>
> John Doty wrote:
>

>>It's pretty clear from the contents of the development kits
>>that the dominant language for small widgets must be C.

>
> Your viewpoint is skewed. Those development kits are for
> *medium*- sized low volume widgets. When the quantities
> become high (some of my designs have reached quantities of
> 100,000 per hour) the usual choices (PIC, 8051, ARM...) cost
> too much. At those volumes you are more likely to see
> microcontrollers made by by GeneralPlus/SunPlus, Elan/EMC,
> WinBond, Sonix, etc, and nearly 100%% of the software is
> written in highly optimized assembly language with
> Forth making some inroads. This is an entire world that is
> invisible to you unless you are a designer of talking barbie
> dolls, computer mice, or musical greeting cards.
>
> References: ...

>Guy Macon wrote:

>>
>> John Doty wrote:
>>

>>>It's pretty clear from the contents of the development kits
>>>that the dominant language for small widgets must be C.

>>
>> Your viewpoint is skewed. Those development kits are for
>> *medium*- sized low volume widgets. When the quantities
>> become high (some of my designs have reached quantities of
>> 100,000 per hour) the usual choices (PIC, 8051, ARM...) cost
>> too much. At those volumes you are more likely to see
>> microcontrollers made by by GeneralPlus/SunPlus, Elan/EMC,
>> WinBond, Sonix, etc, and nearly 100%% of the software is
>> written in highly optimized assembly language with
>> Forth making some inroads. This is an entire world that is
>> invisible to you unless you are a designer of talking barbie
>> dolls, computer mice, or musical greeting cards. ...

>By the way John Doty is involved in the design of scientific
>instruments to be sent into outer space and these need a similar
>style of programming as inexpensive toys even tho they are made
>in quantities of three or four.

>Can't a Forth development system do whatever optimization
>is needed for machine code faster and better than ancient
>assembly language?

> m-coughlin wrote:
>

>> By the way John Doty is involved in the design of scientific
>> instruments to be sent into outer space and these need a similar
>> style of programming as inexpensive toys even tho they are made
>> in quantities of three or four.

>
> You are claiming that a system with multiple transputers
> and DSP chips, each with 20MB or RAM, running a hand-
> crafted RTOS and programmed in C needs a similar style
> of programming as a 4-bit microcontroller with 64 Nybbles
> of RAM, No OS, no stack, no interupts and hand-crafted
> assembly language.

>
>Guy Macon wrote:
>

>> m-coughlin wrote:
>>

>>> By the way John Doty is involved in the design of scientific
>>> instruments to be sent into outer space and these need a similar
>>> style of programming as inexpensive toys even tho they are made
>>> in quantities of three or four.

>>
>> You are claiming that a system with multiple transputers
>> and DSP chips, each with 20MB or RAM, running a hand-
>> crafted RTOS and programmed in C needs a similar style
>> of programming as a 4-bit microcontroller with 64 Nybbles
>> of RAM, No OS, no stack, no interupts and hand-crafted
>> assembly language.

>
>Different parts of the code needed different styles. The SXC processing
>required a carefully hand-crafted assembly language inner loop of half a ...

> John Doty wrote:

>> Guy Macon wrote:
>>

>>> m-coughlin wrote:
>>>

>>>> By the way John Doty is involved in the design of scientific
>>>> instruments to be sent into outer space and these need a similar
>>>> style of programming as inexpensive toys even tho they are made
>>>> in quantities of three or four.

>>> You are claiming that a system with multiple transputers
>>> and DSP chips, each with 20MB or RAM, running a hand-
>>> crafted RTOS and programmed in C needs a similar style
>>> of programming as a 4-bit microcontroller with 64 Nybbles
>>> of RAM, No OS, no stack, no interupts and hand-crafted
>>> assembly language.

>> Different parts of the code needed different styles. The SXC processing
>> required a carefully hand-crafted assembly language inner loop of half a
>> dozen instructions or so. I spent quite a bit of time eliminating a
>> single cycle of pipeline stall, making the difference between code that ...

> because that was his name.
> I first met Mel when I went to work for Royal McBee Computer Corp.,
> a now-defunct subsidiary of the typewriter company.
> The firm manufactured the LGP-30,
> a small, cheap (by the standards of the day)
> drum-memory computer,
> and had just started to manufacture
> the RPC-4000, a much-improved,
> bigger, better, faster -- drum-memory computer.
> Cores cost too much,

> Wow, LGP-30 http://tinyurl.com/2w8ffv that's a blast from the past. I
> remember being mesmerized by one on display at the Regina Exhibition when I
> was in high school (ca 1958). They had it doing highway earthwork
> calculations and the person domonstrating it boasted that it could do the
> calculations faster (80 MPH) than you were allowed to drive on the roads.

> 80 multiplications per hour?

> If we want an inroad in the gcc tool chain (see other threads)
> assembler may be our best choice.
> If we managed to replace the back end of gcc by a Forth
> assembler the whole tool chain might become simpler, as Forth
> may come to the aid of register assignment etc.
> There could be a persuasive argument, execution speed!