nov 2007 17:05:12 -0800 (PST),
Koolsk8boarder@excite.com wrote:
>> If it freezes instead of just running slower, it is far more
>> likely a video card issue, not CPU.
>
>Thanks for the suggestions so far.
>
>The only reason I did not consider that is because for awhile now the
>GPU has run at about 61 C when gaming, and it had never been a
>problem. I even read that the 8800 GTS usually runs that hot, or
>hotter, and shows no signs of problem. I just suspect the cpu due to
>the fact that it seems to throttle a lot for no apparent reason.
What is "it" in the above sentence, do you mean system
freezes or games freeze or you have direct evidence that the
CPU frequency is reduced?
Overheating CPU does not generally cause freezing when it
has a throttling capability. Overheating CPU generally
causes crashing, errors, rebooting, or complete system
shutoff if temp continues to climb. There are rare cases
where a very instable CPU will cause freezing but this is
such a narrow opportunity that it tends to happen outside of
games, typically a system in this state wouldn't even be
able to boot windows at all.
Try running Prime95's torture test, large in-place FFTs
setting. See if errors are produced, and load up a realtime
CPU frequency monitoring utility (Google will find some).
If the Prime95 test doesn't cause errors or change in clock
speed, it would seem you don't have a CPU overheating
problem.
>The
>power supply used on the card is substantial so I don't see that being
>an issue.
If you mean the card's onboard power circuitry, that is
exactly why it is more likely a problem, that there is more
to go wrong, more power used, more heat. The power
subcircuit on a cheap low end card is very simple and hardly
ever has problems unless overclocked a lot, or defective
capacitors are used (or someone tries to make it passive in
a badly cooled system which overheats it).
>I suppose I could look into installing a new fan on the gpu
>to help cool it, but I have trouble believing it is the source of the
>problem as it is fairly new (1 month old).
You would expect it to wait until it's older to overheat?
That seems unlikely to me, unless it was the result of a fan
failure or dust buildup.
On the other hand, maybe this new addition to the system has
strained your PSU, is it a trustworthy brand of adequate 12V
current rating for the system? You might measure voltage
with a multimeter while reproducing the freezing,
preferribly taking the measurement at the video card
connector instead of an unused connector.
>Aside from a benchmark, is
>there any reliable stress test or something that I could run for the
>video card, to help narrow down the problem?
Why not go ahead and run benchmarks to see if it reproduces
the problem?
Anything that causes the freezing would narrow down the
problem, depending on whether that cause is stressing the
video card. Opening the case and pointing a strong fan
directly at the video card while running a test or game that
causes freezing, then comparing the frequency of the
freezing would tend to suggest overheating video card. It
produces a lot of heat, perhaps the temp monitor for it is
wrong and it is not at 61C. Since it takes a little while
for heat to build up, the sooner it freezes when gaming the
more likely it is power instead of heat to blame.
Again I suggest that momentary freezing during gaming is
usually (almost always) a video card problem of some sort,
of it's hardware state (not driver related, usually, but
perhaps if it only happened in a single game then drivers or
a game patch might also be looked into, except that some
games stress different parts of the GPU to a greater or
lesser extent so even trying a few games isn't a direct
indicator of this unless you know if each relatively
stresses the card the same).
If all else fails you might try the card in another system
or another card in that system.