Subject:
|
Re: Mechanical Memory for Computing
|
Newsgroups:
|
lugnet.robotics
|
Date:
|
Tue, 1 Jul 2003 01:46:01 GMT
|
Viewed:
|
903 times
|
| |
|
|
In lugnet.robotics, John Barnes wrote:
> In lugnet.robotics, Anders Isaksson wrote:
> > Depends on what you call 'recent' :-)
> >
> > The DEC PDP-8 had 12-bit words with 6-bit bytes, if I recall correctly.
> > There were also computers with 36-bit words containing six 6-bit bytes in
> > one word (DEC 10/20?), or three 12-bit bytes.
> >
> > The DataSaab D21 (that's a computer from the 1950-60 something) had 24-bit
> > words split into eight 3-bit bytes/nybbles/digits.
> >
> > In the name of internationalization, a character is now 16 bits wide, so the
> > 'byte' may still grow...
> >
> > --
> > Anders Isaksson, Sweden
>
> I just had to join in on this one :)
>
> I greatly enjoyed playing around with Intel's early 4004 PMOS microprocessor,
> one of the very first I think. It was a 4 bit per entity (character?) device
> which was primarily aimed at processing BCD, one digit at a time. After that,
> numerous other 4 bit devices appeared, mainly as the "engine" in early
> calculators. But Ah! it was a great day when Intel unveiled the 4040, a real
> step up from the earlier 4004 :) Access to data RAM could be done in less that
> 12 instructions - you could do it in 7 !!
>
> Then they invented 8 bit devices after which all the programming got almost too
> easy ;) The Nat Semi SCAMP cut the memory access function to a mere 5
> instructions.
>
> But I digress, the original thread content regarding mechanical computing is
> very interesting. I had tried this myself some while back and had serious
> problems getting the "power" in. In order to sustain a continuous computation,
> you have to find a way to feed energy into the system, especially when using
> mechanical parts with the lack of precision common in Lego elements. My attempts
> to couple a "clock" (shaft input from a motor attempting to complete things like
> memory storage strokes), I had too much difficulty in getting different parts of
> the assembly to work in accurate enough phase.
>
> Maybe someone with more time and/or imagination might be able to come up with a
> "clocked" ALU to memory/memory to ALU mechanical state transfer scheme.
Hi John,
I'd forgotten about the 4040. I thought they went from the 4004 to the 8008
to the 8080, but my memory must be failing me. I really only became aware of
microprocessors with the 8080.
I invented a clocked pneumatic scheme where you make up your design using
pneumatic master slave flip flops. By using dual source pressure lines (one for
the masters, and the other for the slaves) that are mutually exclusively
pressurized, you can create synchronous pneumatic designs.
Kevin
>
> JB
>
> (showing his age with knowledge of 4 bit micros)
|
|
Message is in Reply To:
 | | Re: Mechanical Memory for Computing
|
| (...) I just had to join in on this one :) I greatly enjoyed playing around with Intel's early 4004 PMOS microprocessor, one of the very first I think. It was a 4 bit per entity (character?) device which was primarily aimed at processing BCD, one (...) (21 years ago, 1-Jul-03, to lugnet.robotics)
|
5 Messages in This Thread:
  
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
