[rx7_fan]

I love to root for ATi because they are the underdog, but sometimes, I wish they would do a better job with their drivers.

 

[jnev]

Quote:

Originally Posted by asmox Could also be your powersupply. Check the rail voltages.

yup, that's what i was thinking. make sure that your rails are stable and not fluctuating. if they are, time for a new PSU.

 

[NiceCans]

I had a power supply go a while back, it started out like a heat problem, but it quickly got shorter and shorter between my PC rebooting itself.
No errors, no out-of-ordinary tempurature indications, just sudden instant reboot. Of course no errors as the PC was off before it could write them.
Luckily when I wiggled the wires where they leave the power supply I could cause the reboot so knew I was onto it.
Strangely, it was not a bad connection or a short, the caps had cooked and had been leaking, as they got drier they shorted faster. Only the change in timing between the reboots and the wiggly wire led me to it before it died completely.
Good Luck.

 

[dj_mocok]

The Memtest suggestion is pretty good actually. MY second guess would be memory, but the thing is, on weekends when I am at home, I can pretty much run my PC on non stop for more than 12 hours and nothing will happen, works perfectly fine.

But then as soon as I start to play game, it will freeze or restart within 10 minutes or 1/2 hr. So I guess I can rule out overheating eh? Besides I got like 20cm empty spaces at the back of the CPU and no messy clutter inside the CPU.

About CPU, I replaced the stock AMD fan with 3rd party one, which is better than AMD one. Everything runs at stock speed since I don't really know how to overclock.

But I will do this memtest later to figure out whether it has anything to do with the RAM or not.

Illah, ABout the Memtest, which memtest are you talking about? I ran it in Google and it has a few.

 

[Illah]

Memtest86
http://www.memtest86.com/

You'll need to do a bootable floppy or CD for that prog. As for heat, heat output jumps dramatically under load. A game is a pretty CPU/GPU intensive program, so that can push the limits of an otherwise stable system. If it's on for 12 hours just doing web and other light tasks that's one thing, but once you start doing video editing, 3D work, gaming, etc. that really exposes any weak spots.

--Illah

 

[dj_mocok]

I just finished the memtest using this Memtest, did it for one hour as the manual suggested (it was around 340% coverage), and nothing happened.
So I guess I can rule out RAM at the moment then...

Is there any program that tests only the graphic card? I'm not sure whether 3Dmark only tests graphic or the overall CPU performance in general.

 

[eric.w]

As far as I know, any program which tests your graphics card will stree the cpu to some extent, so you can't really test the graphics card on it's own. If your ram's okay, next step is to see if the system can handle several hours of Prime95's torture test with no errors.

 

[asmox]

Which tests did you run with Memtest?

I had the exact same problems that you are having (although with different applications), and they went away after I slightly loosened up my memory timings. You mentioned that the system restarts on its own every now and again even when you aren't playing games.. that's a very convincing argument for faulty RAM. I should also mention that one hour isn't enough for a proper Memtest run.. I've had RAM that didn't generate any errors until 12+ hours of Memtest.

So, again.. try running with one stick, and then try the other one.

Go into your BIOS and note the voltages on your +3.3V, +5V, and +12V rails.. what do they read? An unstable power supply is also a very common culprit in random lockups and ghost system restarts.

For other stress testing utilities - grab OCCT, SuperPi, and Prime95 (as mentioned above). Grab Motherboard Monitor, as well.

 

[Illah]

Yeah do Prime95 next. Let that run for a couple of hours. That will hit the CPU/RAM really hard. Let it go overnight if you can. If that is OK, then do 3DMark. If it really only restarts with video intesive apps then it could be the video card.

PSU is also a possibility but, from my experience anyway, I've done severe overclocking with nothing but generic PSU's for years and never had a problem. Unless your power rails are *WAY* off I find the PSU thing to be somewhat of an old wives tale. When I built 24/7 fileservers at my last company with multiple P4 or Xeon processors, 4GB RAM, and RAID-5 storage they used generic 300W PSU's that came with the case. No brand name fanciness there, and no prolems whatsover (well, none that had anything to do with the PSU anyway). Dell and all the other major manufacturers use generics, and most of those machines work just fine for years.

Brand name aftermarket PSU's probably represent 0.5% or less of the total market for PSU's. The only reason I bought one was for silence and efficiency, not power rails! The minimization of power waste lowers my electrical bill and therefor uses less power, which is good for the environment.

--Illah

 

[star882]

Quote:

Originally Posted by Illah Yeah do Prime95 next. Let that run for a couple of hours. That will hit the CPU/RAM really hard. Let it go overnight if you can. If that is OK, then do 3DMark. If it really only restarts with video intesive apps then it could be the video card. PSU is also a possibility but, from my experience anyway, I've done severe overclocking with nothing but generic PSU's for years and never had a problem. Unless your power rails are *WAY* off I find the PSU thing to be somewhat of an old wives tale. When I built 24/7 fileservers at my last company with multiple P4 or Xeon processors, 4GB RAM, and RAID-5 storage they used generic 300W PSU's that came with the case. No brand name fanciness there, and no prolems whatsover (well, none that had anything to do with the PSU anyway). Dell and all the other major manufacturers use generics, and most of those machines work just fine for years. Brand name aftermarket PSU's probably represent 0.5% or less of the total market for PSU's. The only reason I bought one was for silence and efficiency, not power rails! The minimization of power waste lowers my electrical bill and therefor uses less power, which is good for the environment. --Illah

Some of the generics are actually very good. It all depends on the OEM. Channel Well is used in many generics, Antecs, and Dells. It's one of the best ones available (some other good OEMs include Fortron, Delta, and Mitac). Just keep in mind that many Antecs have the fans wired to less than 12v (likely to reduce dust buildup), and therefore may actually be worse than Channel Well generics.
When picking generic PSUs, pick one that's heavy for the power rating. There's a good chance that it will get you a nice unit. I used a $5 350w Channel Well generic in my dual core P4 server, and it has worked well. I was able to run Fear Factor for 24 hours on that system and it has ran for several months constantly folding.

BTW, if you really want a CPU that fries many PSUs, go for a Britney. I recall a 550w Antec PSU that was used in a dual Britney workstation. It worked running memtest86 overnight, but a minute after running Fear Factor, it burned out. They ended up replacing the Antec with a 650w PC Power & Cooling.

 

[dj_mocok]

I'm at uni now so can't really do the Prime95 test. When I get home tonight and have time, I'll try to run a couple hours of Prime with my PC and see what happens then.

I don't think it's a PSU issue as well.

I'm not sure if this is related or not, but my videocard used to make a VERY loud whirring noise, and I took it to the store, and the guy cleaned the video card fan from dust buildup.
But after I took it home, it still made that sound, so I suspect it's the joint issue or whatever. Took out one of the screw from the videocard, and it hasn't made any whirring sound anymore.

But MAYBE for some reasons after that incident, the fan is not really rotating as fast as it should be thus the instability issue, or maybe I am just unlucky and got a P.O.S. video card.

the card is GeXCube ATI Radeon 9600 XT Extreme.
It's a slightly faster card than the normal 9600XT.
I really like this card (when it's fine) but now I'm so annoyed with it. I'm not sure whether this company is really dodgy or I am just unlucky getting the dud one (less than 2yrs old), or maybe it's not a video card problem altogether, not sure.

Here are the link for the videocard:
Link 1
Link 2

 

[star882]

Quote:

Originally Posted by dj_mocok I'm not sure if this is related or not, but my videocard used to make a VERY loud whirring noise, and I took it to the store, and the guy cleaned the video card fan from dust buildup. But after I took it home, it still made that sound, so I suspect it's the joint issue or whatever. Took out one of the screw from the videocard, and it hasn't made any whirring sound anymore. But MAYBE for some reasons after that incident, the fan is not really rotating as fast as it should be thus the instability issue, or maybe I am just unlucky and got a P.O.S. video card. the card is GeXCube ATI Radeon 9600 XT Extreme. It's a slightly faster card than the normal 9600XT. I really like this card (when it's fine) but now I'm so annoyed with it. I'm not sure whether this company is really dodgy or I am just unlucky getting the dud one (less than 2yrs old), or maybe it's not a video card problem altogether, not sure.

Probably a bad bearing. Try oiling the fan bearing using silicone machine oil (regular oils will eventually gum up).

 

[dj_mocok]

I did 2 hours of Prime95, and no problem at all.
After I stopped at 2 hours, I did that benchmark test, and looked at the website for the average time for Athlon XP 2500, my results are generally faster than the average. (faster = better right?)

So now what, most probably video card now?

Too bad I don't have spare cards to check.

 

[Born2bwire]

Quote:

Originally Posted by dj_mocok But then as soon as I start to play game, it will freeze or restart within 10 minutes or 1/2 hr. So I guess I can rule out overheating eh? Besides I got like 20cm empty spaces at the back of the CPU and no messy clutter inside the CPU.

That sounds like a classic overheating problem to me. My guess would be that maybe the thermal paste between the GPU and the heatsink has dried out and lost its effectiveness. I would try cleaning the heatsink and GPU, applying a new spot of thermal paste, and reseating the heatsink. You could also try reverting to an older driver version that you used before the problem arose to see if it is in the software.

 

[titaniumx3]

The best way forward here is to either try out a new graphics card on your system so as to ensure the rest of the computer if fine, or try out your card on another computer. You really need to monitor your video card core/memory temperaturesn somehow (I normally use Lavalys EVEREST home). If the temperatures are excessively high (in comparision to idle temps) straight after quitting a game, you know you have a overheating problem.