[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: What's your bus?
Thanks for providing information on Spacewire. It appears to be
somewhat similar with IEEE 1394 which is now a common video
transfer method for consumer devices and some storage devices.
The newer standards like Serial ATA or SCSI provide far more
bandwidth at a low cost. These have been built on the work done
by the ANSI T11 Fibre Channel group at 1Gbit/sec. This work
was used by Ethernet for 1Gbit/sec, then Ethernet did the 10Gbit/sec
4-lane XAUI which Fibre Channel also uses.
I did not see when the Spacewire spec was done, but the bandwidth
on the website indicates it is 200 Mbits/sec. Fibre Channel started
in 1997 at about 1000 Mbits/sec and is now at 10K Mbits/sec.
The 8bit/10bit encoding provided by IBM is the common encoding
method uses by most of the high speed connections. I would be interested
in the future planned speed improvements for Spacewire.
I would also like information on Star Fabric since I have not heard of this
As for the original question about what Bus do I like, I like them all.
I am proposing a common physical layer which can be used by most
of them. I am the co-editor of the 10GFC spec and have proposed
Fibre Channel be expanded to provide 20 & 40GFC which may also
be used for Infiniband, PCI-Express, Serial RapidIO and Ethernet.
These are all self clocking 8b/10b encoded data protocols.
I am not an expert on Hypertransport, but I thought I heard it is
unencoded and requires a separate clock. Perhaps someone
can provide feedback on this.
The start of the multilane transfers were proposed within the
IEEE committee as the HARI & SALI interfaces which became
XAUI & XGMII for Fibre Channel and Ethernet.
The various standards had different requirements, so the usage of
the common features diverged. I believe it is now time them to
converge with a standard XAUI & XGMII interface with different
modes of operation to account for the differences. I will be
presenting a detailed proposal at the Fibre Channel meeting
in Huntington Beach in February 2003. Go to WWW.T11.com
for details. Details on the various standards are contained in
My next document will detail the new illegal k-characters to be
used on the XGMII interface to control the new features used
on the various standards: Beacon, Loss-Of-Signal, Receiver-Detect,
power-down TX/RX/Both, Electrical Idle (D+==D-), alignment
character and a few others.
Thanks - Curt
Paul Walker wrote:
> A common feature of the ones you mention is that they are not buses,
> they use switch fabric.
> So does SpaceWire, a derivative from IEEE 1355, that is simpler and more
> flexible than any of those you mention. It was way ahead of its time,
> but the plethora of new standards, all of which follow it and all of
> which are more complicated, just show that its time will come.
> You can reach most of the information about SpaceWire and IEEE 1355 from
> our web site:
> Best regards
> Paul Walker
> In message <OF142CD769.C7EBA262-ON88256C94.000425FB-
> 88256C94.firstname.lastname@example.org>, Alan Deikman <Alan.Deikman@znyx.com> writes
> >I'd like to hear back from anyone with an opinion on this. What do you
> >think of as the ultimate bus after PCI and why?
> >1. Hypertransport
> >2. RapidIO
> >3. Star Fabric
> >4. PCI Experss
> >5. Infiniband
> >Any others that will be players? I get asked this sort of question all the
> >time and I need some new material. :)
> >Alan Deikman
> >ZNYX Networks, Inc.
> Paul Walker
> CEO, 4Links Limited, Chair of the 1355 Association
> www.4Links.co.uk www.1355.org
> 4Links Limited --- Boards, chips, IP and consultancy ... for Links
> P O Box 816, Bletchley Park phone +44 1908 64 2001
> Milton Keynes MK3 6ZP, UK fax +44 1908 64 2011
org:LSI Logic Storage Components;Serial Protocol Development
title:Distinguished I/O Architect, Serial Protocols
adr;quoted-printable:;;1873 Barber Lane=0D=0AMS AH-260;Milpitas;California;95035;USA