Re: Re(2): [colorforth] DOES> How is colorForth different from other

["Samuel A. Falvo II" <kc5tja@xxxxxxxx>]

On Monday 26 January 2004 01:25 pm, mark.bottomley@xxxxxxxxxx wrote:
> From what I can gather about Chuck's design, the "simplicity" is not
> easily reproducable because the design pushes the state of whatever

Wrong.  The simplicity is easily reproducable.  The precise fabrication 
of it may not be.  But who cares?  Even with synchronous logic, at the 
small scales they're using to fab chips today, you can easily get 2GHz 
performance or faster using *dreadfully* simple logic (indeed, 
synchronous logic is much simpler than asynchronous logic).

> the several people implementing their own FPGA versions are clocking
> 25-100 MHz today at much more power. The other major challenge is that

The FPGA solutions use millions of transistors.  Someone can easily come 
up with a synchronous logic ASIC core if they wanted to, and it'd be 
very competitive with the asynchronous core.

> the design is asynchronous l! ogic, which is a black art and generally
> not in the skill set of today's hardware engineers. This unique
> combination amazes me.

Googling for asynchronous logic will yeild a number of current events 
surrounding it.  VLSI CAD software for asynchronous logic is gaining in 
momentum rather quickly, as clock speeds approaching 3GHz and 4GHz makes 
even on-chip clock distribution networks substantially suspect to mutual 
interference and drastically increases power consumption.  These 
asynchronous logic tools also often use dialects of existing VHDL or 
Verilog (as far as I can tell based on context in such articles) 
languages.

That being said, asynchronous logic's features, including power savings, 
have been known for decades.  Synchronous logic won out because of the 
fact that asynchronous logic circuits have roughly twice the number of 
transistors needed to implement a given function, with even more 
interconnects.  This means it's expensive to fabricate.  Note that 
asynchronous logic experimentation is almost as old as digital 
electronics itself, dating as far back as when vacuum tubes and magnetic 
core RAM were the biggest things in computer technology.

I do eventually intend on building my own discrete component processor 
again some day.  A 32-bit, asynchronous MISC is well within my grasp, 
even with ACT-TTL technology.  I just lack the time and funding to make 
it happen.  Depending on how I optimize the processor's circuitry, I 
predict it would be competitive with a 60MHz CPU, despite using discrete 
components and wire-wrap technology.

--
Samuel A. Falvo II


---------------------------------------------------------------------
To unsubscribe, e-mail: colorforth-unsubscribe@xxxxxxxxxxxxxxxxxx
For additional commands, e-mail: colorforth-help@xxxxxxxxxxxxxxxxxx
Main web page - http://www.colorforth.com