[sdiy] Small MCU MIPS, DMIPS?

[Scott Gravenhorst]

karl dalen <dalenkarl at yahoo.se> wrote:
>>Scott Gravenhorst <music.maker at gte.net>:
>
>> small things and their low
>> cost is quite right for microwave ovens, toys, and a
>> miriad of things we don't even remember has them inside.
>> I'd  like an FPGA with 10,000,000 equivalent gates for
>> $2.00.  Since that's  not reality, I do what I
>
>See, your already measure the complexity by dollars, what
>marketers says are one thing what it really cost is another
>and the approach to cost are a third.

I'm curious to know how you would price things.  For me, if I lose money, I
don't do it anymore because it makes no sense.  That thinking is paramount in
any business.  You make some money or vaporize.

So how can one manufacture big complex devices, which require serious
investment, and sell them for what you think is a reasonable price?  I think
it's a valid question given that comment.  Try to start a business and see how
it works.

>I read recently that one of the big Mobilephone suppliers are
>saying mobilephone obsolete in 10 years, every thing will be 
>that cloud something.Most phones are subsidized so they can 
>make money on the time we are blabbering so cost are fictive
>entity, can be set to anything and mean anyting, therefore i 
>want 1Mbyte in a dsPIC for 5 usd and they can do it no problem
>its just a matter of enough amount of people saying the same
>thing just as the revolution in North Afrika. Kaddaffi for
>obvious reason says no while he sits on top of his stack of
>500million USD/euro. So the companies can if they want top.

That's a very odd comparison.  I'll just have you read what I wrote above - how
would you do this, make these devices that you want, pay your employees,
purchase raw material, pay your taxes, obey all regulations, etc. etc. and
still remain in business.  

Surely, I agree with you that I'd love to have a dsPIC with a megaword of
static data RAM in it for $5.  Ain't gonna happen because it's not economically
feasible.  Yes - it's down to money - so what?  Everything is.

>> can with what is actually available.  We hobbyists
>> have always had to deal with
>> parts that sell the best in things unrelated to what we
>> want to do.  That  hasn't changed since the early days
>>and won't in the future  I'm afraid.
>
>It has changed over and over again, we go from one simple
>solution (Z80) to more and more complex, you have just
>missed it. The answer are the FPGA that you hold in your hand. 

Yeah, well, one couldn't really do much DSP in a Z80.  Or did I miss that?

>With uppcoming speed complexity will decrease and in some
>cases it allready have, we have several forces on the market,
>just look at ATMELs current ideas, will it catch on?
>
>Do i really need a DSPIC for a microwave when it can be
>done with a 555?  

Not with a clock, LCD display, motor and microwave generation controller etc. 
But that was only ONE example.  I can't list them all, but there is a plethora
of devices that use PICs.  Perhaps not dsPIC, but they are being used in common
devices all over.  And the number purchased from Microchip (or pick any mfr at
all) makes it economically feasible.

>> Possibly, but it was bound to happen.  And with FPGA
>> technology, I can put an 8 or 16 voice poly synth into
>> a single chip which implements high quality DSP. 
>> Such a chip is less than $20.  Try making a 16 voice
>> polysynth with similar features using CEM for $20.
>
>Im not against functionality, im oposed to complexity
>when it should had been simplified. What if you had to 
>chose between current FPGA and same in DIP16? You 
>obviously take the DIP16.

But I am not the only user.  I am one guy who buys less than a handful of these
things.  The demand made by industry is what drives this by purchases in the
tens of thousands at regular intervals.

In fact, I benefit personally from those mfrs that buy so many - it makes the
cost of production low enough for all of us.  Sure, I have to then deal with
BGA, but that's life.  The demands of industry will always outweigh the demands
of me.
 
>> Right - even less than 50.  But I doubt that Xilinx
>> was thinking about _my_  application when they did that. 
>> It's about volume because otherwise we'd never be able to
>> afford these things.  However, I believe that there is a
>> Spartan-3E 500K gate device available in TQFP and that's quite
>> capable.  
>
>Well we know the basic fundings about volume market, and my 
>experience on professional level dealing with those 2500 balled
>are that very few designs actually use any vast number of balls
>even large telecom products or data com base stations, whatever.

So why do you suppose they make them if there's no reason to?

>For every 50 extra balls you add another layer of PCB
>complexity. And the cost grows seriously.Its like MIPS,
>more balls dont cut if they are just small. :) 
> 
>> Actually, not so many years.  I think about 4 now -
>> but quite honestly, I designed my first working monosynth
>> within the first year.  And I have no degree, no previous
>> DSP experience and do not consider myself a math wizard,
>> basic calculus only - and I can't even do that anymore.
>> You'd be surprised what can be done with a block diagram...
>
>I wished the dev tools just where block diagram based.

WebPACK ISE (free from Xilinx) supports design tools that use schematic view
instead of Verilog.  That's about as close to block diagram as one can get.

>> Not really sure what "should be easier".  I take
>> things this way: "It is what it is", i.e., if you want
>> something bad enough, you'll learn what you need to
>> learn and apply the required effort.  I don't fear
>> that.  I embrace it.
>
>I im not entirely sure if the embrace thing are all that god,
>if things can be done better why arent it when its obvious
>in many cases? 

Show us how.  I don't see what is obvious about that so I'd appreciate your help.

>Selling out just because you want it bad enough
>makes you a simple target for marketeers.

What?

>Whenenver you by something indicate to the company that you,
>better up say you represent a user group that want's this, that
>and this, instead of this, that, and this.

I'm afraid that user groups don't have the thrust of money.  Oh crap, I said it
again.  It's about money - so what?  What isn't?

And if you think it's that simple and easy, well, I'd ask you to please do it
so that we can all have these things for a dime each.

>> wouldn't advocate any single method for all.  Do what
>> you do best and be happy  in that.  If it's analog,
>> then make analog, but understand that there are  market
>> forces that are working against you.
>
>Ah, yeah the famous market forces who have made VST a dead
>end hair dryer 19:99 at Wallmart (Lidl for europeans) made
>in china thing of it all. Not to mention the recycling of
>hardware VA synth like Blofeld, MiniAK, Alesis, etc etc
>and are either discontinued 5 months later or price dropped.
>
>No wonder DSI are making anything except soft synths.
>http://www.kvraudio.com/interviews/interview_with_dave_smith_15841
>

I think this thread has wandered far beyond "why bother with MIPS" so I will
retire now.

-- ScottG
________________________________________________________________________
-- Scott Gravenhorst
-- FPGA MIDI Synthesizer Information: home1.gte.net/res0658s/FPGA_synth/
-- FatMan: home1.gte.net/res0658s/fatman/
-- NonFatMan: home1.gte.net/res0658s/electronics/
-- When the going gets tough, the tough use the command line.