Byte sizes (was Re: 2.8M 3.5' floppy (was: three and a quarter loppy?)

Paul Koning pkoning at
Mon Mar 14 17:19:39 CST 2005

>>>>> "Bjørn" == Bjørn  <bv at> writes:

 Bjørn> On Mon, 14 Mar 2005 13:26:23 +0100, Johnny Billquist
 Bjørn> <bqt at Update.UU.SE> wrote:

 >> On Mon, 14 Mar 2005 Bj?rn <bv at> wrote:

 >>> A byte is the smallest INDIVIDUALLY addressable unit of data on a
 >>> system.
 >> Nonsense! Where did you get that?

 Bjørn> From Donald Knuth.

That's nice.  But while Knuth is an authority on many things, he's not
necessarily an authority on that.

By that definition a PDP-8 byte would be 12 bits, which is clearly not
how the term is used there.

(Then again, on a CDC 6000 the term "byte" does mean 12 bits, it's the
PPU's addressable unit.  Hm.)

 >> Bah. A sixbit character can very well be a byte, it's just a
 >> question of if you choose to call it that.  Byte addressable is
 >> not usable as a definition of a byte.

 Bjørn> It does not matter if you do not find it usable, as long as
 Bjørn> it IS the traditional definition.  The PDP 10 came long after
 Bjørn> it had been established.

Huh?  Knuth's "Fundamental Algorithms" came out well after the PDP-10
(never mind the earlier PDP-6, which is the same architecture).

 Bjørn> The most interesting byte is the one Knuth used in his
 Bjørn> abstract MIX machine, which he came up with when he was here
 Bjørn> in Oslo. To drive home the point that bytes were not
 Bjørn> dependent upon any particular hardware implementation, he
 Bjørn> defined it as able to hold unspecified information, but at
 Bjørn> least 100 different values. I would have thought most people
 Bjørn> here had read his "Fundamental Algorithms"?

Yes.  That's a really cute definition, but it certainly has not been
applied anywhere.


More information about the cctalk mailing list