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

[oc] Verilog vs VHDL vs Other



The V versus V debate has gone on ever since the birth of HDL and it 
will continue to a "topic of discussion" (putting it mildly) for 
years to come.  Both languages have their strengths, but both are 
also fundementally flawed.

Some evidence of this are the countless Perl scripts used to 
manipulate and manage HDL code -- practically every design house has 
a suite of custom tools for this purpose.  There are several other 
language extensions built to either aid in RTL construction or assist 
in modeling.  A short list includes:

ParaCore - http://www.dilloneng.com/paracore.html
RubyHDL - http://www.aracnet.com/~ptkwt/ruby_stuff/RHDL/index.html
MyHDL - http://jandecaluwe.com/Tools/MyHDL/Overview.html
JHDL - http://www.jhdl.org/
Lava - http://www.xilinx.com/labs/lava/
HDLmaker - http://www.polybus.com/hdlmaker/users_guide/

The main purpose of a language -- programing, hdl, or otherwise -- is 
to ease the expression of design.  If either Verilog or VHDL were 
adequate, most of these hacks would not be necessary.

It's not the fault of either.  Both languages are 20 years old.  20 
years ago, they were excellent modeling languages.  Then 15 years 
ago, Synopsys Design Compiler came along and made them excellent 
design languages.  But today, Verilog and VHDL are having a hard time 
keeping up with Moore's law.

So now, Accellera and Synopsys are cramming features in to Verilog in 
an attempt to coupe with the growing pains.  This may help in some 
respect, but IMHO, Verilog-2001 already suffered from feature-bloat.  
Verilog-2001 features are still missing from many tools, and if you 
listen to the rhedoric between Synopsys and Cadence, the later may 
not even support SystemVerilog.

A language does not have to be complex to be able to express complex 
design.  This was our motivator for Confluence.  Even though the 
syntax is 3X smaller than Verilog-1995, it's semantics (recursion, 
first class functions and environments, etc) and declarative nature 
allows design to scale very easily.

I would like to donate Confluence to the OpenCore community.  Maybe 
have the tools on the server side, so anyone could generate Verilog, 
VHDL, and C from any Confluence open core.  Any ideas?
 
Regards,
Tom

-- 
Tom Hawkins
Launchbird Design Systems, Inc.
952-200-3790
tom1@launchbird.com
http://www.launchbird.com/


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