Amiga 3000UX Color X

[David S. Herron]

In article <1641 at amix.commodore.com> ag at amix.commodore.com (Keith Gabryelski) writes:
>>b&w X is only temporary with Amix.  Right now the engineers at C= are
>>working on (and RSN will be shipping) a version of color X supporting
>>all of the Amiga resolutions/colors that use the specialized Amiga
>>graphics chips.
>
>This is all correct.  black and white X11R3 was available in Amiga
>Unix 1.1.  It supported normal and A2024 (Hedley) modes.  This version
>of X also supported the A2410 (Lowell board).
>
>Amiga Unix Version 2.0 will have X11R4 whose port is quite a bit faster
>than the current release (%90-%100 faster).

I (yesterday) got a chance to play with an A3000UX.  They have
them at Computer Showcase in San Francisco, I wandered in hoping
for a chance to look at it and ended up spending ~ 4 hours helping
them set it up with TCP/IP & NFS talking to an A2000.

The X is exactly as slow as X11R3 is on 386 clones that I've worked
with.  That is, PAINful.  Oh, and the screen is much too small
(they didn't have an A2024 around so it was displaying on a 1950 ...).

The speed improvement in moving to X11R4 is *extremely* believable.
I saw an improvement like that when we made the R3->R4 change on
our Sun workstations last summer.  There will also be a big benefit
from compiling everything with `gcc' (like they say they're going
to do) since `gcc' is a really good compiler and specifically
*shines* on 680x0's.

There weren't many of the normal X utilities in /usr/X/bin.  Assumably
this will change in 2.0?

Notice that everything is under /usr/X, like the Intel X11 I work
with on 386 clones here.  Normally X gets installed in /usr/bin/X11,
/usr/lib/X11, etc.  That's no biggie, really, but does it also have
another `oddity' which the Intel X11 has.  Namely:  there is
/usr/X/lib/xdaemon which runs all the time, which X programs consult
for turning DISPLAY specifiers into a connection to a server.  It
does this by running a transport specific program
(/usr/X/lib/net/<provider>/nameserver) to do the mapping.
(The code to do all this is available in source in the MIT X11 R{3,4}
distribution in a directory named "att-nameserver").

This may sound like a kinda weirdo way to things, especially since
it's rather different from every/most other X implementations.
But it's kinda nice since the X client doesn't need to concern itself
with transport specific details -- simply by editting some
configuration details into /usr/X/lib/Xconnections you can
control which "nameserver" is used for each transport system.
Be it StarLAN, TCP/IP, or ISO protocols.

We couldn't figure a way to get "netstat" to display information
about the ether card.  "netstat -i" only showed it for the
loopback device.

AmigaDOS floppy disks can be read with "cat" but that's not very
useful since it doesn't give you *files*.  It should be a fairly
simple manner for C= to provide filesystem drivers for AmigaDOS
and MS-DOG file systems.

Didn't get to try to get any idea of the actual speed of the system.

The normal console screens "repainted" pretty fast in `vi'.  But
I think an AT&T 7300 (Unix PC) also repaints its screen that fast.

It certainly seemed to be a complete Unix implementation...

Unfortunately this didn't settle any of my doubts or thinking
about which system I am going to buy.  (I am looking to buy
a "home workstation" sometime very soon y'see).  *SIGH*, if
only C= could be trusted to not screw this up ... ;-).


	David



-- 
<- David Herron, an MMDF & WIN/MHS guy, <david at twg.com>
<- Formerly: David Herron -- NonResident E-Mail Hack <david at ms.uky.edu>
<-
<- "MS-DOS? Where we're going we don't need MS-DOS." --Back To The Future