cleaned out dust, replaced PSU and still random restart and freeze, help please?

wow, I miss being able to stay in secondlife long enough to do anything. Anyone around experienced enough with computer hardware with any other suggestions I might try to get my computer to behave well enough to stay inworld longer than ten minutes? (or maybe dare I say... resolve the problem?)

It's randomly restarting with no error warning, just blinks off and starts rebooting. It's gradually gotten worse. This morning, it froze up. What little I could, I looked around different computer sites and saw suggestions for cleaning out the dust, and replacing the power supply. The machine was purchased in 2004, so I figured it wasn't a bad idea to replace the PSU.

I ran scans to rule out malware and virus when the random restarts first started happening last week. All clear on several. I did both online and onboard scans. Did a defrag, cleared out all outdated programs. I installed Speed Fan 4.39 and noticed the core and temp 2 both going up to an extreme of 55C just before a freeze up or restart.

2. I unplugged the machine, took off the cover and carefully used the compressed air to clean out any dust, there wasn't much at all. (I held the fans so they didn't spin as I cleaned). With the side cover off, we turned the machine on and noticed all the fans working as they should.

(as per Dana's suggestion: pulled both memory sticks, cleaned them and the slots and reseated them, trying now but both HWMonitor and SpeedFan are showing the heat going up again. Will try seperate shortly).

3. Replaced 250 power supply with 400, no help.

Any other suggestions what might help this machine so I can get back inworld and stay put for a while? thank you for any help!

(Ceka suggested I put this info in here, ty Ceka for showing me where to find it, and it's AMD Athlon)

CPU: AMD (Unknown model) (2194 MHz)
Memory: 2048 MB
OS Version: Microsoft Windows XP Service Pack 3 (Build 2600)
Graphics Card Vendor: NVIDIA Corporation
Graphics Card: GeForce FX 5200/AGP/SSE2/3DNOW!
Windows Graphics Driver Version: 6.14.0010.5316
OpenGL Version: 1.4.1

Sounds very much like one (possibly both, but unlikely) of the memory modules has developed an intermittant problem. Yes they do that.. now & then you'll get a stick where one chip on it goes tits up but the rest of it is fine. The problems happen when that certain block of memory address is used by windows, which may not be all the time or only under certain circumstances depending on how much memory space is available.

1) Reseat your memory modules (as in take them out, blast em off with air, and put them back in all the way, firmly in, and wiggle them sideways in the slot...carefully! to get the contacts to mate up)

2) try removing one of your memory sticks and run with just one to test it. If that still fails try running with only the other one.

how much memory do you have..also what type of video card and how many proccesses are running..

if you go into secondlife and click the help menu then go down to about second life..click that and copy your system specs from that window..then post them in your original opening post

It's sure worth a try, thank you Dana!

Ceka, I posted the info in the OP and will be back here in forums shortly, thank you!

Your graphics card is out of date and below the minimum for SL according to the System Requirements page. I can get by with my nVidia 7300, which is two years younger than yours, but it's sometimes a struggle, as the Windlight stuff puts a heavy load on the graphics card.

If a new graphics card is not an option for you, I would recommend using a pre-Windlight viewer such as Nicholaz Bleeding Edge. Here's a link:
http://nicholaz-beresford.blogspot.com/2008/05/version-overview.html

55 degrees Celsius is not really all that hot. In fact, most chips are able to sustain temperatures above 100 C before being considered "overheating". If you are experiencing random crashes, lockups, and reboots at around 55 C, then it is very likely a chip is already damaged. Pinpointing the problem won't be easy, however, as it could be a faulty CPU, a faulty graphics card, or even a faulty memory card.

I'm not running Windows, so I can't give you a perfect step-by-step for doing this; but you need to go in to the system settings and verify that it is not set to automatically reboot on a STOP error (BSoD). If your computer is actually throwing BSoD's rather than just spontaneously rebooting, then the STOP codes could be valuable in determining where the problem is.

Ya i'm gonna say it's the Graphics card as well..i had the Nividia Geforce5200fx ultra and when windlight was first coming out all i really could do was run it for a little while to take pictures..i mean it ran windlight but i can see from all the updates we have had since that time that the card would shut down now days..it's i'm gonna say way below what you need nowdays..

The latest nVidia driver for that card is 175.19. Not sure what driver you are using with that reported version........ 6.14.0010.5316. Using nVidia's common driver numbering that would be 053.16? Anyway, it might a good thing to update your driver to the latest for that card:

http://www.nvidia.com/object/winxp_175.19_whql.html

I also noticed your OpenGL version is 1.4. That's pretty old......I don't think SL works with that. You really need 2.0 or above and I'm not sure that card will support that even with the newer driver. AGP is all but obsolete now so you may have a hard time finding a replacement card for it......let alone a modern card with OpenGL 2.0 support. You probably don't have a PCI-e slot on that machine but I'm sure you have PCI...........so I would look for something newer for that interface. Perhaps something in the 7000 series like a 7600. Newegg.com and TirgerDirect would be a couple good places to start.

From what you seem to be experiencing it seems like a graphics issue. With 55 celius temps it's NOT a heat issue......that is well below any threshold for shutting down (that temp is usually around 100 C). It looks like your card is either dying or it just does not have the muscle to handle SL with all the graphics processing it requires from a computer.

You can do the memory stick swap stuff......but I don't think it's going to help. If you do, before you put the stick back in get a regular rubber eraser and scrub the contacts till they are shiney......dust off the eraser dust thoroughly before sticking them back in. You might as well do the same for the graphics cards itself too.

For what it's worth.........my opinion is your card is due to be replaced. Sorry for the bad news. Good luck.

Yeah, it sure sounds to me like the graphics card. Although that temp isn't very high at all, the fact that it restarts when the GPU temp rises suggests the card (or something that's warming up, anyway) is in a pretty tenuous state. I wonder if it could have gotten much too hot at some earlier point; I get it that the 55C was measured before this recent cleaning, but maybe something caused an earlier overheating that did some damage. Wouldn't necessarily have been heat doing that damage, of course, but it's a common problem.

It's an AGP card... so now you have the difficult choice whether to replace it with another obsolete card, or replace the whole machine. Sorry 'bout that. Would be nice to get a loaner card to make sure that's the problem before spending too much.

You might try downloading Orthos and Furmark to try stress-testing different components. Furmark is an OpenGL test, and Orthos tests CPU and RAM.

I would start by running Orhtos and selecting the option in the Test drop-down menu to stress RAM, and let that run 20 minutes. If that passes, open one instance of Orthos for each core of your CPU (two for a dual core) and set the test to stress CPU, and use the drop box for cores to set each instance to a different core. Run that for 20 minutes. If Orthos does not crash your system, then run Furmark for at least 20 minutes.

If it is your graphics card, the best AGP card I could find on Newegg is a 7600GS. This card meets the minimum requirements for SL at $80, but I would not expect high details or amazing effects.

Furmark: http://www.ozone3d.net/benchmarks/fur/
Orthos: http://files.extremeoverclocking.com/file.php?f=200
Newegg: http://www.newegg.com/Product/Product.aspx?Item=N82E16814143069

Thank you all for the suggestions, I'll sure try them.

It was working just fine until the last update.

If the last update is when it started, I would be sure to try Malia Writer's suggestion and install a pre-windlight browser. It could just be the new browser trying to do things that your graphics simply can't handle.

SL does work with OpenGL 1.4. That's the highest Intel GMA chipsets support and these do work. With that old machine however you may want to try a pre-Windlight viewer. Not many left after the adult and security changes that work properly. See my sig or Henri's 1.19.

If you still want to stick to that machine, I think there are Nvidia 9600 cards with AGP slot still available. But thats definitely the end of the line.

Thing is, it isn't just secondlife. In fact, I first noticed the problem at a game I play on another site. At that time, I figured it was a borked firefox update. (or maybe I'd missed a flashplayer or java update, so naturally I redid those). Then it happened with Pogo too. Secondlife was basically the latest I'd noticed this happening.

Only this morning did I realize it was right after the last update.

I'm thinking new machine, which might take a while.

Perhaps you should just give a fresh install a try. A new machine is better but only if you can afford it, so make the best out what you have.

There are a number of tools around to check system health. Someone mentioned the memory which may have some corrosion over time resulting in bad contact, or a chip may have died. Look for Memtest86. That's on most Linux Life CD's like Ubuntu. Also there is a compilation of a system maintenance CD called "Hiren's Boot Disk" that has ton's of older system tools. A lot of other free tools you can find with some search. Google is your friend .

If you don't find a HW problem, it may well just be a too old installation that got messed up over time or some malware like a virus, spyware etc. A fresh install would get rid of that. Why not get a new harddisk, without overwriting the existing one, swap them so you would not lose anything. And reinstall your machine.

Yes, if it's a 2004 computer, it's past time for a replacement. Ask Santa Claus for a shiny new computer for Christmas!

Start > Run Eventvwr

System error records.

Look for the errors, make a note of the codes, and Google them.

Also, right click My Computer, Properties, Advanced, Startup and recovery, take the tick out of automatically restart on failure. This way you'll get a blue screen of death when things go wrong, you can make a note of the info on the screen and Google it after a reboot.

I disagree it's most likely the be the AGP card - Just because a card is old doesn't mean it's somehow gone from being slow but working, to crashing the whole OS because SL hammers it. It could be PSU (but most likely discounted as the old one's been swapped out), memory, motherboard (I had a socket A board that would do this from time to time, got worse over a few months to the point the system was unusable - Turned out a couple of capacitors had been soldered on the wrong way round at the factory) or possibly AGP card. But I'd be thinking more motherboard or memory myself.

A visual inspection of the motherboard helps, if any of the capacitors are going bad they often have very domed tops (instead of flattish) and dried brown crud encrusted on them. I've had quite a few board from around 2000 - 2005 that have been flakey through this fault.

Downloading something like Ultimate Boot CD (UBCD) gives you various tools to test the hardware, set it running memtest overnight to see how the memory is coping.

Very good advice. There's no way it can be what version of SL or GL it is because the problems happen all over the place, on stuff not even related to SL or GL, and not in use at the time. I also agree it shouldn't be the video card.. 9 times out of 10 with a bad video card you'll either get no video at all, or a BIOS beep error code warning and it won't let you boot at all. Running Memtest, which is pretty much an industry standard, is a very smart idea before assuming anything.

I'd be careful about the "lack" of specifics here when referring to chip temps. 55C is cold for a video card, yes.. but that 55C is not cold for an Athlon class CPU. Which is it? Without a CPUID string or model number to go by, the cpu almost has to be either a 2200Mhz model (3400+ - 3700+, regular and low power models), or an overclocked 2000Mhz model (3000+ or 3200+ only, regular and low power models). For the regular 89w models the max die temp is 70C. For the low power 67w models the max die temp is only 65C. Mid 50's to around 57 or 58 is on the hot side of normal range for all of them, but if it gets to like 62, 63 or so.. shut 'er down she's too hot. Also keep in mind that older Athlons relied on an external thermistor that touched the bottom of the CPU package, so not all that accurate. Err on the safe side.

it's 3400+

It's 3400+

I'm sorry, I had this problem too.
I tried everything.
And I had to get a new computer.

Looking at towers shortly... skimming the forums as to what I DONT want. Thank you all of you who post headache systems to avoid

dude ur cpu is not hot its ur gpu or graphix card perhaps. speed fan sholda tolda ya the temps on that too (unless 52oo dnt have sensors). it is what was giving me problems with fps but i wanst freezing. sl is high intensive graphix hog. does a geforce fx 5200 have a heat sink, a fan, u prob just need to invest in a gpu fan is all

oh i found a geforce fx 52oo lying around the house. they have an aluminum heatsink and no fan. YOU need either a new graphix card (a geforce 95oogt 1 gig is like 70 bux and comes with a copper heatsink and alil coolmaster fan- took one apart TODAY ) or get an ultra gpu fan for like 25 bux from tiger direct or azahman for like 40 same place.

if ur g card willplay sl with some aspect of windlight working onit at all ur gonna need cooling

It's been a long hard road, and I know this was probably temporary, but something I did made this all worse, but then the report error to Microsoft box kept coming up. When I'd send it, it sent me to the nVidia site. I swear, I clicked to update the drivers many many times since you posted this, and it would NOT let me accept the EULA. We were pricing machines, trying to get me into a custom lean mean dell tower. Meantime, I did some more on this, and whatever I did, I got that stupid send report to microsoft (seriously, that thing would go off several times in a row, stuck in a loop). Alas, once again it gave me a link to the nVidia site. I looked at that figuring it was gonna stick on the "I ACCEPT" and not progress. It did the first time, but as I had the other times, I tried one of the other links, and it worked. I've been playing my favorite games for over an hour now with no blue screen or restart at all.

So, at least for now until I save up enough for that new tower, it's working okay. (I hope it's not just a fluke tonight and back to the BS tomorrow? ) Thanks again everyone!