Performance

[Vojtech Pavlik]

On Wed, Oct 13, 1999 at 08:29:03AM -0700, Brian Macy wrote:

> > I've been achieving best results with 3c905-B and -C lately. 12.5 megs/sec
> > is a good result (unidirectional) with full duplex and for bidirectional transfer
> > I'm getting a bit more than 10 megs/sec. With tulip, the numbers are lower,
> > especially for the unidirectional transfer, with the bidirectional one
> > the difference is not that big.
> 
> I assume you mean Mbit right?

No, of course not. That's megabytes per second. 12 Mbit/s would be a little
too little for a 100 Mbit/s card.

> Anyways, I have had nothing but bad luck
> with the 3c905b under Windows and Linux platforms... in fact I know
> several users and sysadmins who loath the things. Don't know about the
> 3c905c.

I've had very bad experience with 3c590 and 3c900-A, but the 3c905B-TXNM's
I have really work like a charm.

> Which the KNE100TX uses the 21143 and the KNE110TX uses a
> Lite-On. Kingston was at Linux Expo in San Jose and I was happy to see
> them there (I think they were the only consumer networking company
> there)... the techie there even knew what he was talking about :)
> Supposedly there is a bad rev of the KNE100TX that has a nickel sized
> chip (I think the transceiver, not the 21143???) near the top of the
> board (away from the PCI slots). The fixed rev has a smaller chip nearer
> the PCI slots and closer to the bracket. The guys there said if you
> happen to get one of those cards they'll be happy to replace it with a
> newer rev.

That sounds like two different types of a MII transceiver. But without
more (which transceiver is bad) info I think this is quite worthless ...

Vojtech