[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[f-cpu] DATE2002, day 1 report


The DATE2002 exhibition started today in Paris.
This european (and in fact, second worldwide ?)
EDA and microelectronics company gathering is 2x larger
than last year (in Munich), it took me a whole day before i saw
most booths. It is also very positive from the point
of view of F-CPU and "open source HW designs" : i have
a lot of good news.

First i was nicely helped by Cadence engineers on their
booth, they quickly gave me some insights on ncsim
(that is installed on my laptop, which i brought)
and they don't seem afraid of helping an "open source
project made by hobbyists" : they were enthusiastic
about their product and tried to share some knowledge.

It is good to see such a large company having this
behaviour with someone who is not a "customer" and
i apreciated it. Thanks to Martyn Pollard's help,
i was welcome and i hope to continue to work in
such good terms.

I have also met someone from a company called "bridges2silicon"
(using Aptix hardware) who uses Wishbone as a demo
application. He is interested by the "free hardware movement"
so i pointed him to the Opencollector.org site. I was
surprised by his reaction, the opposite from the
ARM representatives in Munich last year ;-) I hope we can
partner in a way or another.

I have seeked help from most emulator makers : too early.
this is a way of saying that making F-CPU run on a FPGA
or a dedicated machine which is connected to a dedicated
online development server is not going to become
a reality soon. Look at http://www.f-cpu.org/EDAserver.html
for more background about what was thought a year ago.
One company's representative was not "disturbed" by this idea,
though. He works for ISYTEC (a german company which is
represented in France by Answer Systems). There are still
two problems : it depends on "heavy proprietary software" to
synthesise the FPGA bitstreams, and the HW is very expensive too.

Concerning the sharing and hosting of VHDL synthesis ressources,
i think that it could be possible if we team with a
french VHDL club that is currently being created in
a french computer school (epita.fr). They have "suitable"
machine rooms (cooled, with enough electric power and
internet connexion) and thousands of students from which
at least a few might help 24/7 (server configuration/administration,
maintainance etc.) but now the problem is to get
an emulator/FPGA board with the corresponding licences
and support software. The shared emulator problem is still
being discussed by the french F-CPU team. There might be
a solution at a time or another. I would be happy to get
help from my last employer, because even an "old"
machine should be enough for fitting a whole F-CPU.
Unfortunately, experience shows that if it was possible,
we would have had it already.

Later, while discussing with a sales/marketing manager,
the old synthesis problem popped up and he directed
me straight to someone from Synopsys. It was the end
of the day so i was not surprised anymore (after all
what happened). I have the address of a person to contact
and i wish that the good star will continue to shine,
because synthesis is currently the #1 problem for F-CPU.

OTOH, being in contact with the 3 leading EDA companies
makes my personal situation difficult. Having worked
for Meta Systems (the emulator branch of Mentor Graphics),
i have to be very careful because these companies are
always in court against each others. I wish we could
work together in good terms and avoid any useless tension.
"Design and let design".

I did not have time to say "hello" on all the booths,
so the panorama is not finished. For example, i have
to return to some booths, such as this of EUROPRACTIVE
where they will explain me the issue of the fundrie's
design kits. This day was very tiring but unexpectedly
positive, in a period of economic degradation.
i thought that my F-CPU lobbying would look misplaced
but in contrary, i was more helped in a single day than
during months.

What looked like an utopia 3 years ago (and it was one)
is now becoming a tangible reality : LEON and OPENCORES
are now credible and F-CPU should follow, since we
have some good code now.

I'll be at DATE tomorrow, where i meet Nicole Ciry (from
APRIL/FSF Europe, who helps us setting up the french
F-CPU association) at 12, near the entrance. If you want
to join, be there :-) [for those in/around Paris]


PS: ok for publication.

PS2: i found a new/better barrel shifter structure
this morning in the metro :-)
To unsubscribe, send an e-mail to majordomo@seul.org with
unsubscribe f-cpu       in the body. http://f-cpu.seul.org/