Data-corruption bug hits VIA chipsets

8 replies [Last post]
johnpr98
Offline
Joined: Aug 20 1999

For Info, Theres a fix on the way
http://www.theregister.co.uk/content/3/18267.html

quote:
VIA has confirmed a data-damaging glitch in its 686B Southbridge chip - a major part of the Taiwanese company's KT-133A chipset - and is working with mobo makers to prepare BIOS updates to fix the problem.

The southbridge part is used in the vast majority of AMD Athlon-oriented mobos, primarily the KT-133, but it can be used with northbridge parts from the Apollo Pro 133, KX-133A and AMD-76x chipsets too. VIA said it is investigating the problem to see how many chipsets are affected.

The bug was uncovered by German hardware site Au-Ja! It's not exactly a common problem: the date corruption affects large, 100MB and up file transfers between two hard drives connected to separate IDE channels exchanging the data by DMA. Having a Creative Labs Soundblaster Live card in place seems to exacerbate the problem.

VIA's BIOS fix works by adjusting a number of PCI settings, which, according to TechChannel, suggests the problem is a result of competitive PCI access.

VIA told The Register that it is a BIOS issue, and it will be posting a fix on its Web site sometime next week. ®

John Price
http://www.johnpr98.com

[This message has been edited by johnpr98 (edited 13 April 2001).]

johnpr98
 
If you have any Forum Suggestions please post them here

johnpr98
Offline
Joined: Aug 20 1999

IMPORTANT FIX New PCI Latency Patch http://www.viahardware.com./

quote:
New PCI Latency Patch
George has released a new PCI Latency patch to address the PCI problems on VIA chipsets with the 686b south bridge. The most noticeable fixes are to the large file transfers and problems associated with the SB Live! This new version brings us up to 1.4 and if you have any of the following problems, I suggest you try this out:

Files copied from one IDE hard disk drive to another are corrupted
The computer stops ("freezes", "hangs", "locks up") when playing sounds, especially to a Creative SoundBlaster LIVE! card
The computer stops ("freezes", "hangs", "locks up"), or files are corrupted, when copying to or from an IDE DVD, CDROM, or CDRW drive
Sounds from a SoundBlaster LIVE! card are distorted when other PCI settings are adjusted in the computer
Before installing, its also suggested that users read the Read Me.

Posted on Friday, April 20, 2001 jasper Comments (4)

John Price
http://www.johnpr98.com

johnpr98
 
If you have any Forum Suggestions please post them here

johnpr98
Offline
Joined: Aug 20 1999

For info (As always)
http://www.heise.de/newsticker/data/jow-20.04.01-000/

German Update

quote:
BIOS-Updates gegen Datenbeschädigung durch VIA-Chipsatz

Die Mainboard-Hersteller reagieren auf den Fehler in der VIA-Southbridge 82C686B, durch den der Chipsatz teilweise Daten während Festplatten-Transfers beschädigt. Sie veröffentlichen einige BIOS-Updates zur Beseitigung der Fehler; doch Tests im c't-Labor zeigen: Diese Updates helfen nicht immer.

Die Aktualisierungen der BIOS-Software befinden sich im Beta-Stadium; einige Hersteller stellen sie nicht öffentlich zur Verfügung, sondern sie liegen bei verschiedenen Hardware-Sites:

Abit: Die Version ZT (Beta) KT7 von der Beta-Seite soll den Fehler beheben.
Asus: Bei Planet 3DNow! sind Updates für das A7V (1008 1B) und A7V133 (1004 2D) zu finden.
Enmic (NMC): Die offizielle Version 8TTX+ ist derzeit auf der Update-Seite noch nicht verlinkt.
Epox: Auf einer speziellen 686B-Update-Seite stehen BIOS-Aktualisierungen für das 8KTA2 und 8KTA3/3+ bereit.
Iwill: KK266, KK266-R (offiziell); bei AMDZone liegen die Updates für das KA266 und KA266-R; ein neues Flash-Programm scheint notwendig.
MSI: Bei AMDZone gibt es ein Update für das K7T Turbo, das den Fehler beheben soll.
Soltek: Die Version K3 für das 75DRV (offiziell) soll den Fehler beheben, doch Soltek selbst erwähnt davon nichts.
VIA: Auf der offiziellen Seite tut sich was: Dort finden sich auch die Dateien von Epox und Iwill, weitere BIOS-Updates haben demnach AOpen, Biostar, ECS/Elitegroup, Gigabyte und Legend QDI angekündigt.
Nach Tests im c't-Labor beheben diese BIOS-Updates den Fehler jedoch nicht zuverlässig. Auch Boards mit neuem BIOS und mit allen bekannten Work-Arounds produzieren Fehler: Sobald Daten über den zweiten IDE-Kanal fließen, während der PCI-Bus unter hoher Last steht, können diese Daten verfälscht werden. Betroffen sind nicht nur Boards mit der VIA-Northbridge KT133A: Übertragungsfehler traten auch auf Boards mit der AMD-Northbridge AMD-761 auf.

Der Fehler scheint auf den zweiten IDE-Kanal beschränkt zu sein: Sobald nur Transfers zwischen Geräten am ersten Kanal oder an separaten IDE-Controllern stattfinden, arbeiten alle Systeme stabil. Doch sobald Geräte am zweiten IDE-Kanal hängen, treten die Fehler auf — auch beim Kopieren von einem RAID-Controller zum zweiten IDE-Kanal der 686B. Falls Datentransfers zwischen Master und Slave am zweiten IDE-Kanal stattfinden, traten die Abstürze sogar gehäuft auf.

Der Nachweis des Fehlers gelingt oft nur mit viel Geduld: Bei einigen Boards trat erst nach mehreren Stunden die erste Störung im Datentransfer auf. Eines der Boards, für das schon ein BIOS-Update existiert, lief im c't-Labor ohne Aktualisierung des BIOS und mit kritischen Einstellungen ohne Fehler. Nicht auszuschließen ist auch, dass mehrere Fehler oder Instabilitäten in die Tests mit einfließen. So soll ein Mitarbeiter eines Mainboard-Herstellers betont haben, dass sich mit jedem Bugfix neue potenzielle Fehlerquellen ins BIOS-Setup einschleichen können.

Nicht endgültig entschieden ist auch, ob die Soundblaster-Karte, die bei fast allen Nachweisen des Fehlers zum Einsatz kommt, für die Störungen verantwortlich ist. Auch im c't-Labor laufen alle Systeme ohne diese Karte stabil. Die Ursache für Störungen können Konflikte zwischen der Karte und dem Onboard-Sound sein. Zu empfehlen ist also, beim Einsatz einer Soundkarte den Onboard-Sound zu abzuschalten – falls möglich (per Jumper und BIOS-Setup). Falls nicht, sollten unter Windows zumindest die Onboard-Sound-Treiber (meist "AC97 WDM" oder ähnlich) deaktiviert werden. Bei der Konfiguration der Soundblaster-Karten könnte auch die SB16-Emulation Schwierigkeiten verursachen: Man muss einige ISA-Ressourcen für diese PCI-Karte reservieren (siehe Hotline-Tipp vom November 1999). Die Soundblaster-Karten erzeugen laut einiger Support-Mitarbeiter eine unnötig hohe Last auf dem PCI-Bus; das spricht zwar gegen die Karte oder deren Treiber, aber der PCI-Bus sollte mit dieser Last klar kommen. Leser berichteten c't, dass sie auch mit anderen PCI-Geräten, die eine hohe Buslast erzeugen, die Datenverluste an der 686B-Southbridge beobachten. Die Soundblaster-Karte sorgt also wohl nur für verstärktes Auftreten des Fehlers, aber nicht für den Fehler selbst.

Im c't-Labor erwies sich als einziger funktionierender Workaround, den zweiten IDE-Kanal nicht zu benutzen. Dann liefen bisher alle Systeme stabil, auch unter etwas härteren Bedinungen (etwa gleichzeitiges Kopieren vom IDE-RAID auf 686B-Kanal, Kopieren vom IDE-RAID auf 100-MBit-Netzwerk, sich ständig bewegende USB-Maus, Abspielen von MP3, Swap-Datei auf 686B-Kanal). (jow/c't)

Translated with my Globalink

quote:
BIOS-Updates against data-damage through VIA-chip-sentence

The Main-board-manufacturers respond on the mistake in the [VIA-Southbridge] of 82C686B, through the the chip-sentence partial data during hard disk-N-transferences deface. You publish some BIOS-Updates to the elimination of the mistake; however test into the c't-laboratory shows: This [Updates] doesn't help always.

The updatings of the BIOS-software are into the beta-stage; some manufacturers put them not public to the possession, but they lie at different hardware-Sites:

[Abit]: The version [ZT] (beta) KT7 of the beta-side should remedy the mistake.
[Asus]: At planet of 3DNow! is [Updates] for the A7V (1008 of 1B) and A7V133 (1004 of 2D) to find.
[Enmic] ([NMC]): The official version of 8TTX+ is at present on the Update-side still not [verlinkt].
[Epox]: On a particular of 686B-Update-Seite BIOS-updatings for the 8KTA2 and 8KTA3 stand/ 3+ ready.
[Iwill]: KK266, KK266-R (official); at [AMDZone] the [Updates] for the KA266 and KA266-R lies; a new Flash-program seems necessarily.
[MSI]: At [AMDZone] there is a [Update] for the K7T turbo, that should remedy the mistake.
[Soltek]: The version of K3 for the 75DRV (official) should remedy the mistake, however [Soltek] mention away nothing.
VIA: On the official side itself does what: There itself finds also the files of [Epox] and [Iwill], further BIOS-Updates has therefore [AOpen], [Biostar], [ECS]/ [Elitegroup], gigabyte and Puts announces [QDI].
To test into the c't-laboratory this remedies BIOS-Updates the mistake however not reliable. Also boards with new BIOS and with all well-known [Work-Arounds] manufactures mistakes: As soon as data over the second IDE-canal flow, while the PCI-bus under high load stands, these data could be falsified. Involves is not only boards with the [VIA-Northbridge] of KT133A: Transfer-mistakes appear also on boards with the [AMD-Northbridge] of AMD-761.

The mistake seems on the second IDE-canal limited to be: As soon as only transferences between appliances at the first canal or at separated [IDE-Controllern] occur, works all systems stable. However as soon as appliances at the second IDE-canal hang, steps the mistakes on _ also at the copying of a [RAID-Controller] to the second IDE-canal of the 686B. Event data-transference between [Master] and [Slave] at the second IDE-canal occurs, stepped the crashes even heaps up on.

The [Nachweis] of the mistake succeeds in often only with many patience: At some board stepped first to several hours the first disturbance into the data-transference on. An of the boards, for the already a BIOS-Update exists, ran into the c't-laboratory without updating the BIOS and with critical employments without mistakes. Not to rule out is also, that several mistake or instabilities into the test with [einfließen]. So an employee of a Main-board-manufacturer should have accentuated, that itself with each [Bugfix] new [potenzielle] mistake-sources into the BIOS-Setup [einschleichen] can.

Not final decided is also, if the Soundblaster-card, that at almost all proving of the mistake to the employment comes, for the disturbances responsible is. Also into the c't-laboratory runs all systems without this card stable. The cause for disturbances could be conflicts between the card and the [Onboard-Sound]. To recommend is therefore, to switch off the [Onboard-Sound] at the employment of a [Soundkarte] too- if possible (per [Jumper] and BIOS-Setup). event not, should be deactivated under Windows at least the Onboard-Sound-drivers (most "AC97 [WDM]" or alike). At the configuration of the Soundblaster-cards the SB16-Emulation difficulties could cause also: One some ISA-resources for this PCI-card reserves ([siehe Hotline-Tipp] of the November of 1999). The Soundblaster-cards generate loudly some Support-employee an unnecessary high load on the PCI-bus; that speaks to be sure against the card or that drivers, but the PCI-bus should come with this load plainly. Readers told [c't], that they also with different PCI-appliances, that a high [Buslast] generates, the data-losses at the 686B-Southbridge observes. The Soundblaster-card worries therefore well only for reinforced appearance of the mistake, but not for the mistake.

Into the c't-laboratory [Workaround] proved works as more only, to use the second IDE-canal not. Then ran until now all systems stable, also under something harder [Bedinungen] (approximately simultaneous copying of the [IDE-RAID] on 686B-Kanal, copying of the [IDE-RAID] on 100-MBit-Netzwerk, itself constantly moving USB-mouse, Plays of MP3, Swap-file on 686B-Kanal). ([jow]/ [c't])

[ version to the printing][ Per [E-Mail] mails][<< Past][ Next>>]

Commentaries:
[Re]: [Hmmmmm]. ([Luzypher], 21.4.2001 of 11: 40)
. and the [User] is the Stupid. ([Luzypher], 21.4.2001 of 11: 38)
[Re]: Second canal= [Secondary] or [Slave]? ([TeeKay], 21.4.2001 of 11: 37)
more.

John Price
http://www.johnpr98.com

johnpr98
 
If you have any Forum Suggestions please post them here

johnpr98
Offline
Joined: Aug 20 1999

http://www.viahardware.com/

quote:
VIA says not their fault
To whom it may concern:

During the past few weeks, problems were reported on several web sites stating that one of VIA’s chipset has compatibility issue with Creative SoundBlaster Live PCI Sound card. After debugging, verifying, and checking compatibility between our VT82C686B and other devices, we concluded that the problem is not totally related to our chipset. We found that the system problem will happen only under certain configurations and certain procedures, and only occur under Athlon processor based systems with VT82C686B chipset and Creative SB Live card. We also checked whether our chipset has any compatibility issues in other configurations, and the result was we did not find any such compatibility issues.

The configuration is described as follows: One IDE HDD ATA-100 or one ATA-66 HDD and enable default DMA mode; one IDE DVD-ROM and enable fault DMA mode; one Creative SB Live PCI sound card; one VGA card; and copy file, W2KSP2.exe (Win2K service pack2).

Consequently, we also developed a new IDE driver to resolve the compatibility issue with the Creative SoundBlaster sound card, and this IDE driver will be posted at our website for users who have Athlon based systems using this configuration.

Best regards,

VIA Technologies, Inc.

Discuss this subject further in our forum.

John Price
http://www.johnpr98.com

johnpr98
 
If you have any Forum Suggestions please post them here

PD
Offline
Joined: Aug 6 1999

Thanks for maintaining the updates on this issue John.

I'm sure everyone has been watching this with interest, and sifting through the often-unreliable noise that gets generated elsewhere on the Web can at best be very time consuming.

On the issue of audio adapters generally, I have seen a fair amount of ignorance prevailing about the integrated audio facilities of some mainboards, where fairly sophisticated integrated hardware solutions have stupidly been dismissed just because they are something that have been acquired at little or no additional cost above that paid for mainboards that either don't have the facility, or rely on the integrated solution directly within, say, the VIA VT82C686x PCI, mainboard Super-I_O Integrated Peripheral Controller.

Of late I have not been able to devote and contribute as much time as I would like to these Forums generally because amongst other projects I have been engaged in some quite intensive testing of a increasingly large number of mainboards.

One of these has been an ASUSTek A7M266, which has an excellent C-Media CMI8738 audio controller.

More on this whole subject later, but suffice to say that whilst I am no audiophile, to me the sound quality is very good indeed which is an opinion shared by others who have heard a brief side by side comparison with a system installed with a Creative Labs SoundBlaster Live!

I doubt whether many people would gain any advantage whatsoever by replacing it with something that wasn’t really exceptional both in functionality and in price!

Cheers,

------------------
Paul

Paul Dutton
DVdoctor R&D

HEXUS.swankyDynamicSignature - Give it a click!

Assimilator1
Offline
Joined: May 12 2001

What about surround sound?

Also don't those on board audio chips (excluding creative ones)rely more on the cpu? ,no good for us gamers!

------------------
Want to search for ALIEN signals?SETI@home.
Want to join a helpful & friendly team? Team Anandtech

Join mankinds ultimate search SETI@Home
Team Anandtech 6th place :)

johnpr98
Offline
Joined: Aug 20 1999

http://www.viahardware.com/download/index.shtm

quote:
VIA Driver Download Page

4-in-1 Drivers

VIA has been offering their 4-in-1 driver for quite some time now. It is designed to solve all compatibility problems with Microsoft OS's and VIA chipsets. As MS has traditionally provided very little support for third party chipsets, the chipset manufacturers must provide their own patches.

The VIA 4-in-1 is updated regularly, as bugs are found and performance is tweaked. It is recommended you install the latest 4-in-1, unless you are experiencing problems. Also, keep in mind that individually downloaded drivers may be newer, as we tend to get updated drivers before they are integrated into the 4-in-1. Update- If you are having issues with the installation program, and its looking for a VIAGART.CAT file, read this post.

4.31 FINAL- Now integrates the new official VIA 686B issue patch.

johnpr98
 
If you have any Forum Suggestions please post them here

PD
Offline
Joined: Aug 6 1999

Last known official postion from VIA Technologies:

*********************************************

To whom it may concern:

During the past few weeks, problems were reported on several web sites stating that one of VIA’s chipset has compatibility issue with Creative SoundBlaster Live PCI Sound card. After debugging, verifying, and checking compatibility between our VT82C686B and other devices, we concluded that the problem is not totally related to our chipset. We found that the system problem will happen only under certain configurations and certain procedures, and only occur under Athlon processor based systems with VT82C686B chipset and Creative SB Live card. We also checked whether our chipset has any compatibility issues in other configurations, and the result was we did not find any such compatibility issues.

The configuration is described as follows: One IDE HDD ATA-100 or one ATA-66 HDD and enable default DMA mode; one IDE DVD-ROM and enable fault DMA mode; one Creative SB Live PCI sound card; one VGA card; and copy file, W2KSP2.exe (Win2K service pack2).

Consequently, we also developed a new IDE driver to resolve the compatibility issue with the Creative SoundBlaster sound card, and this IDE driver will be posted at our website for users who have Athlon based systems using this configuration.

Best regards,

VIA Technologies, Inc.

*********************************************

Interesting...

------------------
Paul

Paul Dutton
DVdoctor R&D

[This message has been edited by PD (edited 18 June 2001).]

‘Since posting we have been contacted by John Gatt of VIA Technologies’ Technical Support operation, and he has asked us to advise that this statement was originally issued on the 2 May 2001, and that the VIA 4-in-1 Service Pack version 4.31 contains the necessary resolution to the incompatibility issue between the VIA VT82C686B PCI Super-I_O Integrated Peripheral Controller and the Creative Labs SoundBlaster Live! audio adapter.

There is a file identified and detailed as “VIA's fix for KT133/A systems with Soundblaster Live sound cards. Should clear up any file corruption problems that may occur with the 686B southbridge.” which is available for download from Viahardware at http://www.viahardware.com/download/viapfd100.exe.

I have checked and this compressed executable contains a file VIAPFD.SYS of the same name, date and size as a file which is contained within the official VIA 4-in-1 Service Pack version 4.31 and the 4-in-1 version 4.32 Beta which are both also available for download at Viahardware.

I assume therefore that no additional benefit is to be gained by the installation of the viapfd100.exe file if you are, as you should be, running version 4.31 or later of the VIA 4-in-1 Service Pack.

John goes on to say that there is no "new" bug at all.

Obviously I wasn’t suggesting that there was.

What I was doing was helpfully informing you that VIA was on the case, they seemed to have identified how to recreate the issue and that there would be a fix available.

Which although this seems to be the case even now I cannot find it expressly identified as such on any of VIA’s English language Web Sites.

In fact I have not been able to find any reference to this widely discussed incompatibility issue on any of VIA’s English language Web Sites and if this whole issue has been confusing to you guys imo it sure could have been less confusing if VIA had posted up a helpful reference of their own.

From a customer service perspective I can’t see the argument for not doing so.

Possibly the most comprehensive reference to this is currently and, with genuine respect to the owners of the site, rather surprisingly at the independent site Viahardware www.viahardware.com.

Johnpr98 has pointed this out in the thread above but I’ll mention it again here; the document to be reading is titled 686B Bug FAQ and is available at this URL: http://www.viahardware.com/686bfaq.shtm.

It does however fail to mention to the issue being exclusively related to the integration of AMD Athlon microprocessors with these components.

I’ll try and see what AMD’s official position is on this.

In any event could anyone with the VIA VT82C686B and SoundBlaster Live! combination who has previously, or continues to experience any file corruption problems please post up your experiences here?’

[This message has been edited by PD (edited 19 June 2001).]

HEXUS.swankyDynamicSignature - Give it a click!

johnpr98
Offline
Joined: Aug 20 1999

Thanks for maintaining the updates on this issue Paul.

Heres a FAQ with some interesting stuff http://www.viahardware.com/686bfaq.shtm

This also looked interesting
VIA KT133/A WPCREDIT Memory Tweak Guide http://www.viahardware.com/memtweakguide1.shtm

I am also coming to the conclusion from various posts around, that enabling
'Enhanced chip performance' in the BIOS is a major tweak for VIA.

Cheers

John Price
http://www.johnpr98.com

[This message has been edited by johnpr98 (edited 18 June 2001).]

johnpr98
 
If you have any Forum Suggestions please post them here