Re: Fat references (Anton Ertl)
Mon, 04 Jan 2010 19:57:25 GMT

          From comp.compilers

Related articles
[18 earlier articles]
Re: Fat references (2010-01-03)
Re: Fat references (glen herrmannsfeldt) (2010-01-03)
Re: Fat references (Jon Harrop) (2010-01-04)
Re: Fat references (Kaz Kylheku) (2010-01-04)
Re: Fat references (BGB / cr88192) (2010-01-03)
Re: Fat references (Robert A Duff) (2010-01-04)
Re: Fat references (2010-01-04)
Re: Fat references (Kaz Kylheku) (2010-01-04)
Re: Fat references (Jon Harrop) (2010-01-05)
Re: Fat references (Dennis Ritchie) (2010-01-05)
Re: Fat references (Kaz Kylheku) (2010-01-05)
Re: Fat references (BGB / cr88192) (2010-01-05)
Re: Fat references (Jon Harrop) (2010-01-06)
[4 later articles]
| List of all articles for this month |

From: (Anton Ertl)
Newsgroups: comp.compilers,comp.arch
Date: Mon, 04 Jan 2010 19:57:25 GMT
Organization: Institut fuer Computersprachen, Technische Universitaet Wien
References: 09-12-045 09-12-055 10-01-003 10-01-008 10-01-009 10-01-016 10-01-019
Keywords: architecture, history
Posted-Date: 05 Jan 2010 13:33:48 EST

glen herrmannsfeldt <> writes:
>In comp.compilers Anton Ertl <> wrote:
>> Yes, 16 bits were called a word on such 8-bit processors (there was
>> another name for 8-bit units: byte), and it was commonly needed,
>> because these machines used 16 bits for addressing their 64KB address
>> space.
>But not quadword and octoword.

Sure, but once they had used "word" for 16 bits, they apparently had
no fantasy left for names of larger units (hmm: phrase, sentence,
paragraph; at least "line" and "page" are used:-).

>> But it also was marketed as a successor to the PDP-11, with various
>> compatibility features. And apparently it's important to keep "word"
>> the same size when doing such successions, and so 16 bits are still
>> called a word even in the Alpha architecture.
>It is also important to show the technology advances. VAX was
>supposed to be DEC's entry to the 32 bit world. Keeping the word size
>at 16 bits dilutes the effect.

It's just a name. I would prefer to use the natural word size for
"word" (i.e., 64 bits on Alpha and AMD64). But I guess there are
groups of people who reuse stuff from the smaller machines on the
bigger ones, and if the width of a "word" changed between the
architectures, interpreting documentation for such things would become
confusing ("Does this documentation mean a PDP-11 word, a VAX word, or
an Alpha word in this place?"), so every manufacturer decided to avoid
changing the number of bits in a "word" even if that meant that a
"word" was smaller than a natural machine word.

>I do remember that in early Alpha C compilers a C (long) was 64 bits,
>but later changed to 32 bits with (long long) as the 64 bit type.

C's long int type is 64-bits on Alpha (and other 64-bit architectures)
on Unix, and AFAIK 32-bits on Windows (even 64-bit Windows), with
64-bit long long int on both. I don't think you get Windows for Alpha

- anton
M. Anton Ertl

Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.