Re: [colorforth] Musings on SEAForth

["John R. Strohm" <strohm@xxxxxxxxxxx>]

You do what someone else did.

You design a meta-architecture, that looks like you want it, such that each 
meta-instruction can be interpreted by a sequence of not more than, say, 32 
core instructions.  You stash those routines in a known place in memory.

You then dedicate one core as a memory controller, that fetches words from 
external memory on command.  You dedicate a second core as a meta-fetcher, 
that requests a word from the first core, decodes it just far enough to see 
which meta-instruction interpret routine is needed.  The meta-fetcher then 
commands the memory controller to pull in the routine, which may be executed 
by the meta-fetcher, or may be sprayed at a third core.

You then write an integer BASIC interpreter, in your meta-architecture, and 
stash that in memory somewhere else.

At some point in this process, you realize that you have reinvented vertical 
microprogramming, the Sweet16 interpreter, and Integer Basic, from the 
earliest days of the Apple II.

One is drawn to speculate that high data bandwidth, how much data you can 
move THROUGH the chip, was not a design goal, but high INSTRUCTION 
bandwidth, how many instructions you can execute per data point, was.  This 
makes the chip suitable for certain very compute-intensive applications, 
such as ray-tracing, and not so suitable for I/O-bound applications, such as 
a specialized image processing problem I worked some years ago.

----- Original Message ----- 
From: "M. Edward (Ed) Borasky" <znmeb@xxxxxxxxxxx>
To: <colorforth@xxxxxxxxxxxxxxxxxx>
Sent: Monday, January 21, 2008 9:41 AM
Subject: [colorforth] Musings on SEAForth


> 2. How one would fit a large complex program to an array of these devices. 
> While Forth is certainly a multi-level language eminently suitable for 
> designing domain-specific languages, I don't see how you could 
> *automatically* go from a high-level problem description to working 
> hardware and software, especially with the cores having a ROM component!


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