Computer freezes

[Amaranth]

Hello,

I previously posted about problems with my computer randomly freezing
especially during games, like Wolfenstein Enemy Territory.

Here are my computer's specs:
Mobo: Asus P5ND2-SLI
CPU: Dual Core Intel D 805 2x2.66GHz
Memory: 3x1GB sticks
Gfx card: XFX 6200LE or XFX 9600 GSO
Hard disk: 3 different types
Sound: SB Audigy or Realtek on board sound.

I've run Memtest overnight without any problems. So memory is OK.

Prime95 was run for a few hours without any problems either. So CPU is
OK.

The crashes seem to be more frequent when using the XFX 9600 GSO card.
I've thus run Furmark to stress the XFX 9600 GSO but no crashes. The
crashes only seem to happen in games and they have happened while
running the lower specced XFX 6200LE card.

The crashes seem to be more frequent with the Realtek on-board sound.
Replacing it with a PCI SB Audigy reduced it but there are crashes on
both sound devices.

The crashes have happened on 3 different hard disks and at least 3
fresh Windows installs.

I'm at the end of my tether now and don't know what the cause might
be. I'm guessing either a dodgy PSU or mobo.

Some crashes give the BSOD (see below), others just freeze entirely.
http://img7.imageshack.us/my.php?image=screenshotxoo.jpg

I haven't had any BSOD crashes since changing my sound from the on
board Realtek to the PCI SB Audigy.

Any ideas what it might be? Thanks.

 

[Edify]

Question, is the power supply adequate for the graphics cards you’re
running? Check the video card OEM site for issues (updated drivers,
compatibility). Also, do you have adequate cooling, do you have a way
of checking case temp while running games? If not, I suggest leaving
the tower open (safely, avoiding possible static discharge and
moisture) with a fan (external) blowing on components.

This sounds more like a voltage issue than driver problem, either the
power supply is underperforming by providing less voltage than what’s
required (or spiking due to faulty condition) or the instability of
overheated MOBO and/or card leads (case temp too high) is causing the
flow of either data and/or current to be unstable. This is just a
guess but following the above advice is a free opportunity to prove
the theory right or wrong. Just don’t touch any internals without
being grounded and don’t allow anything to fall into the case while
internals are exposed.

Happy troubleshooting

 

[Amaranth]

You probably have more than one problem. The freezes are
probably overheating of something or PSU instability, while
the bluescreen crashes with the Driver IRQL message tend to
indicate a driver, or game bug. Replace drivers (especially
sound and video) and try more conservative game settings and
go to the game developer's site and look for patches and
read in the forums for others having similar problems.

The BSOD crashes seemed to have stopped after I switched my audio
system over from the Realtek on-board sound to the SB Audigy.

I've tried replacing drivers for the XFX 9600 GSO (semi-latest, latest
and one that came on the CD in the box) and switching to the XFX 6200
LE. Freezes happen on all of them, but not the BSOD.

 

[Amaranth]

Question, is the power supply adequate for the graphics cards you’re
running? Check the video card OEM site for issues (updated drivers,
compatibility).

I think it adequate power isn't an issue as the crashes have happened
even with the low specced XFX 6200 LE card. The PSU might be unstable
though, as Kony points out.The PSU's brand is Sansun.

Also, do you have adequate cooling, do you have a way
of checking case temp while running games? If not, I suggest leaving
the tower open (safely, avoiding possible static discharge and
moisture) with a fan (external) blowing on components.

Tower's already open on both sides, as I have been pulling out hard
drives and video cards to try to diagnose the problem. Usually after a
reboot when I go into the BIOS about a minute later, my mobo temp is
~35C and my CPU temp is about ~68C.

When Furmark stressed the XFX 9600 GSO graphics card, it was stable at
85C for an hour or so that I tested it.

This sounds more like a voltage issue than driver problem, either the
power supply is underperforming by providing less voltage than what’s
required (or spiking due to faulty condition) or the instability of
overheated MOBO and/or card leads (case temp too high) is causing the
flow of either data and/or current to be unstable. This is just a
guess but following the above advice is a free opportunity to prove
the theory right or wrong. Just don’t touch any internals without
being grounded and don’t allow anything to fall into the case while
internals are exposed.

Happy troubleshooting

I did run this CPU/mobo overclocked at a stable 3.33GHz instead of
2.66GHz in the past, but it is currently running at stock speeds.

On the balance of probabilities do you think it is more likely to be
the mobo or PSU?

 

[Amaranth]

what cpu cooler are you using and what is the temp of the cpu when playing
games

I know the CPU temperature right after a freeze is about 68C. It is
the stock cooler but I am running at stock speeds and it isn't
overclocked.

Is there any tool that continuously writes to HDD the CPU temperature
so that in the event of a freeze I can look at the last known reading?

 

[Amaranth]

Question, is the power supply adequate for the graphics cards you’re
running?

PSU is a 600W Sansun BTW.

 

[Amaranth]

What are your motherboard northbridge temperatures?

If those are within reasonable ranges, it is more likely the
PSU. It's brand I've not heard of, meaning it's probably a
generic of questionable quality and true capacity. On the
other hand, the factor that makes this diagnosis
questionable is that it still freezes when the low-powered
video card is installed instead of the 9600 GSO. Usually
software just crashes, does not make the system freeze, but
it is still possible so I suggest testing with some other
game to see if you can reproduce the problem.

The freezes also happen with the open source game UFO:AI. These are
the only two games I'm playing ATM.

So far the freezes have never happened when I'm in the desktop
working, and I spend a lot of time in the desktop (probably 10-14
hours) as opposed to games (2-4 hours).

If it is a dodgy mobo or PSU, why should it only happen in games and
not when stress testing or in the desktop?

By the way, I'm not sure if it is relevant but in some of the crashes,
the PC speakers make a squeak or static-sounding "bzzzt" sound.

 

[Amaranth]

What resolution are you playing the games at and what happens when playing
games at different resolutions? how big is you screen and what is it's max
resolution?

I have a dual monitor set up.

I'm playing at 1024x786 @ 85Hz for both monitors, using a 19" & 17"
CRT. Max is 1600x1200 and 1280x1024 respectively.

 

[westom]

I have a dual monitor set up.

I'm playing at 1024x786 @ 85Hz for both monitors, using a 19" & 17"
CRT. Max is 1600x1200 and 1280x1024 respectively.

As Kony suggested earlier, you could have multiple problems. And
you did nothing to move components from an unknown to a known good
list.

Well, Memtst was the right idea. But defective memory often does
not appear until Memtst is executed while memory (an associated
interface chips) are heated to temperatures that all semiconductors
love. Use a hairdryer on highest heat settings to heat memory to
temperatures that all semiconductors love - that is uncomfortable to
touch and yet does not burn skin. Only then does Memtst put memory on
the list of 'known good' components.

As noted, 68 degrees is high. But get the actual manufacturer
number. Remember, any PC must operate pig’s heaven happy even in a
100 degree F room. Therefore your processor must be sufficiently below
that max temperature at room temperature so that it is also perfectly
happy in a 100 degree F room.

You still have no idea is the power supply 'system' (not just a
power supply) is good until voltages are measured. That means
measuring DC voltages on any one purple, orange, red, and yellow wires
when the system is most stressed. Not when running Prime95. That is
not stressful. Execute a complex graphic (ie a movie), while
downloading from the internet, while playing a CD, while searching the
hard drive, while moving the mouse, etc. Maximum load occurs when all
peripherals are operating. Now numbers from those power supply to
motherboard wires are useful. And report those numbers here because
those numbers contain other facts you would not know.

Only with those numbers do we know that all components of the power
supply 'system' are definitively good. Only then can we list those
many components on a list of accomplishments and move on to other
suspects.

Appreciate how this works. Remove nothing. Determine what is
definitively (without a doubt) good by using diagnostics and numbers.
Then move on to other suspects.

Also relevant (I did not see it) are reports stored in the system
(event) logs. Were hardware problems reported there as Windows worked
around those problems? Again, critical information.

Considering the CPU temperature, what was the degree C per watt
number for that heatsink assembly? A number necessary to learn if the
heatsink is properly sized or simply not properly attached.

Remember, you could have multiple problems. And the number of
problems would then only increase if you tried to solve them using
shotgunning. Swapping parts assumes only one failure exists AND can
exponentially complicate things. You want to build a list of what you
know is definitively good. Only that list measures accomplishment.
Swapping parts sometimes even creates negative accomplishment.