Sign in with
Sign up | Sign in
Your question

GeForce cards + VIA Chipsets ... *Very Bad News*

Last response: in Motherboards
Share
March 22, 2001 4:06:02 PM

Like many other people, I assembled a computer with an AMD platform, VIA chipset, and the nVidia GeForce2 GTS ... including a 300 watt power supply.

Ever since, I have run into more problems using this system than any other computer I have ever owned. I've posted here previously about the problems I've had with this machine ... and as you will read, it's been an on-going battle.

System Specs:
Win98 SE
MSI K7T Pro - VIA KT133 chipset
1000MHz Athlon
256MB Crucial Cas2 PC133
Adaptec 29106N PCI SCSI Controller
Adaptec AIC-7850 PCI SCSI Controller (for a SCSI-2 scanner)
18.3GB IBM SCSI hard drive
SoundBlaster Live X Gamer
RealMagic Hollywood Plus MPEG DVD Decoder
NIC Card
Antec 300 watt power supply
(5) 80mm fans
GlobalWin heatsink and fan combo
And last, but not least ...
nVidia GeForce2 GTS w/64MB-DDR

The system ran fine for a few months. I overclocked the video card when playing games, even raising the memory speed to 393Mhz. I was pleased with the performance of the computer, and I got really good benchmarks with 3D Mark 2000.

Then ... out-of-the-blue, while playing MechWarrior 4, the computer locked up, the monitor screen went blank, and the computer refused to boot. The only way to get the computer to restart was to unplug the power supply from my UPS ... and wait for a while, sometimes several hours.

I did not realize what was causing the problem, at first. Like other people, I did everything fundamentally possible to ensure that this was not a software issue, from installing Windows on a freshly formatted drive, to the latest drivers for the chipset and video card. I used probes to test for temperatures, just in case this was heat-related. But ... nothing fixed the problem.

(Note: After the first incident, I never overclocked the video card again. You may have noticed I had several fans in the machine. I took 3 of them out. I also had an NIC card installed at the time; but I removed that also. I am now running an USB ADSL modem, so the NIC card wasn't necessary anymore.)

At this point, I had to conclude that this might be a power issue. The GeForce Faq page had information posted that stated a GeForce card needed to have at least 20 amps available for the 3.3v line in order to run correctly. I saw that the Antec 400W specs said it could provide 28 amps for the line ... so I installed it.

(Note: I kept installing Antec power supplies because these were on the AMD-approved list, and a site that had done some testing found these to be the most stable. But, hey ... don't believe everything you read.)

Everything ran fine for two weeks.

Then, while running a low detail multiplayer game online, the system crashed, and again, would not boot up for nearly two hours. I assumed that this was because the power supply needed to cool.

I fought the system for the next few hours, and watched the video card die. All sorts of artifacts on the screen. Eventually, all I could reach was Safe Mode, and even then, I could barely see the display. I observed that in the BIOS, everything was spelled incorrectly ... and this included the hardware being listed in DOS when the computer goes through the POST.

After installing a new video card, I found that my hard drive's electronics had sustained damage, the sound card was damaged, and one of the memory modules was fried.

I was more than convinced that this was a power issue after I saw all the damage!

Antec power supplies might not be the best ... but a power supply, previously, and successfully tested under load, and 400 watts ... that should have been sufficient to drive this system. However ...

When playing games, or running other 3D applications, the AGP port demands more current than when the system is idle. The 3.3v and the 5v lines "share". This means, when the demand on the 3.3v line is high, less power is available for the 5v line, and vice versa.

I also discovered that a regular AGP port (not the AGP pro) can only deliver 25 amps to the video card.

Basically, this means that whenever I played a game, the rest of my system became underpowered, and believe me, that can be just a damaging to system components as too much current. It just may take the parts longer to die.

My conclusion was that this particular motherboard, with the VIA chipset, does a horrible job of regulating current, especially to the AGP port. I don't think that installing ANY ATX power supply, of ANY size, would make a difference.

Over the last few weeks, I have found literally hundreds of people yelling about this problem. The symptoms are nearly always the same ... lockups when playing games, sometimes only a few minutes into the game ... difficulty attempting to reboot after the crash, sometimes artifacts on the screen, spontaneous reboots with Windows first loads and/or Windows won't completely load the desktop, system tray, and icons.

I found an entire section of a forum devoted to this issue ... including information on why this is happening:

<A HREF="http://www.insanehardware.com/articles.php?i=00003" target="_new">http://www.insanehardware.com/articles.php?i=00003&lt;/A>

The first thing you'll see, is that this is happening primarily to people with the KT133 chipset, regardless of the mobo manufacturer. Also, people who are using a GeForce card like the MX seem to experience less problems, as these types of cards require slightly less current, and may even be a more efficient design. But people using these cards may experience problems, nevertheless.

The solution? I really hate to say this ... because I have been an AMD fan, right from the beginning ... but the solution, at least for the moment, is an Intel chipset and mobo, like the revamped 440BX, or the 815e. Unless you wish to use RDRAM, and that's your business ... and your money.

I haven't heard enough about the KT133A to know if this problem has been addressed. I also don't know if the newer GeForce cards, like the Ultra, or the GeForce3 handle power the same way. But I would be surprised if there have been any major changes to the nVidia reference design, and the way the cards handle current ... the company didn't bother to correct the problem with the earlier cards ... so why do so now?

I'm hoping that a couple of things will happen in the future. One ... AGP Pro cards will become affordable. AGP PRO 50 and AGP110 (or AGP Universal) can, respectively, provide 50 and 110 amps to the video card and port, using not only the 3.3v line, but the 12v line as well. Second, nVidia and VIA will fix this problem. Or I'll never come anywhere near a mobo with VIA chipset, ever again. Third, power supply manufacturers will recognize the need for affordable, higher power, cooler ATX units that will fit on existing motherboards and inside ATX cases. How many people do you think will want to cut up their cans, and install redundant power supplies? Heck, how many people would even know how to do that? We are PC users, not electricians.

Right now, I have a replacement KT133 motherboard I obtained from MSI, because the parts were still under warranty. I am using a Creative Labs Annihilator2 with 32-DDR, at least until my replacement GTS card arrives. For power, I installed a 400W Leadman PowMax, which seems to be a better constructed unit than anything from Antec. It was no problem to install new memory, but I'm also waiting for the new hard drive to arrive.

The computer is cranky, but running. I CAN play games ... I tested the system with the new 3D Mark 2001. But that's as far as I went ... I don't want to damage the new card or the mobo. Rebooting is still a problem ... I might have to restart the computer two or three times to finally load Windows. But once all the new parts arrive, I am going to put everything in a new can, install an older video card, and set the system aside as an emergency backup.

I'm going to put an Asus Solano2 815e motherboard in my old can, install a 600W ATX power supply from PC Power and Cooling, load up Win2K ... and I expect that all my problems will disappear. I've built systems similar to this for other people, and they haven't had any complaints, whatsoever. This is what I get for trying to get more bang for my buck, and trying to support healthy competition between chip manufacturers ... hit right in the wallet, and very disappointed.

That's the real deal, whether you are using an Abit, Asus, Gigabyte, MSI, or any other mobo with this chipset. You've been burned. If you want your GeForce to run the way it should ... go to the Intel platform, get a big power supply, and never look back. It'll be cheaper to buy a new proc and motherboard than to gut your entire machine and start over from scratch when the video card dies, weakens the power supply, and the rest of your system bites the dust.

Comments? Additional information?

Yours truly,

Toejam31
Anonymous
a b V Motherboard
March 22, 2001 6:13:59 PM

You're the perfect example of *why* I never overclock.. with Windows I have enough with figuring out how to not crash the software, I don't want to deal with the possible short- or longterm effects of overclocking.. In particular massively destructing things like this. Then again I'm just your average UT junkie and don't think the difference between 48 or 50 fps matter nowhere as much as my skills :) 

Kjella
March 22, 2001 7:11:29 PM

<b>Toejam31 :</b>

Hi there man. I looked over at your rig description, and the very first thing that caught my eye was that you had all (except one maybe, I don't remember how many slots KT7 Pro has, and don't feel like looking it up) your PCI slots occupied, were using high power AGP card and had a case full of fans. Then I read that the system ran just fine for several months. And went down when you overclocked the already oh-[-peep-] power hungry GeForce 2 GTS 64MB.

My personal opinion is that you simply damaged your power-related circuitry on the mobo and PSU. What damaged what : PSU the mobo power regulators or vice versa? I'm not sure (I'd be a liar if I said I am). Prolly fried the VRs on the mobo. They became unstable and refused to operate while hot. Somewhere on the way, because the VRs where out the window, the power surges killed your components (switched power supplies are a very delicate breed, never know what fit they could throw with fried VRs attached).

You went out and got a 400W PSU, with 28A rating on 3.3v (after seing that your AGP required 20A). You prolly forgot that you had a couple more devices in the box, except the GeForce. And in another good move you removed the cooling from the case. So now your fried mobo had to constantly operate at above pleasant temp. The system kept falling down, because although the PSU was more powerful, it now tried running more current through already damaged VRs on the mobo. And you tried doing all of that while having your rig full of power hungry cards. Finally, something on the mobo gave up.

Problems you were having later may or may not have something to do with you re-using some of the previously overloaded components, and may just be the GeForce power problems Instane mention.

Blaming VIA is nice and fun. A scapegoat. In reality, I'm not sure identical configuration would work on any chipset, given identical power-related components used. I think you should have a peek at power related specs of server PSUs and mobos. You may find some interesting things among those. For instance that they use much higher power-rated electronics.

Don't get me wrong, I do sympathize you, dude. All those components SHOULD have operated together. Theoretically. Reality is a much stronger thing though. A slight deviations from standard by each of those manufacturers, and puf! - your system died an agonizing slow death. My condolences.


P.S. For the record, I do have some electronics education behind me, and over 3 years of D-Level maintenance (as in "fixing") of hi-precision electronics devices for the Air Force.

-------------------------------------------
Theory is when you know everything, but nothing works. Practice is when everything works, but no one knows why. Around here, we combine theory and practice... Nothing works and no one knows why!"
Related resources
March 22, 2001 10:30:54 PM

In response:

I'll attempt to clarify the issue.

Yes, as I said, the system ran fine for several months. During that time, I didn't really run any high-powered 3D applications. Running games like this seems to exacerbate the existing problem ... something I might not have noticed for an even longer period of time if I had kept to running strictly 2D apps. At this point, I have been fighting this problem for around five months ... and the video card was only overclocked while playing games ... something I really only did once every couple of weeks. The video card was overclocked, perhaps 5 to 6 times, over a period of a month. And I was very careful when doing so, checking for artifacts, snow, etc. I had no intention of damaging the hardware ... I paid too much for the components to be frivolous.

These past few months have given me plenty of time to ascertain the source of the problem. This includes taking to the computer to a technician with even more experience than I have, and asking him to do some testing. He came up with the same conclusions I did.

This was not a heat-related issue. The average temp for this motherboard is 70F ... before and after all the problems began. I removed extra fans, and my NIC card, hoping to lessen the drain on the power supply. But I left enough fans for sufficient cooling.

I never forgot what components I had installed. I built the rig ... I know what's in there, and how much wattage I needed. I've run Intel platforms with 300W power supplies and 4 SCSI drives, and still never ran into a problem like this.

This isn't about choosing a scapegoat, or in attempting to assign blame. The reason I posted was not so much as to ask for help, but to try and get this information out to the forum, so that other people, dealing with similar issues, could finally have a clue as to the source of the problem.

If you spend some time, browsing the various forum on the Web, you'll see that many people are having similar problems with GeForce cards, VIA chipsets, and power supplies. This includes people who have never overclocked a component, and wouldn't if they knew how.

I appreciate your comments ... but understand, I also have some electronics experience, and have been a certified repair technician for several years. I wouldn't have posted this information, if I thought the problem was due to my own faulty judgement. If that was the case ... I would have just taken my lumps.

This is more than that ... I believe that there is a problem with the power-regulating circuitry on many VIA motherboards. Every test I have run points in that direction. I have also seen far too many people complaining about exactly the same problem, using components that are similar to my home system. If I hadn't found this information, I would've had to assume that I was at fault.

I posted the site link, because this was the first time I had found information that directly explained what my instincts had been telling me all along ... the problem is primarily found on KT133 motherboards. It's a design flaw, or a manufacturing defect.

I've looked at the rated specs for server power supplies ... but the fact is, for home PC's, and using mid-tower and full-tower cases, it's not feasible to expect people to alter their cases so these larger power supplies can be installed. Most of these cases need an ATX/PSII power supply ... or you are starting to talk about rack-mounted units.

I also want to reiterate ... I am now using a new, replacement motherboard, with a new video card and power supply. I currently do not have any components connected to the motherboard that were installed previously, except for the SCSI controller and hard drive. I am still having problems; the system locks up when booting into Windows, the mouse cursor occasionally sticks to the screen, and the keyboard stops functioning. But ... I am assuming (for what it is worth) that this is because of the damage to the hard drive's electronics. Nothing is overclocked.

Toejam31
a b V Motherboard
March 22, 2001 10:41:09 PM

I have a CUSL2 running 5 drives, 2 case fans, a bay cooler, a GTS, and 5PCI cards (inclucing a 2940UW SCSI card), running fine on a quality 250 watt power supply.

Suicide is painless...........
Anonymous
a b V Motherboard
March 22, 2001 11:09:43 PM

Way to go! Press that lemon to the last drop!

"Byte, Kilobyte, Megabyte, Gigabyte, Terrabyte, Pentabyte, Exabyte, Zettabyte, Yottabyte..."
a b V Motherboard
March 22, 2001 11:15:02 PM

BTW, I switched to an Intel chipset solution because I had the same kind of problems as you. My hard drive kept becomming corrupted. Eventually my video card quit. And that was only a TNT2 Ultra. So I gave up on the system after about 100hrs of servicing it over a 3 month period, put an old TNT card in it, stripped it down to basics, and got rid of it.

Suicide is painless...........
March 22, 2001 11:34:51 PM

Toe, are you sure it's the VIA chipset and not the power circuitry on the mobo that's at fault. The link you provided seems to point at the latter, not the VIA chip sets per se. I'm very concerned because I just put together a system with pretty much ths same things, except I do have the KT133A (no reason to think it's anything other than an officially-approved 133 version of the KT133, tho), and I'm "only" using an MX board.
March 23, 2001 7:12:31 AM

Read my tag line:


<i><font color=purple>Running within specs is the key to a stable computer!</font color=purple></i>
Anonymous
a b V Motherboard
March 25, 2001 3:05:22 PM

Hmm
Your computer was just crammed with powerguzzlers. Not unlikely that the powercirquits on the MOBO could handle this for only so long. Do not think VIA has got much to do with it.

I do agrre however that there are a few more problems with VIA-boards.
March 26, 2001 4:02:30 AM

Here, here. I too have no intention of EVER overclocking. I can only pray that will be enough....

sz
!