Announcement

Collapse
No announcement yet.

Data corruption with via chipsets

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Data corruption with via chipsets

    http://www.theregister.co.uk/content/3/18267.html

    I know that sometimes The Register can play jokes on the public but this article looks to be legit.

    Haig

  • #2
    It sure looks legit. Note that the SBLive's presence makes matters worse. Typical....

    Dr. Mordrid

    Comment


    • #3
      I installed My new RAID system today and copied 19+ GB of data from the primary slave to the secondary master (both IBM DTLA drives), then I copied it back to the RAID!

      NO Corruption, (No I have not checked all 19+ GB but enough to be certin) And i Have a SBlive in my system.

      This might be that some one at the register has a thorn in his hide about VIA chipsets OR that he has used a "Bad motherboard maker" tm Motherboard!

      I Dont like VIA, Infakt i hated every MVP3 board i saw and definitly their NIC, but in the same way people dislikes Intel's way of selling slow cpu's at crazy prices i't seems to be "in" to hate VIA chipsets by any reasons.

      For every person who have gothen thier data corupted by a VIA chipset i can find one that has had the same experience with Intel, SIS, ALI & etc.
      If there's artificial intelligence, there's bound to be some artificial stupidity.

      Jeremy Clarkson "806 brake horsepower..and that on that limp wrist faerie liquid the Americans call petrol, if you run it on the more explosive jungle juice we have in Europe you'd be getting 850 brake horsepower..."

      Comment


      • #4
        Technoid, it's not a joke or a vendetta from The Register.

        The bug was discovered at Au Ja! , a German hardware site. Their site gives a very full report of their extensive testing and the hunt for this bug. (in German only I'm afraid)

        - the bug occurs with the VIA 686B version of the Southbridge

        - the bug is that if you have a HD on IDE1 and 2 od the southbridge (NOT on separate IDE or RAID ports, like Highpoints or Promise) and you DMA copy a large file between them, while putting load (like SBLive does) on the PCI bus, you get data corruption

        - other IDE devices may be affected (e.g. copy from DVD drive to HD) but nothing for sure yet

        - there's a BIOS fix (and I'm sure we'll see lots of new BIOS updates appearing next few weeks):
        1. PCI delay transaction OFF
        2. PCI master read caching OFF
        3. PCI Latency 0 (or at least stay below 32)

        This will have an impact on speed but only a very small one.

        If your mobo bios doesn't allow these settings you can do it via WPCREDIT/WPRSET tools. Email me and I'll translate the info for this from the Au JA website if you need it.

        Note: the Au JA website also posted a remark by MSI and Pinnacle: Pinnacle observed that Liveware transfers continuously null-bytes over the PCI bus to the SB Live, even when no sound is played. In the worst case, this led to 30% of the available PCI bandwidth being wasted. This caused the problems during video capture. When one deactivates the MIDI functions of the SB Live, this problem is solved.

        J-kun

        Comment


        • #5
          Yes, it IS REAL.

          Technoid:

          If you're testing this with a RAID card (as opposed to using Win2K or NT4'S software RAID) then you're not using the VIA chipsets internal ATA support.

          The Promise, 3Ware and HighPoint (yech) chips carry that load and so wouldn't be affected.

          Dr. Mordrid


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

          Comment


          • #6
            The File transfere was betwen DTLA drives on the Primary and Secondary Motherboard Ide Conectors.
            I was going to use my secondary 30gb DTLA with another 30 gb DTLA to use raid, not wanting to lose 19+gb of data i disconected my DVD & CD-RW drives from the secondary motherboard conector and atached a fresh 60gb DTLA to it and moved the data!!

            The Promise raid Was not involved in that first move!

            I was using the Internal VIA ATA system in the initial move of DATA!!

            J-kun:

            I always turn of PCI delay transaction because i once read that it buffers ISA --> PCI transfers, that might not be true any more but I do it almost automaticly when i first setup bios.
            Mayby thats why I never had any problems.


            If there's artificial intelligence, there's bound to be some artificial stupidity.

            Jeremy Clarkson "806 brake horsepower..and that on that limp wrist faerie liquid the Americans call petrol, if you run it on the more explosive jungle juice we have in Europe you'd be getting 850 brake horsepower..."

            Comment


            • #7
              I read the original German article and it makes an interesting observation about the drivers sending null bytes to the card all the time (even if it's idle).
              I wonder if this explains the troubles everybody's having with Creative PCI cards? They also write that it can be fixed by disabling the MIDI part of the card !!!
              Resistance is futile - Microborg will assimilate you.

              Comment


              • #8
                Yes, I read the info about the null bytes and the MIDI also. The only problem is that, under W2K, you can't disable the MIDI interface alone. If you disable MIDI, then the Wave interface and the Mixer interface are disabled too. Another stupidity of the Creative drivers. In other words, either you disable the whole card, or you don't disable anything. Which means I cannot check the validity of the info about the MIDI interface being the problem.
                However, I would like to point out that the PCI problem does not exist under WXP. Using the exact same hardware, drivers and capture programs.
                Michka
                I am watching the TV and it's worthless.
                If I switch it on it is even worse.

                Comment


                • #9
                  The SBLive does more than that. It also consantly uses other system resources like the system timer interrupt.

                  Dr. Mordrid


                  [This message has been edited by Dr Mordrid (edited 15 April 2001).]

                  Comment


                  • #10
                    I don't give a rat's ass what other people think of the system components I'm using if they work correctly for my applications and the performance meets my needs.

                    The problem is when a few people for some reason don't see the difficulities others are having and then conclude it can't be the hardware -- adding more heat than light to the issue.

                    My friend owns a small computer/electronics store and sells a lot of Via based stuff -- usually works fine for low-end apps -- none of its has worked for me, Every shipment has the same name and box but the chips and motherboard are different revisions so there is practically no way to tell which Via motherboard is "safe" to buy, thus I recommend against buying Via for intensive applications.

                    Now that some of the manufactures involved are finally owning up to the problems they might be fixed eventually.

                    --wally.


                    [This message has been edited by wkulecz (edited 16 April 2001).]

                    Comment


                    • #11
                      Ok, Dr Mordrid!
                      you are right and I am wrong!

                      But Why don't I notice any of these Problems?

                      The only time i had the kind of problems described was when i had a SS7 MVP3 VIA motherboard and that was before i got my SBlive!(crackling noise, corupted files, green flashes in my captured avi's etc etc)

                      That Via Board drove me almost insane.
                      (Now i can make that POS work by installing the latest "4 in 1"'s but i rather have it securely hidden away in a closet)


                      I mean if someone told you that your favorite soudcard was a POS and your motherbaord was an even greater POS because it used Intel chipsets wouldent you argue back because you thought otherwise?
                      If there's artificial intelligence, there's bound to be some artificial stupidity.

                      Jeremy Clarkson "806 brake horsepower..and that on that limp wrist faerie liquid the Americans call petrol, if you run it on the more explosive jungle juice we have in Europe you'd be getting 850 brake horsepower..."

                      Comment


                      • #12
                        OK, what can I replace my SBLive with? Is there any other sound cards that don't hog resources?

                        ------------------
                        ASUS A7V133, 256MB CAS2 Micron PC133, "Copper" Athlon 1000@1120MHz (8*140), GlobalWin FOP38 with Delta 7000 RPM 38CFM fan. Asus V7700 Deluxe GF2 GTS, 3COM 905C NIC, Adaptec 2940UW Ultra-Wide SCSI, SBLive PCI 1024. Segate Medalist PRO 9.1GB UW SCSI 7200RPM, Seagate Barracuda 4.3GB UW SCSI 7200RPM, Maxtor 20.4GB ATA66 7200RPM 2MB Cache, 8GB tape b/u, Sony IDE Spressa 8x/4x/32x CDROM RW, RAID 0 stripe set on Promise ATA100 with dual Quantum Fireball Plus LM 30GB ATA66 drives.
                        Tyan Thunder K7, 768MB Registered DDR ECC, 2xMP2200+, Radeon 9700 Pro, Adaptec 2940U2B Ultra2 SCSI, TB Santa Cruz, Pyro 1394DV. RAID 0 stripe set on hacked Promise UltraTX2 with dual WD 120MB SE drives. HP DVD200i DVD+RW drive.

                        Comment


                        • #13
                          Ok!
                          It's an hardware problem and there is ways to fix it by turning things off in bios - ok?

                          I don't ignore the problems people have with their VIA boards and SBlive cards.

                          It's just that I have under the resent year fixed to many computers with VIA motherboard's that the customer thought was behaving bad because their new shining motherboard and athlon cpu.

                          The four biggest problems have been:

                          1: Less than 300 watt PSU's
                          2: PC66 SDRAMS that is run at 133MHz
                          3: Not installing any VIA drivers at all
                          4: Windows Bugs that even makes Intel BX boards behave flaky.

                          Besids the asus board that had died because of a 200W PSU i have never had to replace any motherbard because they had any hardware problems.

                          (on the other hand i have replaced a lot of them that the lightning struck)

                          VIA sells Chips to motherboard manufactors that then implement them on motherboards.
                          If they tweak the chips beond their capacity to work reliably is that necessary a hardware problem or VIA's direkt fault?

                          All the releases of 4 in 1 makes me belive that VIA are trying to support their chips to the end user when they could just do as many other chips makers, tell every one that it isn't their problem and that the customer should speak with their motherboard vendor.

                          If agressive BIOS settings that makes motherboards go crasy are hardware faults then all motherboards have hardware faults!

                          There is alot of other arguments that I would like to say but I'm afraid that many would chose to misunderstand my bad english and just think that I'm a Idiot!
                          If there's artificial intelligence, there's bound to be some artificial stupidity.

                          Jeremy Clarkson "806 brake horsepower..and that on that limp wrist faerie liquid the Americans call petrol, if you run it on the more explosive jungle juice we have in Europe you'd be getting 850 brake horsepower..."

                          Comment

                          Working...
                          X