Cell Relay Archive

Cell Relay Retreat>List Archive>month:1998-May> msg00188



[Date Prev][Date Next][Thread Prev][Thread Next]  
  [Date Index][Thread Index][Author Index][Subject Index]

Re: ATM PCI network interface card

  • From: Berry Kercheval <berry@jhereg.kerch.com>
  • Date: 25 May 1998 12:53:57 -0700

Martin Winter <Martin.Winter@oen.siemens.de> writes:
> 
> I think it would be difficult to get a card, where single cells can be
> addressed. From  what I have seen, all cards provide an interface at the
> AAL5 level, but I have never seen an interface at a lower level.
> 
> The reason is that all cards do AAL (all cards AAL5, some cards other
> AALs) in hardware, because it would be impossible to do this in software
> with a reasonable performance. So any interface below that would be of
> no use for "normal" applications.

No, I disagree, it should be simple to find a card that will allow an
application direct access to cells.  It may be difficult to get this
information out of manufacturer's representatives or web pages.  Some
may not support it in their software drivers, either.

I know for certain that at least cards using the NEC SAR chip *can*
access raw cells; when I was at Xerox PARC we build an S-BUS card with
this chip and I wrote the driver.  With a bit of register-poking it's
fairly straightforward to have the raw cells for a given VC (or ALL
VCs, but don't do that unless you're building a sniffer) delivered 
to the driver, and hence the application.  

So at least one chip has the capability to deliver cells.  I know that
at least two manufacturers used this chip, but I'm out of touch with the
marketplace now.

  --berry