"System failed CPU test" ??

[mikedude]

Hi everybody,

I've been having some issues with my system the past 2 days and am hoping somebody can help me with them.

My relevant system stats:
ASUS P4P800
Intel P4 2.8
GeForce 5700 Ultra

I left my system on all last week while I was gone on a trip to Vegas. When I returned yesterday morning, there was a notice saying some XP updates had installed and I needed to reboot. Before I could even move the mouse pointer to click the reboot button, the system rebooted itself. It then got about halfway through loading stuff in my taskbar before it rebooted itself again. This went on 2 or 3 more times and I finally just shut the damn thing off, figuring I'd deal with it after I got some sleep.

When I turned it back on in the morning, XP booted up just fine. However, about 10 minutes in, the screen output went crazy (random blocks and lines of color) and a voice started playing over my speakers but it was too low to hear what it was saying. I rebooted into safe mode and the same thing happened again, only this time the voice was loud and clear "System failed CPU test". I immediately shut down and reseated both the CPU and all cards in the system.

When I turned the computer back on I confirmed that all fans were working and not blocked by dust. It booted back up into XP and worked fine for a little while. I checked all the power supply voltages using ASUSProbe and they all checked out ok. Then all of a sudden the screen lost signal. It came back a few seconds later but then the computer rebooted itself. It lasted again for a while longer before losing screen output again, only this time when it came back it was in a really low resolution with a notice saying that my display drivers had stopped responding normally. I updated the drivers a few weeks back, so I went ahead and rolled them back.

XP lasted for several hours the next time and I thought I'd solved the problem, but then the screen started losing output and the system rebooted itself. Atleast there was no "system failed cpu test" error this time. It lasted all the way through the night on the next boot, but this morning I momentarily lost screen output a few times. For any video playback, the video stream was just a bunch of random bars and blocks of color, but the sound played fine. My mouse pointer also would double or triple in size and become a series of yellow and magenta dots when hovering over certain parts of windows. In Windows Explorer and during the XP shutdown screen, several images were displaying as just big black boxes rather than actual images. I just rebooted a while ago and everything seems to be running ok right now (including video playback), but I'm at a real loss for what to do.

I want to think the graphics card is going bad due to all the crazy screen output, but I don't really get why that'd give me a CPU error (and why the error would suddenly disappear) or why I'd get normal screen output other than video files.

Any help or suggestions would be greatly appreciated.

 

[foozy]

What do your temperatures look like while all this junk is going on?

edit: check the power supply and see if its wiggin' out

 

[mikedude]

I've been monitoring all of my voltages with ASUSProbe and they all check out ok. Granted, none of the readings are exactly right before problems occur, but since I don't know how I could go about inducing the problem the best I can do is periodicially check them as I'm running normal applications.

As far as temps go, my motherboard temp is pretty consistant around 90 F. CPU temp varies, but I haven't seen it go higher than 120 F (when under load from a H.264 movie file).

As for the display problems, I found this site which shows examples of common graphics card problems. From the looks of things, I've been experiencing both GPU artifacts and bus artifacts. I guess I could try reducing to AGP 4x or underclocking my GPU and seeing if that helps alleviate the issues.

 

[nickslick74]

Another option would be to install a different AGP card and see if the problems go away. You may not have on laying around, maybe borrow one from a friend?

 

[mikedude]

Nope, don't have an extra AGP card. I was hoping there would be some way to test the card and motherboard short of having to just replace them and see if the problems went away.

 

[mikedude]

EDIT: Nevermind, answered my own question.

 

[beef_jerky4104]

And what was the problem. Maybe someone else needs to know.

 

[mikedude]

Oh, its that I was running a stress test (Prime95) and I couldn't figure out why I wasn't getting more than 50% CPU usage out of it. Its because my P4 has hyperthreading enabled, so you have to actually run 2 instances of the stress test at the same time to max out the CPU. Its been running now for about 45 minutes with no issues (other than setting off the temperature alarm as ASUSProbe, but that's because I decreased the max temp a few degrees from what Intel recommends just to be on the safe side) so that's good so far, but I know that sometimes it'll take hours before a CPU problem will appear.

I know everything so far has been pointing to the graphics card, but I'm just doing any checks I can on other potential causes just to eliminate things before I start the process of trial and error with replacement parts. Assuming the CPU checks out, its down to the graphics cards or the motherboard in general. And since I don't know any way of checking the motherboard, I'm gonna have to hope its the graphics card (which is also the cheaper of the two to replace..).

 

[mikedude]

Well, ran fine on Prime95 for several hours, but then the screen output went all crazy again (I heard a 'click' coming from the tower when this happened). Rebooted again and started getting artifacting all over the screen, followed by the return of the dreaded "SYSTEM FAILED CPU TEST" voice.

I'll be off to Best Buy tomorrow to pick up a random AGP card and see if that solves the problem. I pray to God it does..

EDIT: And now there is a new symptom.... after trying to reboot the system back up, I got an alert telling me "The NVIDIA System Sentinel is reporting that the NVIDIA-powered graphics card is not receiving sufficient power. To protect your hardware from potential damage or causing a potential system lockup, the graphics processor has lowered its performance to a level that allows continued safe operations." This was, of course, followed by the screen completely freezing up. Now.... should I still take this to mean the card itself is going bad.. or should I be looking into trying a replacement power supply instead? I've been checking my voltages this whole time and they've always been ok, but this alert would seem to suggest an intermittent problem with power output (perhaps the 'click' sound I hear from the tower when the problems manifest?).

 

[mikedude]

Well, I replaced it with a Radeon X1050 that I picked up at Best Buy (cheapest AGP card I saw). So far so good..

If it holds up for a couple of days I guess I'll return it and pick up something better off of Newegg.

 

[mikedude]

Well.. it wasn't the graphics card.

It was good for a while, but then everything started getting choppy. The mouse movement was glitchy, and video and audio files were choppy as they played. Restarted and since then its been nothing but problems. On one boot I even had Windows running ok for a while, but the whole time the "system failed cpu test" voice was playing over the speakers. The system lasts for a variable amount of time on each bootup; either it reboots itself or it freezes and I do it myself. Sometimes its just a couple minutes.. on the current boot its been up for about 20 minutes though. I managed on one boot to queue a scandisk (well, XP's version of scandisk) on my C drive and didn't come up with any bad sectors or anything. I also did a full RAM test that also yielded no errors.

I did notice this weird clicking/popping sound when things started going screwy. Sounded different from an HD though.. maybe its the sound of something shorting out? That'd make it sound like the motherboard, but that's really weird given how variable the problem has been the past couple of days. I thought when motherboards went that they just totally died.

I don't know where to go from here. I could try a new power supply next, even though my voltages were registering just fine. I could try new RAM, even though I haven't been able to get any errors on tests and it wouldn't really account for some of the problems I've been having (especially the warning about the graphics card supposedly not getting enough power). I could also try to track down a replacement socket 478 motherboard, but that almost seems like more of a hassle than its worth.

 

[MetalX]

I'd say it's time for a new motherboard. Get a new motherboard, and if that doesn't solve the problems, keep the new motherboard and get a new CPU too. When strange things like that happen to a computer, I always tell people to replace as many parts as they can.

 

[HPCE_Larry]

How long are you monitering the voltages? Is there anyway to log them over a period of hours? The power supply could only be having intermittent errors. The mobo may be having trouble as well, because it would be the thing actually giving the power to the video card.

 

[mikedude]

I've been monitoring the voltages using ASUSProbe. It sets off an alarm if anything goes more than 5% out of expected voltage. Though I guess if the problem is intermittent, I'd imagine the system would lock up before the program could alert me to it? Is there a better way to monitor the voltages more directly?


The problem with then moving onto a new motherboard is finding someone that still makes a good socket 478. My current motherboard is 3 years old and I don't know how easy it'll be to find an appropriate replacement. I feel like if I'm going to be buying all these random replacement parts hoping they fix the problem, I may as well just bite the bullet and build a new system altogether.

EDIT: Also, with regards to my card saying it wasn't getting enough power, I wasn't really sure what that would indicate. Yes, the card is hooked up to the motherboard through the AGP, but it also has a separate power connection directly from the power supply. Thus I didn't think the warning alone pinpointed a particularly part as the problem.