> did they give a reason for not using the 3c905 cards. I am using 5 of these
> right now in 2 different machines with no problems but if it is not a good
> idea to use these, what should I use in the next set of Linux machines that I
> setup. One thing though, to make those 3c905 cards work, I did have to get a
> newer 3c59x.c driver file than the one that came with the kernel source that I
> used, (v2.0.30)
there are no docs available for these new 3COM chips which resulted
in very bad drivers for this cards (it's a surprise for me that Don Becker
even managed to get these cards working). so _many_ people suffered
from complete network lockups and/or very bad performace on 100MBit
when not using busmaster dma (which fixed lockups).
same happened to us when we bought 10 boomerang cards for PPro machines
in a 100Mbit net end of last year. and I spent _much_ time to get an idea
whats really broken and trying to hack and fix the driver and testing new drivers
over and over again (and hacking cron jobs every minute to restart
locked up networks) until I finally realized that there are no chip docs at all
so there was not too much hope to get a reasonable driver in reasonable time
and thus I gave up and replaced all cards with DEC tulip chips in May/June
(from LANTRONIX, works great and very cheap!).
now I know that the most recent 3c59x.c driver from Donald seems to be
(somewhat?) better for boomerang (but not for vortex) but it's still
the problem that 3COM doesn't supply docs and that a similar disaster
might happen again for the next chip generation so I decided for myself:
no more 3COM in the future and I told 3COM managers about this decissions
and it's reasons...
Harald
--
All SCSI disks will from now on ___ _____
be required to send an email notice 0--,| /OOOOOOO\
24 hours prior to complete hardware failure! <_/ / /OOOOOOOOOOO\
\ \/OOOOOOOOOOOOOOO\
\ OOOOOOOOOOOOOOOOO|//
Harald Koenig, \/\/\/\/\/\/\/\/\/
Inst.f.Theoret.Astrophysik // / \\ \
koenig@tat.physik.uni-tuebingen.de ^^^^^ ^^^^^