alamarmalomar

Howdy everyone, this is my first post here. I'm essentially migrating from forum to forum, looking for ways to solve this problem that lots of people, including myself, seem to be having with their ATI AGP card.

First off, system specs and maybe some relevant information:

Computer is a Compaq SR1230NX with an AGP Radeon X850XT installed and another 512 MB RAM stick installed. Both additions were installed in March 2006.

RAM: 1 GB PC2700 RAM (2 x 512MB)
Video card: AGP Radeon X850XT
Processor: AMD Athlon 3200+
PSU: Not exactly familiar with the brand, but it's 450 watts I believe. I have Everest trial version, is it possible to find it hidden in there somewhere?

CPU runs at 53'C when idle.
GPU runs at 45'C when idle.


Timeline:
1. A week or two ago, my Steam games refused to start, giving me the error "Failed to Create D3D Device!". Examining their support page, they suggested updating DirectX and updating my drivers. Fair enough, I hadn't updated my drivers in a while and DirectX released a new version a week prior to this error.
2. I install the latest version of DirectX and update my drivers to 8.9 Catalyst.
3. I play Half-Life 2, and while playing through it, my computer restarted about three times through. Shrugging it off, I think nothing of it.
4. The error occurs in Garry's Mod 10. Now, I play on a friend's server with a bunch of other friends of mine, and this error is starting to get on my nerves. Play it a few more times and it keeps restarting.
5. I buy Warhammer Online and install it and play. It's...relatively stable. As stable as this machine can run it. At least it doesn't crash, and the game is mad fun. However, a week after that, I get a restart. I got my second one today, but I play this game a ton, so it's relatively not that bad.
6. I start going through various tech support forums, and I learn the ways of memory dumps. Searching through them, I noticed the Half-Life 2/Garry's Mod 10 errors are caused by ati3duag.dll, while the Warhammer ones are caused by ati2cvag.dll (I believe, purging my computer of wasteful material apparently got rid of my old .dmp files, more on that later).
7. A solution to use driver cleaners since ATI isn't good with installation is proposed. I try the manual solution first by tinkering with the registry, but no dice (Probably a bad idea in retrospect). Fair enough. I actually never cleaned up old drivers on this machine, at least between reformats (I've reformated twice). Using the latest version, I clean my computer. A lot. I'm sick of the restart button now. Still, the same errors occur.
8. I learn that for my video card, the latest drivers aren't necessary the best. Uninstall, driver clean, etc. I do this with the 8.8 drivers (current version of Catalyst is 8.9), there's a restart error. 8.8 AGP hotfix, restart. 8.9 AGP drivers, restart. 7.10 drivers (apparently they're the best for my card), my old D3D device error resurfaces. 7.11 drivers, D3D device error resurfaces. Omega drivers (forget which versions at the current moment), restart. Thorough driver cleans between each installation.
9. I find the touted Windows updates that supposedly fix the problem. The creator of it apparently had these restarts with explosion sounds and the dump related it to the ati3duag.dll file. I then have an epiphany, and realize all the Garry's Mod 10 errors arise with explosives, or so I think. To test this, I mess around in Garry's Mod, making damned sure I don't use any explosive. After 15 minutes or so, I spawn a pod and 8 sticks of dynamite. I get in the pod so I don't die and detonate the dynamite repeatedly. Computer restarts in about five seconds. Did the same thing with a meteor storm SWEP and a nuke SWEP. Restarts. These times, the computer beeps for about 3 seconds, beeps for half a second and stays on, but nothing shows up. I turn it off and wait and then turn it back on. Still an ati3duag.dll error.
10. I install the updates, thinking this will actually fix it. No dice. Getting really annoyed at this point. I reinstall the regular 8.9 Catalyst drivers. I find out this may be a power supply problem due to the explosions restarting the computer instantly.
11. I find out about CCleaner from a trusted person on DriverHeaven.net. I try it out. I clean my computer thoroughly. Still no dice.

So here I am, with a ton of what was supposedly useless files cleaned out. I have the 8.9 drivers installed, and the restarts still happen. I can't play Half-Life 2, Episode 1, Episode 2, or Garry's Mod 10 without restarts. Warhammer restarts (but rarely). I haven't run memtest86 due to a lack of blank CDs at the present moment.

The question is, based on this, is this a video card problem, RAM problem, or a power supply problem? What else can I do to test this?

Tumbleweed36

Sounds to me like a heat problem from the temps you reported. Your rig is way too hot IMHO.

Just to double check, download and run SensorviewPro and report your temps and voltages as soon as you boot up, then play a game and check the temps. Let's use this for comparison with what you just posted.

BTW, have you considered changing the thermal paste on this rig?

alamarmalomar

I have, actually. A friend of mine suggested it may be a heat problem. I've never changed the thermal paste on this thing, and I've had it since 2004.

Right now, with Youtube running a video, Ventrilo and Trillian running, here are my current statistics:

CPU: 56'C/133'F
SYS: 42'C/108'F
HD0: 35'C/95'F
CPU Usage 10%
Memory Usage 62%
HDD Usage 0%
NIC Usage 1%

CPU Fan: 1089 RPM
SYS Fan: 2281 RPM

VCoreA: 1.65V
+3.3V: 3.34V
+5V: 5.16V
+12V: 11.40V
-5V: 5.11V
-12V: 0.64V
+5VSB: 4.97V
+VBAT: 3.30V

Tumbleweed36

If under stress, those temps don't seem as bad as reported earlier. Also, check the bios temps, because they are normally the most accurate. I do think if this were my rig, I would change the thermal paste and see what happens with the temps. Here is a link if you decide to do that:

Applying Artic Silver Thermal Paste

alamarmalomar

Well, damn. With Warhammer and Ventrilo running, I get these readings:

CPU: 61'C/142'F (+6'C/+9'F more than before)
SYS: 47'C/117'F (+5'C/+9'F more than before)
HD0: 37'C/99'F (+2'C/+3'F more than before)
CPU Usage 82% (+72% more than before)
Memory Usage 93% (+31% more than before)
HDD Usage 100% (+100% more than before, holy cow)
NIC Usage 1%

CPU Fan: 1133 RPM (+44 RPM from before)
SYS Fan: 2312 RPM (+31 RPM from before)

VCoreA: 1.65V
+3.3V: 3.34V
+5V: 5.16V
+12V: 11.40V
-5V: 5.11V
-12V: 0.64V
+5VSB: 4.97V
+VBAT: 3.30V

Lots of resource usage. Is this any kind of indicator of anything?

Tumbleweed36

I see two potential problems. First of all, you are way to hot, it should not get over 60c ....change the thermal paste and talk to me about your fans...what fans, where located, which direction are they blowing.

The +12 volt line on your power supply is weak. +11.40 is the lowest acceptable power that you should be running, so that is also an issue. Take the side of the case off and see what the brand name is and what wattage it is.

Report back with questions/concerns.

alamarmalomar

I took the side off, but nothing appears on my power supply unit. Is there any way I can tell with software what my power supply is?

As for the fans, I can't remember the brand names, but one is blowing out the back (I think that's the default one) and one is near the top middle blowing towards the side (which seems dumb since the vent's at the bottom-middle of the tower, but eh).

Tumbleweed36

Fan placement:

Back Fan - pulling hot air out of case and throwing it out the back of the case
Front Fan - Pulling cool air in the front of the case.
Side Fan - Blowing in to the case and pulling cool air in to the case.
Top Fan - Pulling hot air out of the case (heat rises)