Chuck's SVFig talk part 6, questions and answers

[theFox@xxxxxxxxxxxxxxxxxxx (Jeff Fox)]

Dear MISC readers:

This is section 6 and the final section of Chuck's
presentation to SVFig on 5/22/99.

The full HTML version is at http://www.UltraTechnology.com/cm52299.htm
and is complete except for the addition of some graphics and final
edits.  For section 6 this ascii version is fine.

Question about iTV's status and funding.

(something about deals are in progress.) Negotiations are delicately
proceeding.
One consequence is that we have not had the funding to make any chips
for the last four of five months so there have not been any chips made.

I have done some work. I have added features.
I came up with a whole new chip that is the best of all chips.
But at the moment I don't know if that chip is relevant. I don't know
what other chips might be relevant.  And I am sure that someone is going
to
come to me someday and say "This is what we want." And my reaction will
be,
"But this is different than anything you've ever said."

I am kind of waiting.  I am in the document mode.  I have got some very
pretty HTML screens gates in it. Showing off the various kinds of gates
the system uses. There is a bug in it. I can display this fine with I.E.
but
my color inkjet printer won't print it with the right colors.  I showed
you the color table there.  Those are the colors and I know they are
right because I displayed them on the screen. But the printer won't
print them that way. I don't know who is responsible for putting out the
colors. Is it the printer driver or Windows? But someone is dropping
the ball.

I will show you that sometime. It will be on a website sometime.
It is very pretty. And it is much easier to understand silicon
design when you can see it rather than when you have to read
the netlist.

We are still at our old site, the second site, in Redwood City.

Dr. Ting ask questions about a USMC foundry and the shuttle
program and how they are looking for customers.

That sounds to me very attractive.  I mentioned it to Gary Langford
and he says that USMC has a very active legal department defending
themselves against intellectual property claims.  He doesn't want to
go there. He is afraid of losing whatever proprietaryness we have
in our design.

I am more sanguine about this is an invulnerable design. You can't
replicate
it without the design tools. But he doesn't want to bet the company on
that.
If that turns out to be the only alternative it is the way we will go.

We are talking to some Israelis.  The Israelis are not known for
honoring intellectual property either so I don't understand
the distinction that he is making.  These kind of choices
tend not to be economic but political.  If we are dealing with a
company that is dealing with a company that knows somebody in Israel
then that is where we will go. We have few choices. I've never had
a good choice.  First there was ORBIT. They were the only one I knew
about
and they were cheap. Then there was Mosis with HP.  And now Mosis has
three or four others that we could pick.  But it is a 5 to 10 thousand
dollar
shot and with that kind of money you can pay somebody for a month for
what
it costs to do a fab. That is
what we are doing.

John Rible: Have you had any time to make OKAD work in a hierarchical
fashion?

Chuck: No.

Q: What do you mean by hierarchical?

Chuck: The chip layout is flat. All the latches are laid out in the
registers and t here is no way I change the layout of a latch and have
it propagate everywhere there are latches. But I am not strongly
motivated to do that. It might have been a better way to start.
But it is very important that I have the flat layout in order for
the simulator to simulate.  So I have to flatten everything anyway.

Given that to make it hierarchical is more work not less.
There are a number of changes I want to make to OKAD but it is not clear
that there will ever be a convenient time to do that.
You start down a path past the branchpoints and where do you end?
More importantly, I would like to get the a ..
John has suggested changing the tiles from the 16 different patterns of
corners and crosses and things to about 4 different patterns of
lines and corners and that is higher on my list than a hierarchical
layout.

Dr. Ting asks a question about million gate FPGA availability and doing
chips
in FPGA. "At least you have something that runs at 1/10 the design
speed."

John Rible answers, "You may as well use the software simulator because
the only thing that that will give you in this environment is the
logical
correctness. It won't give you any other information."

Chuck says The timing would be completely different and some of the
tradeoffs would be different. If I were just starting out I think
that would be a very interesting path to follow, things get complicated.
But given when I am now I don't need any practice.  I've got a simulator
that I'm convinced now tells me true. And if nobody wants to make
production quantities of these chips I've got other things to do. I'm
not going to beat a dead horse.

My goal at the moment is not to design another chip, it is to sell
some chips. Selling some chips means your doing a million, if you
are not doing a million it is not worth doing. The economics of
the industry have changed in the last ten years.  More and more
reliance on large production runs. And you get great concessions
from the manufactures if they anticipate a large production run.
And no one is interested in 10,000 parts, it is not worth anyone's
trouble; designing or manufacturing. If you want 100 parts then yes,
FPGA,
but there is still a big gap between 100 parts and 10,000 parts
that is completely empty.

Q: What are you going to do with Color Forth?
Chuck: I suspect that I will publish it on a website somewhere as soon
as I get it where I like it. It will be very small. 1K, 2K something
like that.  It will boot from a floppy or CD-ROM, if you give me your
whole hard disk I can boot from hard disk. I doubt if I can live with a
partition. Which is why I say this is whole nother opportunity for old
computers to come back to life.

Chuck: Good question. What would I do if I didn't have to work for a
living?
I would do exactly what I am doing now. I am very happy with this.  I
wish
we could get into production. If we get into production there is a
disturbing
consequence. If we get a chip into production it will be the chip we
used
last year.  It will not be my best chip because it has no history. How
can
you prove the value of a new chip when nobody will let you make one
because
they are busy making the old chip that you know doesn't work?  And with
large production, with the pressures of production, I think that a huge
amount of innovation will get lost. This is the conundrum, but even so
I'm tired of innovating.  I want to see some production.

Color Forth looks to me to be a lot of fun. I will try to re-address
these
issues of my wild irresponsible claims as to how small software can be.
But I will repeat them lest anyone misunderstand.

I do not think it is intrinsic in the nature of software that it has
to be large, bulky, buggy. It is in the nature of our
society. We now have millions of programmers who are relying on
their paychecks and they have to keep busy. Y2K is a boon for the
software industry. You can drag old retired programmers back and
put them to work. There is so much economic motivation for
producing this ... bloatware, that I kind of feel irresponsible in
standing up and saying the emperor has no clothes.

But at least in my small way I will accumulate a library of examples
to prove the point to anyone's reasonable satisfaction.
I don't expect it will sweep the industry.

There is one way that it could. That is if a small group of people
got together and decided to challenge Microsoft.  We could replicate
their software in 1/10 of 1% of the code.  We could feasibly do that
with a small team in a year or two.  But again why destroy an industry?
That is probably not a wise thing to do.

J. Thomas: I don't think you could do that. To do it effectively
you would have to replicate their bugs.

Chuck: Oh no, I'm not running applications.  I'm replacing the
whole shmear.

Jet: Oh, you could that.

Chuck: That's much more practical.

Dr. Ting: The Chinese are doing that.  They are very much worried about
Microsoft controlling every aspect of the industry so they are trying
to develop their own OS.

Chuck: You said the evil word. If they are starting from the OS they
have made the first mistake. The OS isn't going to fit on a floppy
disk and boot in ten seconds.

Actually the greatest problem to having a Forth come up nicely with
a picture on the screen before you blink is the damn BIOS.  It powers
up in BIOS it doesn't power up on the floppy and it takes you ten
seconds to get to the floppy.  So you have this delay. But
I'll give you the PROMs that you can plug into the sockets that will
do that. None of this is hard.

There was a book, it was translated from the Danish, written about 1989.
I forget the author but the name was, The User Illusion. This should
ring a bell for everyone. The User Illusion is what Apple exploited
to make the computer look different than it really is.  To make it
friendlier, easier to use.  So the user has the illusion that he
is dealing with the software when really he is dealing with the
hardware.
He viewed this as a good thing.  Among many of his good comments this
was a misunderstanding I think.

I would like to dispel the user illusion and make the interface
transparent and attached to the hardware.

I repeat the claim that I made long ago that it is easier to write a
floppy disk driver in machine language than it is to interface to the
BIOS routines. And this is true of every device.

All the devices together, now that is a different problem. But you are
not dealing with all the devices together. I am not going to give you
every printer driver.  I am going to give you the HP inkjet printer
driver and if you want a different one you will have the knowledge and
tools to build one.

Q: Is Color Forth going to have the same colors as the iMac? (laughter)

Chuck: I have been thinking. comp.lang.forth has been talking about
a Forth logo.  If you follow that, it kind of got voted down, an anime
Jedi warrior girl, scantily clad. But that got me thinking.

What we need is a uniform! (laughter) How much more impressive it would
be
for me to stand up here looking at you if you all had your cadet
uniforms on.
I think it is pretty clear that color should be that butternut color
that monitors are made of so you could blend in with your computer
system. You could have white pipeing on the ANS Forth programmers
and red and green pipeing on the Color Forth programmers
so that you could tell at a glance as to the essential information
about your audience and their level of sophistication.
And these would be compatible with Color Forth.

John Rible: FIG Forth programmers could have silver and gray pipeing.
(laughter)

Chuck: I wore this tee shirt to get in the uniform mood.  But I noticed
dearth of tee shirts. Is anyone selling Forth tee shirts?

George Perry: You are the second one to ask.

Chuck: Are we missing a whole marketing opportunity? Let's have a
logo contest and everyone will have to produce a prototype tee
shirt and wear it next time.  If we have a Forth logo then we
can have a Forth flag. There is a whole world out there anxious
to buy stuff. (laughter)

I do have several talks full of questions backed up that I will not
have time to do to here. I did have another thought. The advantages
of Forth in a team programming environment. We have spoken the fact
that a small group of Forth programmers can do an application that
would otherwise take many more and that there are efficiencies
of inverse scale at work here. So it a good team approach.

There is another advantage that I  haven't heard mentioned, they
can do it faster. Time is often a concern in getting out applications.
And if you want to minimize the time it takes to do an application
move in a team of Forth experts, hopefully they have Forth on
your platform already, blitz the problem working twelve hours a day
for two weeks and go off with it finished.  This is a custom programming
model that Forth Inc. has used very effectively and could be applied
in a much wider range.

When you are working for a company doing an application it can become
a career. And heaven forbid that you finish the application your hanging
on the end of loose limb. And that is again a cultural motivation. It
may not be the way it ought to work but that is the way the world
does work. Moving in a team of a Forth commandos. (question) It depends
on your rank. (laughter)

Bob Barr ask a question about efficiency. The natural order isn't so
efficient. How efficient is an apple tree or society?

Chuck: Yes, I agree that economic motives are important but
they are not the only ones. There are issues of personal
satisfaction which ought to be at least as important as the economic
value of what you are doing.

This code that I showed you. I worked on it two or three times as
long as I ought to have in order to get it right.
I am never going to execute it often enough to get a pay back
but it feels good. And that is important especially in this
increasingly corporate oriented world where people go to work
to earn a living instead of seeking the job satisfaction.
I don't know that we should be satisfied with a world full of mediocre
programmers. Increasingly our culture is depending on computers in an
important way as Y2K is pointing out, perhaps there is room for some
professionals to do it right. I don't know how to sell that.

Question about Y2K.

Chuck: I think the answer to the Y2K problem is not the patches that
they
are doing now.  I gather that what they are doing is windowing it so you
will have another problem in the future. The answer is to make the
specialty embedded microprocessor self-documenting. So that you can
walk up to one and ask it what is your take on this particular problem.
And the way you do that is with source code instead of binary.
And the way you do that is that you code it in Forth and compile it
at runtime. And then you know that the program that you are running
is the program that the source described and that it works that
the way you expect. Especially if you are going to look at that
program fifty years from now for the first time and try to figure
out what it does.

I think if the Y2K comes out badly there will be regulations about
what you can do.  (groans) We will just have to get political
influence to mandate Forth.

John Carpenter asks a question.
Chuck: Unfortunately this margin is too small.

Question about ADA being demandated.
Chuck: I hear that ever since they demandated ADA that it is
more popular than ever. But the pressure will be on compactness because
you don't want to store any more in ROM than you have to.

George Perry calls for a hand.
Dr. Ting calls for plans for coffee at Fry's or dinner.
(end of talk)