Re: [oc] Inquiry

[sdb@cloud9.net (Stuart Brorson)]

Greetings --

I can't say that I agree about the quality of the cores on opencores.
I have played around with some of them, and they all worked well as
far as I could tell.  Of course, I would never use an opencore without
doing a lot of modification to it in order to integrate it into a
larger design.  This is the beauty of using an openocre over a
proprietary -- and probably encrypted -- core: you can hack it to suit
your needs.  Anyway, during the modification process one can discover
and fix any bugs in the core.

The problem I have with opencores in a commercial environment is that
the licencing of the core is problematic.  Some of the cores are
GPLed, which means that you are supposed to distribute both the core's
source code, as well as your own after integration.  Other cores use
different licences which impose other limitations onto commercial use.
There are a number of problems with this:

*  First off, when you do a chip, any licence requiring you to expose
the source code is a non-starter.  If you are a chip vendor, the code
is your company's commercial value, and can be worth tens or millions
of dollars.  Giving it away is foolish.

*  Secondly, supposing you wanted to release your FPGA's source code.
How would you do it?  Would you provide your customer a floppy of the
FPGA's source?  Include it in a tarball with your software images (if
you distribute software)?  And what would your customer do with it?  I
suspect that he would think you were a lunatic if you gave him several
directories of raw Verilog on a floppy.

*  Since there are several different licences in use, you have to
become a legal expert in the detailed restrictions of each type of
licence.  People on Slashdot love to argue about the relative merits
of the Mozilla licence vs. the BSD vs. GPL, vs. Netscape,
vs. etc. etc. etc.  Personally, I don't want to have to worry about
any of these issues.  

*  Finally, even if you did want to be a good doobie and release the
GPLedsource, how would you -- as a working engineer -- ever explain
that to your boss?  Telling anybody in the management line that you
want to release the source code for your chip would be professional
suicide.  


I do know that lots of designers look at the opencores page for
ideas.  They may indeed be using the cores in their work, for all I
know.  But if they do, they will never admit it because it opens up
Pandora's box of licencing issues.

Of course, all these issues also exist in the software world.  The GPL
was originally deviced to protect academic hackers from major
companies (i.e. Apple, Sun) who wanted to expropriate unix
freeware, put it in their products, and then put a commercial licence
on it to prevent third parties from hacking the source ever again.  To
this end, the GPL is a wonderful thing -- it preserved the right of
academics and hobby hackers to play with their code.  

However, the GPL makes it difficult to build a business on open source
code.  RMS's notion that you can give the software away and make money
on services is manifestly not working -- more Linux companies have
gone bankrupt than continue to survive.  And Red Hat -- the gorrilla
of the industry -- made barely $100K last year.  You can't run an
industry like that. 

Stuart



> 
> thanks a lot for your feedback ! I think all
> feedback is relevant !
> 
> Thanks !
> rudi
> 
> On Sun, 2003-05-04 at 10:53, John Dalton wrote:
> > Hi Rudi,
> > 
> > I'm not a 'commercial user' as such, but am looking
> > into using opencores to help prototype some research
> > results.  Not really relevant to your request.
> > 
> > Over the last couple of years I have had quite a
> > few discussions with people about application of
> > opencores.  There seems to be a common perception
> > that cores from opencores are of 'low quality' and
> > not 'up to scratch'.  To the extent that at least
> > one author refused to upload their work to opencores,
> > in case opencores' reputation detracts from their own.
> > This particular person cited questions about the
> > correctness of the cores, lack of testing and seemed to
> > be of the view that by the time you debugged and figured
> > out how to use an opencore (lack of documentation)
> > it was easier to do it from scratch.
> > 
> > The above is not necessarily my own opinion, just
> > what others have said to me.  Also please don't
> > take the above as a judgment on to whether the
> > reputation is justified.  I'm just the messenger
> > 
> > Regards
> > John
> > --
> > To unsubscribe from cores mailing list please visit http://www.opencores.org/mailinglists.shtml
> -- 
> rudi               
> -------------------------------------------------------
> www.asics.ws  -- Solutions for your ASIC/FPGA needs ---
> ---------------- FPGAs * Full Custom ICs * IP Cores ---
> * * * FREE IP Cores  --> http://www.asics.ws/ <-- * * *
> 
> 
> 
> --
> To unsubscribe from cores mailing list please visit http://www.opencores.org/mailinglists.shtml
> 

--
To unsubscribe from cores mailing list please visit http://www.opencores.org/mailinglists.shtml