Cab Crashes...What to do...What to do...

[bkenobi]

I've had my cab up and running for over a year and have been working on a random hang-up problem for that time as well. For reference, I'm running an Athlon XP 3000+ with 512Mb (or maybe 1Gb, I forget...I'll check) and an ATI 1300pro. The monitor is a 27" TV, so it adds a bit of heat and air flow restriction, but it's not the problem.

I originally thought it was heat related, but after installing a pair of 80mm intake fans pointing at the open back of the case and installing a vent pipe to duct the heat away (via the psu), I still have crashes. I ran speedfan and recorded a log to see if anything was going on over time or just before the crash, but everything looks fine (cpu fan speed, all voltages, temp is a bit high, but level at 51C). I figured the only other thing it could be was bad ram, but I have been running MEMtest86 for 22 hours with 24 passed tests and 0 failures.

Random crashes are usually heat, bad ram, or some type of memory hole (though less random). I've seen crashes in times as little as a couple hours or as much as overnight. I'm really just looking for suggestions of what to try. I was thinking a general system burn in test, but I can't seem to find a freeware/shareware one.

Any thoughts?

[Nologic]

Are you using a stripped down version of windows?...do you have your services tweaked...do have any emulators playing in screensaver mode...and if you do...are they up to date?

Have you ran an adware & spyware application on the system to check if something may have slipped in?

[Creek]

If you haven't already, I would suggest you tried a different PSU and/or a different graphics card. Even though the voltages look alright, it doesn't rule out that the PSU is at fault. And I've had several illogical problems with graphics cards over the years, so it wouldn't surprise me, if that was the cause.

But generally, any piece of hardware connected in or to your pc, could be at fault.

[LB11]

Try running orthos for 24hours as well... Should tell you if its a CPU problem at least.

http://www.overclock.net/downloads/138142-...-v20060420.html

[bkenobi]

Full version of XP Pro with some services disabled. I don't think they should be the problem, but I could go back to default settings on them. No emulators running in screen saver and it only locks in MAME screen saver mode. Haven't checked ad/spy ware as I don't really go online with this machine. It's running Avast anti virus which catches a lot of that stuff anyway.

Haven't tried a new PSU recently, but I have a different one I could try. I don't think it's the video card (unless it's drivers for some reason) since I have another one that works fine. This one was running for a long time as my main video card in my primary PC, so I don't think it should be an issue. I guess I could do a quick swap of the two cards since they are identical though.

Anyone know of a software package that I could run that would stress test the cpu, video card, etc? I don't have a long term use for it, so I was hoping for a freeware/shareware/trial ware package. Thanks for the help!

[LB11]

Orthos will do CPU and memory at the same time, for graphics, I guess one of the 3dmark suites. There is also this program that nvidia had me use when I had a problem with one of there cards.

VideoCardStabilityTestSetup.zip

[HooPZ]

Anyone know of a software package that I could run that would stress test the cpu, video card, etc? I don't have a long term use for it, so I was hoping for a freeware/shareware/trial ware package. Thanks for the help!

Been awhile since I used it, but SiSoft Sandra is free. Not all features are available, but it tests almost everything. Not sure if there is anything better now or not. May be worth a shot.

[bkenobi]

I'll give these a try and see what it says. Thanks!

[LB11]

Also, like someone mentioned. The power supply can be a pig part of crashing. If you have a spare I would try that before I bothered with anything else.

[bkenobi]

I know about PSU problems being hard to detect! I had one that seemed fine, but I started getting a random system beep once in a while (once or twice a week when I was at the machine). I never found the problem until a drive started to die. I was in bios and noticed one of the voltages randomly drop out of spec and turn red just as it beeped. I swapped PSU's but it was too late for the drive...lost ALL of my important documents and 99% of my photos!

I'm running one of the stress tests now to see if I can get a hang up. If not, I'll try the PSU swap and the other two tests suggested. Thanks!

I hate these unpredictable problems. They are so hard to diagnose. That's why a lot of mechanics tell you to just wait till something fails. It may be a more expensive fix part-wise, but in terms of labor, it's much cheaper.

[jthompson]

I know about PSU problems being hard to detect! I had one that seemed fine, but I started getting a random system beep once in a while (once or twice a week when I was at the machine). I never found the problem until a drive started to die. I was in bios and noticed one of the voltages randomly drop out of spec and turn red just as it beeped. I swapped PSU's but it was too late for the drive...lost ALL of my important documents and 99% of my photos!

I'm running one of the stress tests now to see if I can get a hang up. If not, I'll try the PSU swap and the other two tests suggested. Thanks!

I hate these unpredictable problems. They are so hard to diagnose. That's why a lot of mechanics tell you to just wait till something fails. It may be a more expensive fix part-wise, but in terms of labor, it's much cheaper.

Don't rule out the ram in the system being at fault. I've seen some pretty weird stuff with memory and the only test that detected the memory problems was running windows itself. I can't tell you how many times I've seen memory tests pass, but the problem still there until the memory gets replaced. But definitely try a different video card, and for my money I would say the power supply would be my last swap test. Power supplies can do some weird shit like lb11 described, but in my experience they are usually all or nothing. I have also personally experienced a hard drive that I had that did the same thing you were describing, but mine was the drive and not the PS. I got an exact replacement for it as it was still under warranty and still have it in my computer today, and that was three years ago. My hard drive was the main drive and the system would randomly beep and reboot with no warning whatsoever. It could take two hours or two days before it would do it again and it was very frustrating and hard to figure out what the hell it was until my hard drive really shit the bed one day. I haven't had that problem again since replacing the drive and I never replaced the PS. Any ways what I'm getting at is don't rely too much on stress testing it as swapping stuff out will probably save you some time if you have the extra parts lying around. Like Creek said "any piece of hardware connected in or to your pc, could be at fault". Swap stuff out and narrow the possibilities down.

Jay T

[Creek]

The Ultimate Boot CD has everything you need for testing.

Though, trying the different PSU and GFX card before starting the tests could save you a ton of time.

[Brian Hoffman]

Id get a diffrent stick of memory and try that.

Especially if you mix and match brands.

[bkenobi]

I can't remember what I put in it, but I think it's a pair of 512mb KVR. I'll have to check, but that might be it... I ran memtest for over 24 hours and the video card stress test for overnight and had 0 problems with either. Memtest is dos or linux based. The video card test is windows based. So, it seems stable with both. I launched Orthos Prime (CPU and Ram tests) this morning for a 24 hour test. If it's still running at the end, that will be over a day and a half with no errors under harsh test environments on memory, cpu, and video cards. The PSU could be an issue, but perhaps not. I wonder if it's possible there could be a problem with MAME itself that could cause a hang? It only hangs in MAME afterall (well, I've only really seen it in MAME when it's in screen saver mode of GameEx). Perhaps some bad roms causing it? It's a good set from PD, so I would bet against it, but...

[jthompson]

I can't remember what I put in it, but I think it's a pair of 512mb KVR. I'll have to check, but that might be it... I ran memtest for over 24 hours and the video card stress test for overnight and had 0 problems with either. Memtest is dos or linux based. The video card test is windows based. So, it seems stable with both. I launched Orthos Prime (CPU and Ram tests) this morning for a 24 hour test. If it's still running at the end, that will be over a day and a half with no errors under harsh test environments on memory, cpu, and video cards. The PSU could be an issue, but perhaps not. I wonder if it's possible there could be a problem with MAME itself that could cause a hang? It only hangs in MAME afterall (well, I've only really seen it in MAME when it's in screen saver mode of GameEx). Perhaps some bad roms causing it? It's a good set from PD, so I would bet against it, but...

The next time it happens try the run last game in the gameex start menu and see if you can duplicate the error. On second thought though, I'm not sure if that will actually run the last game you initialized or if it also runs the games ran in the screen saver. Worth a shot though.

Jay T

[bkenobi]

Yeah, I had thought of that, but haven't tried it yet. I think the log does record what the last game run was, so I just have to boot up without booting GameEx (so it doesn't overwrite the last log)...which may be a challenge since the machine is shelled. I'll probably just write a script to copy the last log file to a backup folder so I can view it on the next boot without have the chance of losing it.

I'll know in the next couple hours if it crashed during stress testing (still at work), so I can give that a try too if it hasn't.

[bkenobi]

Hmmm, it only took an hour to crash Orthos

I don't understand how to interpret the log though:

Type: Blend - stress CPU and RAM Min: 8 Max: 4096 InPlace: No Mem: 767 Time: 15
CPU: 2163MHz FSB: 166MHz [166MHz x 13.0 est.]
10/15/2007 6:36 AM 
Launching 1 thread...
1:Using CPU #0
1:Beginning a continuous self-test to check your computer.
1:Press Stop to end this test.
1:Test 1, 4000 Lucas-Lehmer iterations of M19922945 using 1024K FFT length.
1:Test 2, 4000 Lucas-Lehmer iterations of M19922943 using 1024K FFT length.
1:Test 3, 4000 Lucas-Lehmer iterations of M19374367 using 1024K FFT length.
1:Self-test 1024K passed!
1:Test 1, 800000 Lucas-Lehmer iterations of M172031 using 8K FFT length.
1:Test 2, 800000 Lucas-Lehmer iterations of M163839 using 8K FFT length.
1:Test 3, 800000 Lucas-Lehmer iterations of M159745 using 8K FFT length.
1:Self-test 8K passed!
1:Test 1, 560000 Lucas-Lehmer iterations of M212991 using 10K FFT length.
1:Test 2, 560000 Lucas-Lehmer iterations of M210415 using 10K FFT length.
1:Test 3, 560000 Lucas-Lehmer iterations of M208897 using 10K FFT length.
1:Self-test 10K passed!
1:Test 1, 4500 Lucas-Lehmer iterations of M17432577 using 896K FFT length.

Oh, and just to point it out, I have 1Gb of ram installed (2x512Mb).

[bkenobi]

Oooohh, I may have figured out big problem! This may not be *the* problem, but it's definitely not a good thing...

Looking at Speedfan, I just noticed that my -5V is only supplying +1.7V... Check my math, but that's not the same as desired

[LB11]

The log doesn't really mean anything. Its just the type of calculation its doing. All it does is try to find prime numbers and matches it to known prime numbers. If the calculation matches, the test continues. If not, the CPU made the wrong calculation. The larger FFT use alot more memory and a miscalculation could be the result of the memory supplying the wrong info as opposed to the CPU calculating wrong.

Most likely its your CPU that is failing. Possibly because of the power supply.

Change out the PSU and try again

Just to mention though, speedfan does not always read the voltages correctly. A lot of motherboards don't read correctly either, especially the - voltages.

If the power supply doesnt work, try lowering the FSB or CPU multiplier if possible. See if it works then.

[bkenobi]

Hmmm. I checked my main system and it said that the 3.3v was 0.0v. I'm not sure I really believe speedfan afterall for voltages. I'll swap the PSU and see what happens.

Btw, I ran GameEx overnight with MAME as the only screen saver and it hung. I wonder if the PSU could be providing the wrong voltages and causing the hard drive to not respond correctly. That would be bad since I don't have a backup of most of my settings. I think it's about time I make one though!

[bkenobi]

I ran another program to check the voltages and temperatures and found that they were not even close to what speedfan was reporting. The CPU temps speedfan spit out were around 50C, but with this other program, they were anyhere between 35C and 90C. I don't know which is right, but I'm convinced that speedfan is wrong when it comes to voltages. I'm not sure that my motherboard reports -5v, so speedfan is just spitting out crap.

I tried a couple things last night on the software side and for whatever reason, the cab was able to run all night without problems. I don't know if it's crashed yet, but when I left for work this morning it was happily playing some vertical shooter (that's at least 8 hours without a problem). Odd...

[LB11]

The problem with speedfan or most monitoring tools is not actually the tools, problem. But the motherboards problem. There is so many different monitoring chips, and all motherboards might have sensors mapped to different components.

That is my experience anyways.

What program did you use that shows a different value?

The best thing to do is check the idle temp in the bios And compare that as soon as you boot into windows and see what is most accurate. You have an athlonXP or athlon64? The athlon64 has an internal diode. not all motherboards can read it. But coretemp program would give you the best reading if its available.

[bkenobi]

I used Everest as the second program. It was a trial version, so I wasn't able to get all readings since that's a limitation of the trial.

I just launched MBM5 for the first time and it has a LOT of stuff there. It does show that -5 isn't monitored at all, so I guess that's why speedfan reported the wrong value. It looks like the CPU (AthlonXP 3000+) is running at ~118F and the case (actually the MB ambient) is running at ~104F. I guess I'm not overly worried about the temps if that's correct (which seems about right).

I wrote an addition to my startup script that starts speedfan and appends the date and time of the crashed session to the GameEx log file. The cab crashed just after I left this morning, but I have a couple log files that show some interesting results. One thing I found was that for whatever reason, GameEx exits the screensaver randomly when there is noone using the system. Between midnight and 6am this morning, the screensaver kicked out and started a few times...while I was sleeping

I'm going to try the power supply, ram, and video cards next since I have parts that I can swap from my other cab for testing.

[hansolo77]

thing I found was that for whatever reason, GameEx exits the screensaver randomly when there is noone using the system. Between midnight and 6am this morning, the screensaver kicked out and started a few times...while I was sleeping

Just to be stupid, because that'sthe way I am... do you have cats? Or maybe (heaven forbid) a mouse/rat problem? Cats like to climb all over stuff. We still have our cats jumping up on our computer desk and walking on the keyboard because they want attention. If you have cats.. they could be jumping up on the control panel, and triggering GameEx to exit the screensaver, then when they walk away it starts up again.

Of course, if you don't have cats, or mice/rats... maybe it's ghosts?! LOL

Actually, funning aside.. If you think the problem might be the PSU, does it really hurt to swap it out with another one? Just temporary. I did that with my computer. I swapped out my PSU with my dad's, and voila! Everything started working again with no problems. So I gave it back to him and just bought a new one. Haven't had trouble since. You might also look into buying one of those PSU testors. I've read some good reports on some of them.

And finally, when was the last time you took the parts out of your cab and gave them a fresh spring cleaning? A can of compressed air to blow out the dust and cobwebs could save you too. My auntis computer was having the same problems as you, and we couldn't figure out what it was. Then I head a story at work about a guy who spring cleaned his computer, and LITERALLY hosed itdown. It got me thinking, and sure enough.. Her computer was massively overheating. All of the air vents were clogged, 2 fans had burned out and weren't running, and her case was almost too hot to touch.. Replaced the fans, and blasted out the dust.. her computer has run fine ever since, and doesn't lockup/crash/reboot/etc anymore.

[LB11]

I wouldnt bother with a power supply tester (besides, you can get a cheap power supply for the same price for testing). They really only test if the voltage is right. Alot of time if a power supply is going bad it will still supply the voltage under that circumstance, just not the amps it used to.