Heh. Yeah, I'm a hardware geek. :)
Here's another one...a little (not much) less technical, and maybe an urban legend,
but perhaps a bit more interesting from the human-technology interface perspective...
Are Religion/Mythology responsible for "modern" word sizes?
The story:
So MIT was supposed to provide the onboard computers for NASA's Gemini program.
But some authority figure in NASA balked because the plan was for a 13-bit data
word. "Too unlucky!" Back to the drawing board. The irony is that the final Gemini
computers, from IBM, used a 13-bit instruction and a 26-bit data word. Heh. A
trilogy of 13's!!!
Rumor has it that the original MIT design was scaled back to 12-bits to make
the superstitious people happy, but it took too long. The design eventually
became the PDP-8 minicomputer.
Having learned from its mistakes, MIT designed the machine used in Apollo to have
a 16-bit word instead.
Whee!
-Chris
On Thu, 14 Apr 2005, Ruffin Bailey wrote:
> On Apr 13, 2005, at 11:57 PM, Chris Wilkson wrote:
>
> > On Wed, 13 Apr 2005, Ruffin Bailey wrote:
> >
> >> Why would the number of bits in a byte have to have anything to do
> >> with
> >> powers of 2? The max number, whether 7 bit, 13 bit, or 8 bit would
> >> still be a binary number. ??
> >
> > Because decoders and encoders using bits as control lines will always
> > address 2^n locations. Either you implement them, or you don't. But
> > if you need 5 states, you need 3 bits to represent that. Then you have
> > created 3 bogus states. You need to define them or else "Bad Things"
> > can happen. You can define them by hardwiring, but you might as well
> > make them useful if you can.
> >
> > In the case of decimals digits, you need 4 bits to encode each
> > character
> > with. If your system is already 6 bits, just to encode the text
> > characters,
> > you're wasting 2 bits when doing BCD. You might as well make your word
> > 8 bits wide so you can encode 2 digits without waste.
>
> ... thus earning the mit.edu behind his email nick. ;^)
>
> Ruffin Bailey
>
> Archives (includes files) at http://www.biglist.com/lists/stella/archives/
> Unsub & more at http://stella.biglist.com
>
>
Archives (includes files) at http://www.biglist.com/lists/stella/archives/
Unsub & more at http://stella.biglist.com
