problems with 3com and intel 100MB cards
becker at scyld.com
Thu Oct 10 00:00:12 PDT 2002
On 9 Oct 2002, Marcin Kaczmarski wrote:
> I try to make calculations from materials science using parallel code
> with lam library. I use 6 athlon cluster connected with fast ethernet
> cards. I use 3com3c905c-tx cards with dual channel bonding connection.I
> use redhat linux with 2.4.x kernel and 3c59x driver. Everything works
> quite good but unfortunately I observed that my program (CPMD) simply
> dies without any info in logs (only confirmation from lam that process
> died and that`s all). I tried also intel eepro100 cards and with no
> channel bonding and the results are the same.
This sounds like an application bug.
> I consulted this problem
> with some man from Germany who is admin of 60 pc cluster and also uses
> the same program - CPMD. I got the information from him that 3com and
> intel cards are very unreliable, they cannot bear extremely high
> network load.
While we ship my driver set because many of the modified drivers in the
kernel have minor problems, calling both the 3c59x and eepro100 drivers
"very unreliable" is horribly wrong. A typical problem is that 3c59x
driver won't work with one specific version of a CardBus card. There is
nothing subtle about this type of problem -- the NIC won't work at all,
or will emit many warning messages.
You should ignore any other advice from this person. (Feel free to
ignore my comments as well, but I might have some experience in this
area ;-) ).
> I know about SCI dolphinics cards but they cost > $1500 each. I also heard
> from the same man that the only one reliable 100 MB cards were that with
> dec tulip chip.
Except that the modified tulip driver in the 2.4 kernel had been broken
with the real 21143 Tulip chip when used with SYM transceivers... you
have to download my driver to get those boards to work.
> I also heard about linux alpha clusters that fail to operate while
> running with 3com cards.
Nope... Scyld Beowulf for Alpha was qualified with at least five
different 3Com NICs.
Donald Becker becker at scyld.com
Scyld Computing Corporation http://www.scyld.com
410 Severn Ave. Suite 210 Scyld Beowulf cluster system
Annapolis MD 21403 410-990-9993
More information about the Beowulf