Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/5/84; site ism780c.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!gamma!epsilon!zeta!sabre!petrus!bellcore!decvax!ittatc!dcdwest!sdcsvax!sdcrdcf!ism780c!tim
From: tim@ism780c.UUCP (Tim Smith)
Newsgroups: net.micro,net.arch
Subject: Re: Re: 386 Family Products (8086 sucks
Message-ID: <154@ism780c.UUCP>
Date: Fri, 13-Dec-85 22:07:19 EST
Article-I.D.: ism780c.154
Posted: Fri Dec 13 22:07:19 1985
Date-Received: Mon, 16-Dec-85 04:27:55 EST
References: <133@daisy.UUCP> <97800013@ima.UUCP> <464@looking.UUCP> <879@ecsvax.UUCP> <466@looking.UUCP>
Reply-To: tim@ism780c.UUCP (Tim Smith)
Organization: Interactive Systems Corp., Santa Monica, CA
Lines: 31
Xref: watmath net.micro:13089 net.arch:2298

In article <466@looking.UUCP> brad@looking.UUCP (Brad Templeton) writes:
>
>But all this aside, is the 8086 the processor to run these heavy floating
>point applications?  Certainly not without the 8087, and if you're that
>keen on speed but insist on 8086 you should have gone to 286, where it is
>actually possible to have segment 2 64K beyond segment 1, allowing full
>sized objects with little work by the compiler.  I don't know if anybody
>has done this, though.
>
It could have been done with no work from the compiler if Intel had
put the bits in a reasonable place.  A full pointer has a selector and
an offset.  Here is what they look like:

    Selector:
    --------------------------------------------------------
    | Segment Number ( 13 bits )  | Other Stuff ( 3 bits ) |
    --------------------------------------------------------

    Offset:
    --------------------------------------------------------
    |             Offset in Segment ( 16 bits )            |
    --------------------------------------------------------

If they had swapped the Segment number and the Other Stuff, then one
could stick 64k segments one after the other, and do reasonable stuff
with pointers, and everyone would live happily ever after.  Could someone
from Intel please explain why it was done this way instead?  The only guess
I have heard is that since the Segment Number is an index into a table of
eight byte things, they saved having to do a shift by three.
-- 
Tim Smith       sdcrdcf!ism780c!tim || ima!ism780!tim || ihnp4!cithep!tim