[net.micro.16k] Gee, 68k lovers in .16k; DSI news and optimizer

jk@utastro.UUCP (John Krist) (12/05/85)

     Golly gee-whiz, Batman! People sure don't like the NS 32000
  series.  It seems there's a bunch of 68k lovers sneaking onto
  net.micro.16k.  I've gotten my share of hate mail.  Will those with
  32000 series chips in their computers mind speaking up and telling
  me what they think about it?

     Mr. Gilmore seems unimpressed with the optimizer on the C and
  Fortran compilers.  The compilers are highly optimizing without
  specifying the optimizer at compile time.  It's just that setting
  the optimizer seems to invoke the "let's make the code a little
  longer for more speed" mode of the compilers.

     I correction to my first review: 
        The graphics demo required loading the ANSI.SYS file in
  order to switch modes.  My mistake. Sorry.
        The Thousand Oaks BBS does work, and I have used it about 3 times
  to get new info and to talk to the DSI tech persons.

  * The Dhrystone benchmark results will take a little time to bring
    to those who wanted them.  Right now, to use the UNIX times and
    time functions, you have to have a Interupt Control Unit on the
    board and the appropriate loader software to use it.  The DSI
    people have told me that a revised loader is almost done which will
    make use of the MSDOS timer.  Meanwhile, I have mailed the
    Dhrystone to them to run on their machine. *

  The UNIX being tested is Version 7.  You need a ICU and MMU to run
  it.  ( I don't see the need for UNIX when the MSDOS shell is fine. )

  The graphics interface is being developed for the Hercules and EGA
  cards.  This will allow multiple screens to be set up in DSI memory
  and dumped into IBM memory.

  I must say that the people at Definicon have been very helpful, and
  I am very impressed with their help.  They also have assurred me that
  they will stay up to date with new NS 32000 products.

  For those who seem sick and tired of me talking about this on the net,
  I am just excited about the product and impressed with what it offers
  for the price.  All you 68k people seem to have your noses stuck in 
  in the air when someone mentions the 32032, but who had a working
  full 32 bitter when Mota and others of similar INTELlect were in 
  development ( Don't tell me of the 486, 'cause I don't see it around
  anywhere). I can't wait for those 68k flames!

                                        Ready and waiting,
                                         John Krist
                                         U. Texas Astronomy Dept.
                                         (jk@utastro.UUCP)

mdb@aicchi.UUCP (Blackwell) (12/06/85)

In article <136@utastro.UUCP> jk@utastro.UUCP (John Krist) writes:
>
>
>  The UNIX being tested is Version 7.
>
I have been talking with the company doing the Unix port for Definicon,
and they assure me it is System V r2.  I have some literature about it
(I know... what does that prove?...:-), and it looks real nice.  It is
based on National Semi's validated port called System V/Series 3200.
I also have the manuals for the Unix, sanz a few update pages for the
DSI-32.  These are AT&T manuals with updates for the 32k and DSI-32.

>  You need a ICU and MMU to run it.

As far as new chips go, you will need *specific revisions* of the
CPU and MMU.  These, for all practical purposes, will be available
only through Definicon as an upgrade package.  This because the
chips needed are still in limited production.  I tried to get them
through the NS & TI Distributers with no success.  No word yet as
to whether there will be a cost for the upgrade.

>     ( I don't see the need for UNIX when the MSDOS shell is fine. )

This all depends on what your idea of fine is. 
If you just want to run one thing real fast (*not* a bad idea!) the MSDOS
shell *is* just fine.   However, if you have a need for concurrent tasks,
its another story.  The people at Definicon swear by DRIs Concurrent CPM/
DOS.  However, my experience has shown several fatal (for me)flaws in CCPM.  The
main one is that CCPM does not work with the Cartridge Drives I have
installed in my system.  DRI ackowledges this, but has no fix or explanation,
though one of their techs said something about buffers not being flushed.

>
>  For those who seem sick and tired of me talking about this on the net,
>  I am just excited about the product and impressed with what it offers
>  for the price.  
>
>                                        Ready and waiting,
>                                         John Krist
>                                         U. Texas Astronomy Dept.
>                                         (jk@utastro.UUCP)

I agree, lets here some more from DSI-32 owners!  Anybody got some nifty
code they want to share?  How about ideas for *using* all this power, now
that we've got it?  Even queries on how to do things...   Lets hear from you!

					Mike Blackwell
					mdb@aicchi.UUCP
					(..ihnp4!aicchi!mdb)