News

[jfox@xxxxxxxxxx (Jeff Fox)]

Dear MISC readers,

Penio asked:

>I am trying to build a new board around MuP21, and I have some problems.
>
>Does anybody have a list of the electrical differences between the DIP and
>PLCC packaged MuP21s?
>
>For that matter, are there data sheets for the two version of MuP21?  If
>not, what is the closest to a datasheet I can get? 

Except for pinout there should be no electrical differences.  The first
PLCC chips were simply run on boards designed for the DIP by using
a socket adapter. 

Internally the PLCC is the same as the plastic DIP in that nop
should be used before A! unless it is in slot 0.  This is not a 
restriction on the ceramic dips and it is thought to be due to 
increased impedence due to plastic being in contact with lead wires.

I published an ascii version of the information on the spec sheets
I got to this MISC mail list a long time ago.  It had the timing
signals etc.  Offete Enterprises Inc. is the source of datasheets and
the chips also.

****************************

Eugene Leitl asked:

>How about a short report on the TV set top box progress? 
>
>(I am not asking about i21/F21 availability, however tentative). 

Sure, I have been meaning to do this for some time.  iTV is still
making progress we have had no major breakthroughs to anounce
since I published Chuck's fireside chat about iTV and the technical
problems with i21.

Just to give people an idea how exciting this all is I should tell
people how the day after Chuck's fireside chat to FIG the staff at
iTV was told that they should be looking for another job.  As had
happened before our investors had failed to deliver on financing
and there was no money to pay bills or salary.  This was the same
situation we had been in before, but this time there was no reserve
and we were told we could not expect to get any more paychecks.  The
reason this time had been that none of the investors would actually
deliver money until all did, and one had decided to wait again.  This
one had decided that we were not far enough along in product (hardware
or software) development and felt they might consider investing again
in a few months.
We had all just finished a set of 90 hour weeks to provide demos for
them before they made that decision, but we decided that perhaps if
we could somehow push our schedule forward a month or two in the next
week we might be able to pull them in.  So rather than spend time 
looking for work we stepped up our efforts dispite the fact that we
were not expecting to get paid for our 100+ hour work weeks.  We in
fact did push our schedule forward and did impress the right people
enough for them to once again sign an agreement to fund our company.
The situation has not really changed, except we have been able to
get some bridge financing to keep going.  We are still told every
week that this is the week that we are suppose to close on financing
and that the investors decided again for various reasons to delay it.
We still hear some complaints that we are not as far along as our
schedule called for and that this is a reason to delay investment.
Of course since we have gone for over a year and a half waiting
for our funding to come in next week this has made it somewhat
more difficult to keep up with a schedule based on being funded.  
Our schedule called for employees, equipment, and last but not least
fabrication runs to debug our technology.  Needless to say when you
can't hire people, or buy equipment, or pay everyone's salaries, or
have prototype chip fab runs made forward progress is more difficult.

The chip Chuck was talking about at his fireside chat was i21i.  He
also talked about i21j and i21k that he was about to submit.  i21j has
been fabrictacated but is not yet here at iTV.  Our investors would
like to see this chip come back and be working correctly, and we
would like to pay for the fab run, get the chip back, and test it,
and we will...
 
As for Ultra Technology Inc.:
Based on i21i being mostly functional and confidence that i21j would
be even better based on the things Chuck talked about in his
fireside chat to FIG I asked him to prepare F21c for fabrication
back in January.  He said he didn't think he could make the Feb
sumission dates at Mosis and we didn't.  We are trying to get an
F21 back into fab at the next opportunity.

*********************************

Penio asked:

>Not to mention, that the F21 die is 3/4 empty and I bet that Chuck is
>working on putting memory there, which will raise the throughput at least
>twice -- to the true 400 MIPs.  Jeff, is that right? 

Yes.  On chip RAM and/or ROM will reduce or remove the memory
bottleneck.  We are currently at about 500mips internally but SRAM will
limit you to about 200 and DRAM closer to 100 at least with external
memory and the current memory setup time on these chips.  And of course
the other limiting factor is .8 micron technolgy.  Chuck has said that
he gets about a doubling of speed per 20% reduction in component size
and .8 is now old technology.  If we ever get technology sharing
contracts with some of the big investors we will see MISC chips with
lots of on chip memory and running at much higher speeds.

As for price volume still remains the main issue.  At fab run quantity
(without factoring in any development costs) just the fab run costs
come to about $300 per chip.  There is a curve and as the quantity
manufactured goes up the price per unit goes down.  If I get 10k
made the manufacture cost drops to $15.  This means if that many are
made the price will be low compared to $100 chips, but not compared to
$2 chips.  If millions are manufactured then the price starts to look
nice.  In very large quantity they _could_ be in the $1-$2 price range,
but don't hold out for it.

As I have said from the beginning the main advantage of this technology
is the potential for low cost.  There are lots of expensive, big, high
end chips that have more instructions, bigger busses, faster memory
interfaces, and more througput.  There are lots of chips that are
smaller and cheaper.  However there are none that are as fast and
cheap and integrated.  In fact we are orders of magnitude ahead on
some real world measures like mips/$/watt.

*********************************

M. Edward Borasky states:

>You forgot to mention one *major* difference: Texas Instruments is a =
>major vendor who will ship large quantities of C6x chips on a credible =
>schedule.  As far as I'm concerned, the F21 is still vaporware; the few =

Quite right.  It is not yet a product and you cannot get shipment of
large quantities like you can with chips that are products.  

F21 chips are not vapor, they do exist, however they have bugs.  And
a lot of work and money has been invested in development but the product
is not here yet so it is fair to call it vaporware.

>resources that were working on it were reallocated to the I21 because =
>that's where the financing was.  And the I21 is available only packaged =
>in a TV / Internet interface that is not user programmable.  Even to a =
>hobbyist such as myself, such things matter.  For my application, =

I21 isn't available at all right now.  It is not clear yet if it will
be available as a chip, but I don't expect iTV to sell small quantities
of chips or develpment systems.  As for the iTV box not being user
programable it is true that the initial program distributed on
flash will not include Forth script, but we do expect to add this
feature in a later release.

>computer music, were I to start building today, I would be forced to =
>choose between a P21 and some homebrew D/A hardware or a DSP chip that =
>might be slower, and might not have Forth, but that probably would have =
>the D/A hardware bundled on an evaluation board.  Or I could spend a =
>*lot* of money and get a refurbished 486/DX4 and a conventional sound =
>card.  I don't want to spend the money for the 486/sound card, and I =
>don't want to build *any* hardware other than soldering phono jacks on =
>audio cables.  So where does that leave me?  With a DSP board controlled =
>by a serial line from my HP100LX; the control code would be in Forth but =
>the DSP code would probably be in assembler.

Yeah thats about it.  I expect you can several good Forth implementations
for various DSP chips that do exist at this time.

Jeff Fox

P.S.  MISC readers:  I am phasing out this jfox@netcom.com account.  Please
use:

jfox@dnai.com              Ultra Technology Inc.
http://www.dnai.com/~jfox

or 

jeff@itvcorp.com           the iTV Corporation.