home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!olivea!hal.com!decwrl!adobe!usenet
- From: zstern@adobe.com (Zalman Stern)
- Newsgroups: comp.arch
- Subject: Re: What's in a name?
- Message-ID: <1992Jul27.165807.29014@adobe.com>
- Date: 27 Jul 92 16:58:07 GMT
- References: <1992Jul25.075130.7192@spool.cs.wisc.edu>
- Sender: usenet@adobe.com (USENET NEWS)
- Organization: Adobe Systems Incorporated
- Lines: 33
-
- I'll make the following points to support Matt Frank's criticisms of Intel's
- backward compatibility in particular:
-
- 1) The 486 still supports an incredibly baroque segmentation scheme that no
- interesting piece of 32 bit software uses. (I.e. Segmentation should have
- been left out of the 32 bit architecture and the gates and other resources
- used to do something productive.)
- 2) The 286 protection bits in the middle of an address debacle in which the
- Intel architects insisted on sticking with protection model that did not
- (does not, and never will) jibe with software people really write and use.
- (I consider the 286 responsible for most of the pain in IBM PC clone memory
- models.)
- 3) The performance anomalies in which the previous generation of an Intel
- Arch. run MS-DOS software faster. Especially so when there are much faster
- clockspeed variants to be had. like the AMD 286 and 386 parts vs Intel 386
- and 486. (If your goal is to run old software fast, then you better had do
- that.)
-
- There are examples of better ways to do things as well. On the extreme edge
- there is the HP3000->precision architecture and DEC VAX->Alpha. For a good
- example of thinking ahead in architecture design, look at the R3000->R4000
- 32 to 64 bit architecture expansion. (The jury may still be out on this
- though. The architecture could have been reworked for 64 bit mode, perhaps
- profitably.) An interesting thing about the R3000->R4000 is that MIPS II
- binaries (32 bit) will run in MIPS III mode (64 bit) unless they depend on
- previously undefined opcode raising an illegal instruction exception. Is
- this true for any other architecture transitions? (286 code in 386 mode?)
-
- Backward compatibility can be good, but it can also be an excuse to keep on
- doing stupid things "'cause we've always done it that way."
- --
- Zalman Stern zalman@adobe.com (415) 962 3824
- Adobe Systems, 1585 Charleston Rd., POB 7900, Mountain View, CA 94039-7900
-