Via's VT82C686 A and B south bridges

[Guest]

Hi Everyone,

Tom's article on the KT-133 chip set indicated a problem with Via's VT82C686(B) South Bridge. This problem effects it's North Bridge which controls 3D and Open GL performance of the system. Tom then wrote another article on the new KT-133A chip set which uses the same south bridge, but didnt mention if this problem was corrected. Does anyone have any info on this, if it was corrected or if it still exists??

Thanks

 

[Guest]

I asked the same question today (3/13). Hope I get more response than you did.

 

[HellDiver]

"Tom's article on the KT-133 chip set indicated a problem with Via's VT82C686(B) South Bridge."

People will probably will beat me up, but I'll say it anyways...

Neither Tom's, nor AnandTech, nor other hardware-oriented websites on the net AREN'T the ultimative authority on the net. I do believe a lot of folks running such sites and posting reviews are professionals, but :
1) Not all of them, by far not all of them. Mind it!
2) Even professionals go wrong. Just look at all the bitching about P4 when it just came out : all hardware reviewing sites claimed P4 was worse than 486, it couldn't add 2+2 in less than an hour, etc. After all, it turns out that P4 actually rips Athlon a new one in Quake-3, and it's not that P4 is that bad, it's just that you need to optimise your S/W a bit to make it run more than on par with Athlon, and in a year or two, it'll turn out, that P4 is actually a much faster core. Optimized FlasKMPEG was just one proof to it.
3) None of those sites has neither time nor will to perform thorough testing of the hardware. All they care about is being the FIRST to publish the scoop. (Applies to great majority, but NOT all sites). 3 runs of Quake3 Timedemo, Sandra, UT if the site wants to brag, and off the review goes. Putting a rig for 24/7 load of tests for a couple of days? DUH! Tomorrow there'll be a new card to test, no time for sissyness!

Tom's results with those 686A/B southbridges may in fact be quite wrong, and may very well be a result of lame BIOS version or some other issue. See this article on RealWorldTech : "http://www.realworldtech.com/page.cfm?ArticleID=RWT012301000000". It includes test results SHOWING NO PROBLEM WHATSOEVER with 686B.

NOTE : some exaggerations may be present in this post.

 

[Guest]

Helldiver,

Thanks for the info!! Im not sure there isnt a problem with the KT-133 chipset. The best example is a recent game I downloaded, Mech warrior 4, that has known problems with this chipset, inwhich, there isnt a solution. These problems cause your system to lock up, even if you have to latest bios version,a7v1007, and Via drivers 4in1(28a), which I do. No matter what their instructions said to do, it didnt help.

I then went to my PIII-600eb, also using an ASUS mbo and VIA chipset, which ran the game without a problem. I then took the same game to work and ran it on two other pentium systems without VIA chipsets, etc, all ran flawlessly.

So Why then does this game not work on the A7V with the VT82C686(B) southbridge but works on every other VIA and Non-VIA chipset?? The problem Tom mentioned with the North and Southbridges conflicting would cause this exact problem.

What do you think??

-Insert clever phrase here-

 

[HellDiver]

"I'm not sure there isnt a problem with the KT-133 chipset."
Did you read me stating that there are absolutely no problems with KT133 chipset? 'Cause I don't remember typing that! Yes, VIA chipsets occasionally exhibit less stability, or should I say they're less tolerant to various programmers abuses and misuses of hardware. There are also software vendors who don't endorse AMD CPU products. There are also whites who don't respect colored folks. Would you like to change the world? BTW, 686B is used only by KT133A, see below:

"game not work on the A7V with the VT82C686(B) southbridge works on every other VIA and Non-VIA chipset??"
For starters, A7V runs on 686A, NOT B. A7V133 sports a 686B southbridge. Then, I would avoid statements like "every other". Did you personally run it in "every other" configuration? If you have specific links to specific locations describing the specific problem MechWarrior4 exhibits on KT133A, I'd like to read through those.

The only relevant inputs to the 686A/B discussion would of course be about mobos that differ ONLY in their southbridge (several companies have mobo models that differ only in southbridge and some other irrelevant to the topic cosmetic changes).

In the mean time, I saw no further evidence supporting Tom's findings, so far.

 

[Guest]

Tom didn't find an overt malfunction, as you seem to have. Only slightly slower performance (43.4 fps vs 46.4 fps w Light-04, whatever that is).

http://www6.tomshardware.com/mainboard/01q1/010117/kt133-17.html
He goes on to say:

"This correlation is evident particularly in the case of MSI. The manufacturer offers two versions of the MSI K7T Pro2. The classic K7T Pro2 has the ATA/66-Southbridge (VT82C686A), while the K7T Pro2-A is based on the ATA/100-Southbridge (VT82C686B). Apart from this difference, the boards are constructed identically. The result shown above reveals that the VT82C686B appears to slow down the Northbridge."

I wonder if it does the same thing with AMD's north bridge. He does not cover that point.

I think you may be right about the urge to publish rather than the urge to run things down to a 'final' resolution. Tom never came back with definitive MPG Flask results - just a piece at a time of new test cases w their results as incidentals in at least 2 later articles.

I have emailed Tom and Via for an update. Neither has replied to me. Imagine that!

 

[HellDiver]

diehard :
"Neither has replied to me. Imagine that!"
Yes, Diehard, I can imagine that very easily - Tom never answered any of my email queries either. And I have sent more than one.

If you ask me, I think in the last year Tom's Hardware Guide made too many harsh statements about various products/technologies, yet failed to back them up with any hard fact follow-ups. Apparently the place is turning into another web-based tabloid : run the scoop on the front page as fast as you can, realize you blew it big time, never go back to the issue, because no one will remember it anyway - tomorrow you will run a new scoop. And the few serious technical articles that occasionally show up are just a leftover from times when the site wasn't so commercialized.

Bummer... :-(