broadcast recieve problem with FA310TX and RH 5.2

Donald Becker becker@cesdis1.gsfc.nasa.gov
Mon Mar 29 19:10:45 1999


On Tue, 30 Mar 1999, Anthony Rumble wrote:

> > There are alternatives.  Generic tulips seem scarce (which is a shame).
> > I've got a source of $20 RealTek cards, though, and hope to post numbers
> > on them very soon.  If they perform at par (that is, 97% efficient or so
> > for large packets) I'll convert to them exclusively.
>
> These price dynamics are not the same here in Australia..
> NetGear cards are about AUD $80
> 
> Intels are around AUD $90/$100 each..
> A small price to pay for something that works..

I would pay 20% extra for the Intel cards, but in the US it's 2X or more.

> Realtek have Tulip based cards?
> I thought realtek only had Ne2000 clone garbage..

The RTL8139 10/100 chip is better than their 10mbps-only NE2000 clones, but
it's still a low performance chip.  You have to copy every Rx packet from
the single Rx ring buffer, most transmit packets must be copy-aligned into
bounce buffers.

It's too bad.. RealTek was a leader in integrating the transceiver onto a
single chip, implementing 802.3X flow control (even on the RTL8029 PCI
NE2000 clones!), and Wake-On-LAN.  They just didn't get the data transfer
part of the chip right.


I haven't been able to reproduce the broadcast Rx problem with the Netgear
cards yet.  If the reports are accurate -- it only happens with some
machines -- then it might take a while to figure out what is going on.

Donald Becker					  becker@cesdis.gsfc.nasa.gov
USRA-CESDIS, Center of Excellence in Space Data and Information Sciences.
Code 930.5, Goddard Space Flight Center,  Greenbelt, MD.  20771
301-286-0882	     http://cesdis.gsfc.nasa.gov/people/becker/whoiam.html