home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!ibmarc!drake.almaden.ibm.com!drake
- From: drake@drake.almaden.ibm.com (Sam Drake)
- Newsgroups: comp.unix.aix
- Subject: Re: AARRRRGG! No one can answer this question!!!
- Message-ID: <1884@coyote.UUCP>
- Date: 29 Jul 92 07:59:56 GMT
- References: <76399@ut-emx.uucp> <Brts9v.1ss@news.cso.uiuc.edu> <76531@ut-emx.uucp>
- Sender: news@ibmarc.UUCP
- Organization: IBM Almaden Research Center
- Lines: 30
- Nntp-Posting-Host: drake.almaden.ibm.com
-
- In article <76531@ut-emx.uucp> ddt@ccwf.cc.utexas.edu (David Taylor) writes:
- |>: Were the IBM developers so boneheaded as to *not* memory
- |>: map the display?
- |>
- |>Believe it or not, many modern graphics displays aren't memory-mapped.
- |
- |I choose not to believe it. IBM PC's and Sun SparcStations are memory
- |mapped. That's a healthy start and accounts for the most popular PC's
- |and workstations in the world.
-
- Well, that's the low end. How about at the high end? For example,
- does SGI memory map? (Just asking, I don't know.)
-
- See, the argument that says, "low end 2-D consumer grade systems all use
- memory mapping, therefore 3-D million-vectors-per-second systems should,
- too" is specious. Different requirements and performance points may imply
- different design choices.
-
- As for the contention that fast general-purpose processors makes the need for
- special-purpose graphics coprocessors vanish: I don't think so ... since
- the same advances in techology that make GP processors faster from year
- to year effect special-purpose coprocessors in the same way. If a coprocessor
- is faster than a GP processor today, chances are it'll be that way 2 years
- from now, too. Remember, the i860 (a popular chip for building MP
- systems) was designed as a graphics coprocessor, NOT as a general purpose
- computer.
-
-
- Sam Drake / IBM Almaden Research Center
- Internet: drake@almaden.ibm.com BITNET: DRAKE at ALMADEN
-