Re: Forth Day and SEA Platform news?
- From: foxchip <fox@xxxxxxxxxxxxxxxxxxx>
- Date: Wed, 21 Nov 2007 08:10:28 -0800 (PST)
On Nov 20, 10:41 pm, Ian Osgood <i...@xxxxxxxxxxxxx> wrote:
On Nov 20, 6:31 pm, Jason Damisch <jasondami...@xxxxxxxxx> wrote:
Samuel Falvo also wrote up Forth Day in his blog.
Samuel's blog is good. I could make a couple corrections
and a couple of additions. Bill's talk could have been about
many things based on the description but primarily it involved
a simple solution to the problems of navigating source code
directory trees on Windows, Mac, and Linux platforms without
using conditional compilation.
Chuck's first comment was that he liked seeing colorforth on
the screen without a windows frame around it. T18 and
colorforth okad code for SEAforth24 had been shown in earlier
presentations running hosted in Windows while Chuck prefers
to use the stand-alone version. Chuck's comment that he likes
colorforth without windows was appreciated by the audience.
Chuck has always been a big fan of self-modifying code and I
knew he would enjoy seeing how the !p+ opcode had been used to
do it in SEAforth. As a former Phd candidate in Mathematics
at Stanford Chuck has a bit more formal background in math
and logic training than most of us. He has said that he
thinks that it should be possible to prove mathematically that
self-modifying code will always be the smallest and fastest
code possible. (hardware doesn't allow it excepted of course.)
I spoke about the !p+ instruction history because it was one of
my babies coming from the F21 design. It was essentially the
most recently discovered and unpublished opcode from F21 which
came back in c18 with a a different name and a much more
valuable use in port executed code.
While some people don't like dramatic code optimization examples
I find that some other people say they are grateful to be
reminded of ways to make their programs smaller and faster and
let them be more productive in their work. One example was in
ANS Forth and was described as symptomatic of much code in c.l.f.
It served as nice example of how this testbed function had been
sped up 256000 times.
I also showed some simple SEAforth code in colorforth to control
servo motors used in robotics. I showed a first cut controlling
ten servos from one node and pausing every 2ms to process command
messages, and a second version supporting eighteen servo motors
and with pauses for command messages every 20ns. I showed how
the node could control up to 35 servos if I put more code on a
neighbor node with no pins. The example was to familiarize
people with colorforth code again on something simple and easy
to understand and to show how code evolves in a couple of hours
of work.
Chuck's 9X 3x3 board has with each 24x cluster chip 1 MW not
1 GB of DRAM for each cluster.
It was nice that Samuel could attend and he did a nice writeup
of his impressions of an SVFIG Forth Day. Thanks Sam.
I had tried to provide a skypecast of the proceedings but I had
never done it before and skype seemed to go down just before it
was suppose to start. Maybe someday someone at SVFIG will plug
in a webcam or provide a skypecast or remote webmeeting access
to presentations. Thanks for the url to Samuel's blog. It is
nice to see a writeup of a Forth Day.
Best Wishes,
Jeff Fox
.
- Follow-Ups:
- Re: Forth Day and SEA Platform news?
- From: helmwo
- Re: Forth Day and SEA Platform news?
- References:
- Forth Day and SEA Platform news?
- From: Jason Damisch
- Re: Forth Day and SEA Platform news?
- From: Ian Osgood
- Forth Day and SEA Platform news?
- Prev by Date: iTV and Asus eee
- Next by Date: Re: iTV and Asus eee
- Previous by thread: Re: Forth Day and SEA Platform news?
- Next by thread: Re: Forth Day and SEA Platform news?
- Index(es):
Relevant Pages
|
