In message <Pine.HPP.3.93l.961005163637.5011C-100000@unix26.andrew.cmu.edu>,
Andrew Y Ng writes:

>well, I couldn't either.  I dunno if the bug is reported or not, but I know
>many ppl in here (well, everybody who uses the ep*, i think) have the same
>problem.

No, *not* everyone.  I have never, ever, ever seen anything like
this,  with several 3c595s running on 100-TX.  I regularly hammer
them with ttcp throughput tests.   These tests send an burst of
256 Mbytes of data from a sender to a receiver.    I think that counts
as a large number of packets.

I have not seen the PCI ep driver freeze.

That's why I've been asking about media type and hardware revisions;
to see if there's any correlation with, say, AUI.

A few more questions:

	* has anyone seen this with the 100Mbit PCI version of the boards?
	  (I haven't, I don't think Mika has either).

	* has anyone seen this with the EISA version of the board?


	* Is port-i386 really the right place for bugs with PCI boards?