It is true that the "nodma" option did solve the VIA disk corruption. This was in turn caused by a network hang-up the cause of which we have not yet located. Nothing was written to the syslog file.
I have found 2 MBs for AMD processors that do not use the VIA chipset: (a) MicroStar 745 Ultra uses the SIS 745 chipset. (b) MicroStar K7N420 Pro uses the nVIDIA MCP-D chipset. Does anybody know if these chipsets have their own problems with the Linux kernel-2.4.18 ? Bjarne On Sun, 2002-04-14 at 11:11, Guy Zelck wrote: > Jose M. Sanchez wrote: > > >Just to chime in on this subject... > > > >The AliMagic and VIA based motherboards have "broken" chipsets. > > > >I've spoken to both companies engineers about this issue. > > > >This discussion arose because both chipsets have problems with DMA > >overlays when using the BT8xx type TV tuner cards... > > > >The problem is with clock timing (the famous broken clock timer chip!!!) > >and DMA xfers. As you push the udma speed the spurious clock pulses > >wreak havoc with the memory writes. > > > >The Windows drivers try to make the OS work around the problem... > > > >This may be why people are seeing corruption. Couple this with Civilme's > >statement on WD drives and lack of CRC DMA data checking, and it's a > >wonder that Linux even runs on these systems. > > > >Turning OFF DMA or reducing UDMA to 33 seems to help with this > >problem... > > > >Your milage may vary though... > > > >-JMS > > > > > Hi, > > Valuable input. I've got an Intel chipset but indeed a bttv tv tuner and > a Firewire card from VIA technologies and suffer the same problems. > > Guy. > > > > > > ---- > > Want to buy your Pack or Services from MandrakeSoft? > Go to http://www.mandrakestore.com
Want to buy your Pack or Services from MandrakeSoft? Go to http://www.mandrakestore.com
