Computer help.

I've been working on this computer for years now, and recently I've been having more trouble then usual. I'm probably going to chuck this sytem soon for a system a friend might give me [this system is too old/slow for today's games anyway], but I'd like to give this one last try. It could still make a decent backup system. Maybe you guys can help give me some ideas.

First, I'm going to list the parts that I have, and can switch:

2 x [AMD Thunderbird 1100MHz K7 Socket A]

[MSI-6389] ; [MSI-6330 KT7Pro] ; [MSI-6340]

3 x [300w PSU]

3 x [256MB PC133 SDRAM]

[GeForce 2 MX 400/400 64MB] ; [GeForce 2 Ti 64MB DDR]

[Western Dig. 30GB HDD 7200RPM] ; [Western Dig. 15GB HDD ?RPM]

[all setups using Win2k, all updates, service pack 4, latest video drivers for both cards, DirectX9.0c, via 4in1 installed]

That should be enough. The problem started with this setup:

MSI-6389, 256MB SDRAM, GeForce 2 Ti.

It worked fine for a while, but last year the sound would mysteriously turn off, and the system would lag real bad afterward. I turned the sound off in the bios, and it worked fine after that. I put a cheap sound card in it, and got the same problems again. Took it out, and just didn't use the sound.

Then the graphics started to go. I started getting very fractured graphics, and weird glitches. Then, in November, the system just wouldn't power up.

So, this is when I bought the MSI-6340. I had to turn off the AGP 1WS Read/Write in the bios to get the video to work properly, but after that, everything worked great. Smooth games, normal lag, I was happy. I got cable internet a couple weeks later, and the sh*t hit the fan. I installed an old NIC I had from before I replaced the MB, and the system wouldn't power up.

I remove the NIC, and when I power up again, my CPU fan isn't working. So, I replace the CPU fan, and the NIC, and now my system is having very similar problems that I had with the MSI-6389. I also bought the KT7Pro at this time thinking the old NIC may have damaged the MB.

I've redone all the software, but it doesn't change anything. I've used multiple variations of the hardware above with no change. I've tested all three PSU's, and all are fine. I stripped the system down to bare necessities (CPU/FAN/MB, 1HDD, 1SDRAM, GF Ti, turned off on-board sound), to see if maybe the PSU wasn't powerful enough, but that doesn't change anything either.

Games I test this with:

KOTOR - Main screen/gameplay runs extremely slow. Frequently pauses, occasionally locks up the system.

Kill.Switch - Main menu graphics are so distorted the options don't show up. This was originally fixed with disabling the AGP 1WS Read/Write, but now that does nothing.

Homeworld/Cataclysm - Just runs real slow.

Even Flash/Shockwave games run slow.

I also tested all memory with memtest86. DxDiag shows no problems, and all tests run fine. No Spyware/Adware/Malware.

Any thoughts appreciated. I'll pretty much try anything at this point.

sounds like a bad or going bad motherboard.. perhaps the PS is going also.. it happens.

Have you tested out your CPU?

Also, what version of gfx driver did you use? I would try to get drivers that were released within a generation or two of your gfx card.
When you redid all your software, did you do a clean windows install?
Also, make sure that your PCI bus are running at 66mhz, since the PCI clock runs on a divider of the FSB, and depending on your mother board, the divider may not be working so well and you are runnin the PCI bus too fast.
I had that problem with my Asus Nforce1 board, and it was fixed when they released a BIOS that let you clamp down the PCI bus speed regardless of your current FSB. (it still had problems, but definitely helped)

Sacrifice a chicken on it. It worked in Major League.

Have you tested out your CPU?

I don't really have a way to, other than having two of them. They could both be defective for all I know. Any suggestions on how to test are greatly appreciated.

Also, what version of gfx driver did you use? I would try to get drivers that were released within a generation or two of your gfx card.

If I remember right, 71.84. I had a little trouble finding a version that supported my GF2 Ti [later versions will work with GF2's, but not the Ti], but this version worked great before I installed that NIC, so I'm assuming these drivers are fine.

When you redid all your software, did you do a clean windows install?

Yes. I have a separate partition where I keep backups of files I wish to keep, so the partition with my OS is typically easy for me to format. I keep the bigger updates in my backup partition to save on download time. Takes about 45minutes for me to go from Format to Completely Updated.

Also, make sure that your PCI bus are running at 66mhz, since the PCI clock runs on a divider of the FSB, and depending on your mother board, the divider may not be working so well and you are runnin the PCI bus too fast.

This is definately something I'll have to try when I get home. I hadn't thought about that. Thanks for this suggestion!

karmajay wrote:

Sacrifice a chicken on it. It worked in Major League. :)

Original or extra crispy?

Sounds like you need some known good parts to work with friend....

To test your CPU, Prime95 would be good. Try running it for 24 hours .
prime95 site
That site should have all the info you need to start stress testing your cpu.
more direct link to help on how to setup a torture test properly.

Thanks Lethial, I'll set that up tonight.

word of caution: you may not like what you may find tho.
Also don't be too worried about getting your system 100% prime stable. People have built many stable systems that would fail prime torture test in few hours.
I'd use it iteratively to see if you get any incremental stability improvements.

Oh and prime95 doesn't do anything to stress yourr GPU, so if you want to do that you will need things like 3dmark, and if you really want to burn your gpu
that is a program that will heat things up fast. Though it is mostly designed for more recent graphic cards, so I don't know if it will run for you. YMMV

I'm going to run the test to see if anything major errors pop up. I figure, if it's the CPU, it'll probably generate an error near immediately.

Not interested in testing the GPU. The fan gave out on it a while back (after I started having problems), but since it's only a GF2, I never bothered replacing it, as it doesn't overheat. The fan does run a little, but more-or-less is just a heat sink at this point. The other card I have (the GF2 Mx 400/400) is stripped from a working computer, so I'm not going to bother testing it either.

One more thing PureEvil be sure you have a temperture monitor like speedfan when you are torturing your system.

If you really want torture your cpu, or you don't want to wait for prime95. Here is a program that will heat up your cpu, your ram in no time.
S&M tester
Please excuse the name, it is a Russian program, so they may not know the connotation of the name. (I didn't know either, until I looked it up)
But just remember that you will DEFINITELY need a temperture monitoring program when running this.

The KT7pro can monitor temp, so if Prime95 doesn't show anything worthwhile, I'll check this out.

PurEvil wrote:

The KT7pro can monitor temp, so if Prime95 doesn't show anything worthwhile, I'll check this out.

Actually I would highly recommend speedfan for temperture and voltage monitoring. Also, S&M tester will stress your system much faster then prime. The complete test takes few hours on my system.

rthdribl is a DX9 showcase, HDR mostly....it either wont run with a GF2, or wont affect the system the way you want it to.

I would worry too much about temps either...at best you are going to get an ambient case temp as there is no on die sensor included on a T-Bird. That readout is a sum of all parts (including room temp), and really wont help you find the culprit. If you are worried about heat....run it with the side off and point a house fan at it.

What I would do is try replacing these parts with something you know (or can at least be fairly certain) works. This is where retail stores can be very useful. Go to a shop that sells this stuff, tell them what you think the issue is, and politely ask them if you will be able to return the item if it does not fix your problem. The answer should be yes, and if its not it will be at the place next door. Morally if it does fix the issue you probably ought keep the item, however thats your call.

pol wrote:

rthdribl is a DX9 showcase, HDR mostly....it either wont run with a GF2, or wont affect the system the way you want it to.

I would worry too much about temps either...at best you are going to get an ambient case temp as there is no on die sensor included on a T-Bird. That readout is a sum of all parts (including room temp), and really wont help you find the culprit. If you are worried about heat....run it with the side off and point a house fan at it.

What I would do is try replacing these parts with something you know (or can at least be fairly certain) works. This is where retail stores can be very useful. Go to a shop that sells this stuff, tell them what you think the issue is, and politely ask them if you will be able to return the item if it does not fix your problem. The answer should be yes, and if its not it will be at the place next door. Morally if it does fix the issue you probably ought keep the item, however thats your call.

Not worried about heat at all. Every time I've check the temp in the BIOS, after running it a while, never reads above 40C.

I'll try a couple of things in this thread, but I've already decided not to waste any more money on this system. Like I said in the first post, I've got a friend who is about to replace one of his machines. He's gonna give me his oldest, which would run better than mine even if it were completely fixed. If I were to spend the money to take it out the store, I really wouldn't feel right just using it to test a machine, then returning it. I don't really live up to my name, do I?

I've already decided not to waste any more money on this system

heh good call

If I were to spend the money to take it out the store, I really wouldn't feel right just using it to test a machine, then returning it. I don't really live up to my name, do I?

I consider this to be the final boon of retail. Return policies are the only reason I even bother going to brick and mortar shops anymore. A lot of things I buy I am more comfortable buying from somewhere that I can go back to, talk to the guy I bought it from, and have them deal with any problems that might come up. And dont think for a minute that the stores dont make money this way. I did fully commissioned electronics sales for about 2 years, and if something goes out the door, even with the relaxed return policies chances are very good it isnt coming back. You can also be sure that if that percentage starts to change, then the policies will change as well.

Anyway not everyone agrees with my outlook, but it is what it is.....and I for one dont mind stikcin it to tha man when the polices work out in my favor >)

I thought retail's return policies are getting much more strict now, because of all the returns and all. i guess that is not true?

But pureEvil, you cannot judge your temp by checking the BIOS after running it for a while. To give you a point of reference, my CPU can heat up to 60oC while running a lot of stress testers, but the second I stop them, the temperture drops back down to the 30's

I only mentioned rthdribl because the older versions aren't just about HDR, and it is a great tool to see graphic artifacts. I used to have one for DX8.

Oh and finally, PureEvil, you are not so evil after all I am very much the same way though so I am no better

Well, I downloaded speedfan last night so that I could get started with the other tests. Honestly, speedfan gave me my problem...

My first temp reading for the CPU was 144C! So unless the program isn't working right, I'm overheatin' pretty damn bad. The other two temps were 52C and 36C... I had assumed that the paste that came on my new CPU fan would be enough, but apparently not. I applied some of my own this morning, a silver composite paste, but I couldn't get the OS to boot up all the way (besides safe mode). I ran speedfan from safe mode, but the temp hadn't changed.

I just didn't have enough time to mess with it this morning, so I'll mess with it some more this afternoon. But this has definately got to be the problem. I'll probably switch the CPU's, just in case the heat damaged this one.

Hate to ask, but any tips on how to configure the fans in speedfan would be very helpful.

I double-checked the CPU fan, to make sure it could handle my CPU, but it looks like mine would be at the very low end of the chart. Shouldn't be overheating at all.

Spire SPA04S4-U

evil. I just double checked, your motherboard should be supported by speedfan, so you sholdn't have problem getting the correct values for your temperatures.

Go to the Info tab, and click on the URL to register at their web site, from after that you can get user created configurations matching that of your motherboard.

From MSI website, you have a KT133, and the SMbus that it uses is indeed supported so you should find a configuration that is right for you.

edit: just triple checked and I find a matching configuration for your MSI6330 board

Manufacturer : MSI
Motherboard : MS6330 (K7T Turbo2)
Language : English

There are 3 configurations, try them out.

OK, I'll get this fixed up tonight. It's actually brightened up my day a bit to finally know why this machine is so unstable/slow. I honestly just assumed the fan was working properly, and [as n00bish as this sounds] I never realized I could check the temp while in windows. Hopefully tweaking the fan speed will cool it down.

My first temp reading for the CPU was 144C

dont panic dude....thats not possible.

Speedfan does the same thing on my system, as does every other util I have used to test temps. You have to just kind of look at what the temp is, and make an educated guess as to what it corresponds to. Your proc will abosolutely not function upwards of about 70C....I am guessing if you look at that temp its rock solid no matter what you do. One way or another it just isnt right, as at 144C it would have melted through the mobo, burrowed through the case, set the carpet on fire, and generally come to rest about 3 inches below ground level

I thought retail's return policies are getting much more strict now

Yeah they are depending on where you go, but good ones will still have this kind of policy in place. Just make sure to speciffically ask, and to get the name of the person you talk to. I have done this with Best Buy lately, and they are fine with it....Circuit City is a grey area, as I have been told both yes and no....the local privately owned PC shop here is always happy to take returns if things dont work out. It can be hit or miss, but as long as you are honest with your intentions you cant really go wrong

That temp reading for the CPU is definitely a mis read. I just hope that you dont' have a bad sensor. From the look of it for your mobo, the sensor is in the center of the socket, which tends to report temps on the warm side.

However, your other two temps kinda worries me, I am assuming that the first one is for chasis, the second for your HDD? If so, 52oC seems kinda high, were you running the stress tests when you got these temps?

Pol: I am surprised that CC is more strict then BB. Did you talk to the Customer service Desk people or the techies at BB?

Well, I couldn't get speedfan to give me a correct reading on this MB, so I switched back to the MSI-6330. Now I've got 46C, 31C, and 21C. I had a little problems when I first got it working (temp 1 was about 60C), but I figured out that the CPU sink was just too tight. I loosened up the clamp... a lot... and now it's at a very acceptable temp. [I hope I'm correct in assuming I didn't crush my CPU, as there is no visible damage, and it boots up fine.]

Unfortunately, this doesn't solve my problem. Still getting major lag, and lock-ups. I'm working a normal day today, so I'll be able to stress test the CPU when I get home to see if I get any errors.

good luck and keep us posted please

Well, after all the crap I did to my system yesterday, I give up.

S&M Tester: All passed.
Prime95: Ran for about an hour. No errors.
FreshDiagnose: No errors, All benchmarks [CPU, GPU, HDD, Memory] completely normal.
DxDiag: No errors, Direct3D and DirectDraw tests fine.
WesternDigital Diagnostic: Both HDD's running completely normal.

The GPU tester you suggested wouldn't work. My card doesn't support Pixel Shaders v2. So it just produced an error and quit.

I wouldn't be mad about this system if I knew why it wouldn't work. It's so frustrating that I can play any game, and have near immediate problems with it, yet every test I throw at it comes back completely normal. I switched out all the cables, and switched out the PSU. Now it's fairly stable in single player games, and doesn't freeze very often. But I tried playing the free trial of EVE online, and I couldn't go 5 seconds without it locking up for 10-15 seconds at a time. I didn't get very far before I gave up. I thought maybe this was indicative that the NIC was another dud, but alas, the system still runs the same without it.

Oh well. This computer lasted me 3 years, and I built it from budget parts to begin with. Chances are, I was lucky to keep it running as long as I did. Meh.

sorry to hear that. Though did you actually run the S&M tester with the stress test setting? and usually people run prime95 for 24 hours or more

Anyway, just one last stupid thing, you did make sure that the NIC is not jammed in the riser card slot on your board right?

Could you run 3dmark2k3 (maybe 3dmark2k1?)

Sounds like somewhere the power is screwed up. I think you made a good choice just don't worry about the system, time to move on.

suggestion: take it out to the back yard and destroy it "office Space" style

suggestion: take it out to the back yard and destroy it "office Space" style

Oh man that would be awesome!

I'll try to get my hands on 3dmark2k1. I doubt it'll show anything, but hey, you never know

Anyway, just one last stupid thing, you did make sure that the NIC is not jammed in the riser card slot on your board right?

I completely understand why you asked, but man... that's just messed up. Can you even jam a PCI card in a riser slot? Looks like it wouldn't line up right.

Sounds like somewhere the power is screwed up. I think you made a good choice just don't worry about the system, time to move on.

I agree, I just don't understand why even when I tested the PSU's they all came out fine. The only thing I can think to do at this point would be to move the computer to another outlet in my apartment and see if I still have the same problems. I could also try switching the power cables out... at least I think I have a couple extra laying around =\

How did you test your PSU btw. I don't recall you mentioning that you measured the voltage and amperage using a multimeter

Oh that 3dmark2k1 thing is just to see if that thing will crap out just like your games does.

First post:

PurEvil wrote:

I've tested all three PSU's, and all are fine.

Robear lent me a little tester. All the lights came on, so I figured they were fine. He lent me another voltage tester which I should have used to make sure I'm getting full current, but I didn't get around to it.

I am not sure about that tester... Does it monitor the voltage variations and current fluctuations? If it does, try to use it when you are stressing your system, and see if it pass or fail.

To give you a reference 5% deviation on the +12V line could cause you a world of stability issues over time