Re: MISC-d Digest V98 #48

["GARY B. LAWRENCE" <garyl@xxxxxxxxxxx>]

MISC-d-request@pisa.rockefeller.edu wrote:
> 
> Subject:
> 
> MISC-d Digest                           Volume 98 : Issue 48
> 
> Today's Topics:
>          News
> 
>     ---------------------------------------------------------------
> 
> Subject: News
> Date: Sat, 7 Nov 1998 16:16:07 -0400
> From: jfox@dnai.com (Jeff Fox)
> To: MISC
> 
> Dear MISC readers,
> 
> The list has been pretty quiet but a lot of things have been
> going on.  I don't think I have posted anything here since September.
> I had shipped a few copies of the videos of Chuck and some copies
> of the new simulator and emulator and demos from the Ultra Technology
> online store.  I thought some people might make some comments on the
> simulator or emulator or the demos that Stas and I did for them.

 I want to wish you luck and hope you get a working chip. I also wanted
to mention that after buying the F21 EMULATOR I had trouble at first
running it under WIN-95, it would crash until I checked my memory setup
and found that it was not set for EMM. The documentation did mention
using EMM memory, but you might want to put in a early message to have
EMM memory setup. Other than that I had no trouble running the demos and
even added a few lines of code to one to see if it changed the program.
  It was impressive to see the display on the dos screen and it will
help me understand code sequences. One that at first did not make since 
to me was the one DUP DUP -OR COM . It takes four forth words and does
not need an out of page memory address and produces -1 for decrement 
of the count that was dupped. I will need some time to understand the
tradeoffs between off page adressing and multiple program words that 
produce the same result faster. Old 8 bit processors like the 6502 did
not have fast and slow memory cycles in ram because the processor did
not outrun the memory.  

 I wanted to add that after sending some email to Jeff about software
I thought that would be needed, he responded that the P21 FORTH had most
 of the things I had mentioned and could be setup for the F21 with out
too many changes. I felt very foolish because I have a MUP21 with the
P21 FORTH ! I have spent the last week or so looking over the manual and
running a few simple words and it has most of the code I asked him 
about. Things like basic sreen editor words, multitasking , single and 
double math, dumps, basic assembler and block graphic words. 
   I would like to set up an F21 with a one hand keyboard and a small
flash disk and it seems that the 500 or so words in the P21 FORTH would 
 make it possible to start with mutimedia smart card commands very
quickly. I have not yet found the San Disk mutimedia cards in Denver 
 but the smart flash cards are in the range of 15 megabytes for $100.
I may have to compress some of my database ( 40-50 megabytes ) but I
know that the program scource can fit in about 300 - 600 screens. The 
multimedia cards were written about in the october issue of CIRCUIT
CELLAR. 

 For now I will try some work on the MUP21 and hope to buy an F21 board
early next year. I have been trying to load source code from F-PC to my
MUP21 board by redirecting the output of the fload word to the rs232
port. I slowed each byte transmitted so that the MUP21 could display and
load each character as it came. I then put in a carriage return word
that is detected by the redirected fload word. When it is detected fload
 sends a carriage return character and then pauses about 600
milliseconds to allow the MUP21 FORTH time to load the code. I noticed
that I needed to delay because the command given was a fill of a large
area for graphics and the MUP21 missed the next command. I think that
 I need to load from a file using xon/xoff protocol or to load the file
straight to memory in the MUP21 and then have another word in the MUP21
 load the scource similar to the way block screens are loaded. 
   If I am missing some words already there that do the same thing for
me let me know. I also have trouble getting multiple files to load from
one F-PC file. I would like to do that since I have used that method in
both F-PC and WIN32FOR . My code sends multiple files out the serial
line, but it also sends the names of these files out. I am using fload
and my redirected fsend at different levels and I am not getting
something redirected right. I will work on this since loading from files
 or screens allows for good structuring of code.

> With no video being generated and running sequential
> stack instructions in fast SRAM F21 can get a max of 222mips.  When
> running in DRAM the max without video is 111.  With video being generated
> and the CPU in DRAM the mips rate can drop to 20mips or lower.  This
> means that the emulator running on a very fast PC will now execute at
> close to the speed of a real F21 in DRAM and giving most of its bandwidth
> to coprocessors.  This makes it more useful for debugging complex and
> detailed programs.  It also makes it more suitable as a virtual machine
> implementation for the five bit virtual machine Forth that we execute
> in hardware on the MISC machines.  I have experimented a little with
> using it as a helper in Netscape this way.
  I wanted to know if operations running in static ram were affected
 much the other processors. A difference between 20 mips and 222 mips
would be enough to design inner loops to run in static ram if the video
processor was being used. In my case I just want to take a database that
 is growing to big for 640k F-PC and try the F21. My program is running 
 on a D2 66 486 so if it only runs at 20 mips I might be able to stand
it ;) I started to put it in WIN32FOR and found out I needed to learn a
lot about WIN95 api's just to get started. I like the idea of having one
sheet with the wiring diagram, one small book for assembly code, and the
P21 FORTH book. I have lots of old books on dos, pc assembly code, and 
 hardware and had to get 6-7 books on WIN95 just to start to understand
 WIN32FOR and WIN95. IF some number of similar applications can be done
with the F21 quicker than on a more complex platform, then others may
beging to understand it's utility. 
    I think that Chuck may be right when he sees the future as having
custom processorsand custom code for many applications. I see the
complexity of the wintel system making it hard to change to meet the new
ideas. 

> I also understand that the Ultra Technology and DYOP web site URLs
> are going to be listed in an upcoming article by Don Lancaster in
> the Jan 99 issue of Electronics Now.  Unfortunately Don was given
> www.dnai.com/~jfox   instead of www.UltraTechnology.com
> I really wish people would use the registered domain name because
> the site could always move from the original address and it is
> easier for people to remember UltraTechnology.com than dnai.com/~jfox
> which ultimately may have nothing to do with Ultra Technology, who knows.
> 
> Oh well.  So much for the news report.  Wish me luck with getting
> the chips back and packaged and paid for and tested and working and
> to do the right demos to impress the right people.  Is that too much
> to ask for?
> 
> Jeff Fox
> Ultra Technology
 Let me again wish Jeff luck in getting a working F21D chip and let me 
 recommend the videos that show Chuck discussing his chips, they help
 to explain how he thinks and give some explanation for why he strives
 for a simpler hardware and software solution.
	Gary Lawrence