PDP users are probably the ones who will agree with you the most. :)

Just to spin slight off-topic...
PDP is actually not a good term to use here, since there were (are) so 
many different machines named PDP, with nothing in common.
However, naming 16 bits a word is something the PDP-11 do, and which VAX 
then continued on. Hence longword (32 bits), quadword (64 bits) and 
octaword (128 bits). Most other people tend to use word to mean 32 bits, 
and usually say halfword for 16 bits.
But on most other PDP machines, a word is actually the native size of 
the machine. Hmm, thinking of it, this is actually true for the PDP-11 
as well. It's just that for a PDP-8 for instance, a word is then 12 
bits, while on a PDP-10 it's 36 bits.

Now, bytes on the other hand could lead to a very funny discussion that 
many today don't appreciate fully.

	Johnny


Chuck Swiger skrev:
> On Dec 11, 2006, at 10:15 AM, Patrick Welche wrote:
>>> Also, there are only a few 64-bit architectures that can make use of 
>>> true
>>> 64-bit addressing.  While the compiler may allow 64-scalar addressing 
>>> in the
>>> source code, somewhere in the build process, the address is usually 
>>> munged
>>> down to the size of the MMU quantum which can be 36-bit, 40-bit or 
>>> 56-bit
>>> depending on architecture.
>>
>> So going further off-topic - what is the size of a "word"? I think it is
>> the same as sizeof(void *) so I would get 64, but should I get 36/40/56 ?
> 
> The term "word" is highly overloaded, but many people consider a word to 
> be two bytes, which means it will always be 16-bits in size (at least if 
> you have 8-bit bytes; PDP users need not apply).  Other people consider 
> the size of the word to be the preferred native register size, which is 
> commonly either 32 or 64 bits in size in modern CPUs.
> 
> The bizarre sizes quoted above for MMU sizes appear due to the strange 
> and wonderful history of the x86 architecture; almost all other 
> platforms have a more sane MMU design.
> 
> ---Chuck
>