[comp.unix.microport] Pity the poor 286 folk

rcd@ico.ISC.COM (Dick Dunn) (09/22/88)

John Plocher was following along the 286 virtual memory and virtual/real
bug discussion when he said:

> This is really moot, tho, because the 386 has none of these problems [but
> it has a few of its own], and a 386 box onow costs the same as a 286 system
> did a year ago...

Sure, but that doesn't mean it costs as much as a 286 box today.  More to
the point, a 386 box costs a lot of money and I already have a 286 box, so
the 386 isn't of any help.  [This is me personally, not me as part of ISC!]

I hope Microport isn't planning to leave 286 users out in the cold just
because the 386 is in vogue...there are a lot of 286 boxen out there
(aren't there? I hope?! hello?) which want a good UNIX.  Saying, "Hey, just
change the memory (and add some more) and change the processor/motherboard"
isn't a useful response.  I want to upgrade software (from 2.2.0!), not
hardware.

Of course, if somebody wants to trade me even, 286 for 386 motherboard, I'm
listening!
-- 
Dick Dunn      UUCP: {ncar,nbires}!ico!rcd           (303)449-2870
   ...Worst-case analysis must never begin with "No one would ever want..."

hedrick@athos.rutgers.edu (Charles Hedrick) (09/22/88)

The whole discussion about doing VM on the 286 may have been
misleading.  The fact is, it isn't entirely Microport's decision.
They just do device drivers and various details for what is basically
an ATT/Intel product, System V/286.  I don't think Microport has the
resources to do their own 286 port of System V from scratch.  If I'm
right, we won't have VM on the 286 until ATT and Intel decide to
produce a 286 port of SVr3 or SVr4.  I've not seen any evidence that
they are interested in doing that.  ATT seems to be concentrating on
putting new features into SVr4, and on getting the 386 port right.
They aren't even doing VAX ports anymore.  So I find it hard to
believe that they're going to do a new 286 port.  We can reasonably
expect Microport to fix the various problems with SV/AT, including
serial line performance, floating point, give us support for newer
disks, and do various other neat new things.  But I'd be very
surprised to see them do a basic new kernel.  Happily surprised you
understand, but surprised.

jfh@rpp386.Dallas.TX.US (The Beach Bum) (09/23/88)

In article <Sep.22.04.57.32.1988.5403@athos.rutgers.edu> hedrick@athos.rutgers.edu (Charles Hedrick) writes:
>The whole discussion about doing VM on the 286 may have been
>misleading.  The fact is, it isn't entirely Microport's decision.
>They just do device drivers and various details for what is basically
>an ATT/Intel product, System V/286.  I don't think Microport has the
>resources to do their own 286 port of System V from scratch.

uPort probably couldn't do a port from 3B2 sources to PC/AT on their
own, but there is little reason they can't hack in some support for
"segment swapping".  the fact is, the number of routines inside the
kernel which are affected by paged out segments is probably very small.
trap.c is probably the most affected, followed by various members of
sys[1234].c and three or four others.

virtual memory for any system which can support it is hard, but far
from beyond the reach of a company the size of uPort.
-- 
John F. Haugh II (jfh@rpp386.Dallas.TX.US)                   HASA, "S" Division

      "Why waste negative entropy on comments, when you could use the same
                   entropy to create bugs instead?" -- Steve Elias

jack@swlabs.UUCP (Jack Bonn) (09/23/88)

In article <9695@ico.ISC.COM> rcd@ico.ISC.COM (Dick Dunn) writes:
>Sure, but that doesn't mean it costs as much as a 286 box today.  More to
>the point, a 386 box costs a lot of money and I already have a 286 box, so
>the 386 isn't of any help.  [This is me personally, not me as part of ISC!]

What ever happened to the 386 that would plug into a 286 socket?  I remember
seeing a description of it in the literature, yet the 386SX seems to have
none of the qualities of the 80288 as it was called.  Is there another chip
waiting in the wings (or has its wings been clipped)?  I suppose that a
daughter board would probably do the trick, but my guess it would cost as
much as a 80386 mother board.  Any further info?
-- 
Jack Bonn, <> Software Labs, Ltd, Box 451, Easton CT  06612
uunet!swlabs!jack (UUCP)	jack%swlabs.uucp@uunet.uu.net (INTERNET)

mann@intacc.uucp (Jeff Mann) (09/24/88)

In article <9695@ico.ISC.COM> rcd@ico.ISC.COM (Dick Dunn) writes:
>I hope Microport isn't planning to leave 286 users out in the cold just
>because the 386 is in vogue...there are a lot of 286 boxen out there
>(aren't there? I hope?! hello?) which want a good UNIX.  Saying, "Hey, just

Yeah!  My uport286 system is pretty good, for what it is, now that 2.3 has
reduced the almost daily serial port panics, and I figured out how to use
getty with 8n1.  But it is a pretty insular universe.  I wanna run MS-word!
Is that too much to ask, without having to get some kind of klunky dosmerge
(if there even is one that works)??  Are there any decent applications I can
run?  I have spent the last year writing code on it, but I can't do everything!

I called our dealer to ask how to connect into an appletalk network, he said:  
"there is NO support for ANY kind of networking with V/AT!"  What kind of UNIX 
can't do networking???  The only thing I can think of is to use the KA9Q tcp/ip 
package over the serial port, that's really gonna please my MS-Word users!

I want a tape backup.  I'm *real* sick of swapping 50+ floppies around, but
I sure as hell can't afford $2000.00 (Canadian) for the ONLY drive supported.
That's more than my whole system is worth!   And what about SCSI??????

Well, that's what I want, but I don't think I'll get it.  I doubt that the
sales volume is there to justify all that work, especially if there is a
generic merged V/386 coming up soon which will support lotsa 3rd party
applications, networking, and devices.  So I guess Microport can't really
be blamed for leaving us "out in the cold".

Brrrrrrrrr........    wanna buy a used AT?
-- 
| Jeff Mann - Inter/Access, Toronto           ...uunet!mnetor!intacc!mann  |
| "A picture is worth 256 thousand words"  {utzoo, utgpu}!chp!intacc!mann  |

sjb@dalek.UUCP (Seth J. Bradley) (09/24/88)

In article <2577@swlabs.UUCP> jack@swlabs.UUCP (Jack Bonn) writes:
>In article <9695@ico.ISC.COM> rcd@ico.ISC.COM (Dick Dunn) writes:
>What ever happened to the 386 that would plug into a 286 socket?  I remember
>seeing a description of it in the literature, yet the 386SX seems to have
>none of the qualities of the 80288 as it was called.  Is there another chip
>waiting in the wings (or has its wings been clipped)?  I suppose that a
>daughter board would probably do the trick, but my guess it would cost as
>much as a 80386 mother board.  Any further info?

The 386SX is available, but as yet is in short supply.
While it is compatible with the 286 architecture, it
does require a few logic chips to interface it to the
bus.  My guess is you'll see boards with a 386SX (with
a socket for a 387SX) and a few logic chips that will
pug into a short slot just for power.  A cable will
then connect this board to the 286 socket on the motherboard.
I'm not making any guarantees, but this arangement should
cost much less than a 386 adapter board or motherboard.
The disadvantages are you are stuck with a 16 bit data path
and the processor will run no faster than your system clock.
But if you are running UNIX, you can upgrade to the 386 version
which will run 2-3 times faster even at the same clock speed!
-- 
Seth J. Bradley     UUCP: uunet!lll-winken!dalek!sjb
		Internet: lll-winken.llnl.gov!dalek!sjb

vandys@hpindda.HP.COM (Andy Valencia) (09/25/88)

/ hpindda:comp.unix.microport / jfh@rpp386.Dallas.TX.US (The Beach Bum) /  9:07 pm  Sep 22, 1988 /
>...the fact is, the number of routines inside the
>kernel which are affected by paged out segments is probably very small.
>trap.c is probably the most affected, followed by various members of
>sys[1234].c and three or four others.

    This one seemed too topical for me--here I sit waiting for a build to
finish for a critical VM system bug which puts me on the critical path.
On a weekend.  When my wife is in San Francisco at a museum, having fun.
No, VM is not easy to do right.  It starts working, and then you stress it
just a bit and it breaks.  So you fix those, and write tests, and accidentally
stress it a bit in a different way.  And it breaks.  And once it starts
getting solid they start asking for more performance.  And you make it go
faster, but then it breaks on something else again.  No, VM systems are
pervasive, and hard to keep under tight control.  I respect uPort's move
(or rather, lack of a move) in this area.

					Andy Valencia

vandys@hpcupt1.HP.COM (Andrew Valencia(Seattle)) (09/26/88)

/ hpcupt1:comp.unix.microport / mann@intacc.uucp (Jeff Mann) / 12:27 pm  Sep 23, 1988 /
In article <9695@ico.ISC.COM> rcd@ico.ISC.COM (Dick Dunn) writes:
>...  I wanna run MS-word!
>...
>"there is NO support for ANY kind of networking with V/AT!"

	Oh, come now.  It isn't THAT bad!  My wife is a writer and does just
fine on our AT.  Microemacs, configured to a 'T' for her, along with tpr
(which I hacked to give her the features she wanted) seems to carry a pro-
fessional writer pretty well.  If her formatting needs ever get too fancy
(she used to use Final Word II--she knows what she's missing, but doesn't
seem to care yet) I'll take a look at migrating her to TeX, troff, ditroff,
or something along those lines.  There seem to be a LOT of options, even
leaving out value-added sources like SCO's Lyrix package.

	I received an offer from uPort just a bit ago for a TCP/IP package.
I'm not saying it worked, but I am saying I saw one for sale.  For that
matter, I believe KA9Q's TCP/IP package works fine over 802.3, so there
might not be a need to go serial out of your AT.  Worth looking into.
Keep the faith!  The resources available for a UNIX box are enormous--but
sometimes they're a little hard to find.

					Good luck,
					Andy Valencia

davidsen@steinmetz.ge.com (William E. Davidsen Jr) (09/27/88)

In article <2577@swlabs.UUCP> jack@swlabs.UUCP (Jack Bonn) writes:

| What ever happened to the 386 that would plug into a 286 socket?  I remember
| seeing a description of it in the literature, yet the 386SX seems to have
| none of the qualities of the 80288 as it was called.  Is there another chip
| waiting in the wings (or has its wings been clipped)?  I suppose that a
| daughter board would probably do the trick, but my guess it would cost as
| much as a 80386 mother board.  Any further info?

  I believe that a duaghter board will be in the $200-400 range, with a
socket for the 80387. My rumor factory says that the 386SX has the same
problems with a 287 as the regular 386, and that use of the 387 will be
suggested for systems being used in protected mode.

  How about some info, vendors?
-- 
	bill davidsen		(wedu@ge-crd.arpa)
  {uunet | philabs}!steinmetz!crdos1!davidsen
"Stupidity, like virtue, is its own reward" -me

rcd@ico.ISC.COM (Dick Dunn) (09/29/88)

> In article <9695@ico.ISC.COM> rcd@ico.ISC.COM (Dick Dunn) writes:
> >...  I wanna run MS-word!
> >...
> >"there is NO support for ANY kind of networking with V/AT!"

But I wrote nothing of the sort, and I can't imagine any circumstances
under which I would possibly have the slightest desire to run MS-word.

Nor did I make any sort of claim about V/AT networking.

If you're going to edit parent-article text in your followup, try to get
the attributions so they appear consistent.
-- 
Dick Dunn      UUCP: {ncar,nbires}!ico!rcd           (303)449-2870
   ...A friend of the devil is a friend of mine.

mann@intacc.uucp (Jeff Mann) (09/30/88)

In article <10770001@hpcupt1.HP.COM> vandys@hpcupt1.HP.COM (Andrew Valencia(Seattle)) writes:
>In article <9695@ico.ISC.COM> rcd@ico.ISC.COM (Dick Dunn) writes:
>>...  I wanna run MS-word!
>>...
>>"there is NO support for ANY kind of networking with V/AT!"

Actually, it was me, not Dick Dunn, that said that.
>
>	Oh, come now.  It isn't THAT bad!  My wife is a writer and does just
>fine on our AT.  Microemacs, configured to a 'T' for her, along with tpr
>(which I hacked to give her the features she wanted) seems to carry a pro-
>fessional writer pretty well.  If her formatting needs ever get too fancy

Yeah, I've got uEmacs running here, and to me it seems great.  I don't know
*why* people in my office prefer to use MS-WORD on a DOS XT, but they do, and
there's no way I'm going to be able to convince them to learn a new word 
processor just so they can work on this machine instead.  For a professional
writer, like a professional programmer, the time spent learning a new system
is easily made up for by time saved by that (better) system.  But for the
casual user who just has to write a few letters every week, the overhead of
learning a new program isn't justifiable.  

My point is, how many software/hardware vendors are going to be interested in
making their product work with Microport?  And how much time and effort is
Microport going to be willing to put into improving the '286 port?  Not much! 
Only because the user base (sales potential) isn't there.  When (if?) the
merged XENIX compatible generic 386 UNIX is available, you can be sure that
there will be a multitude of device drivers and applications available.  For
instance, a TOPS interface so I can hook up all the Macs and Dos machines 
here.  And it's gonna run MS-WORD!  In terms of integrating a uPort '286
system into an existing "standard" office environment, well, it just ain't
possible.  And I'm not going to hold my breath.  But for specialized
applications, like the videotex database we are running on the V/AT here,
it's pretty good (if you can write your own code).  And of course it will
run most of the stuff off usenet.

>	I received an offer from uPort just a bit ago for a TCP/IP package.
>I'm not saying it worked, but I am saying I saw one for sale.  For that
>matter, I believe KA9Q's TCP/IP package works fine over 802.3, so there
                                                         ^^^^^
But there are no drivers available for this, as far as I know... are there?


-- 
| Jeff Mann - Inter/Access Videotex, Toronto  ...uunet!mnetor!intacc!mann  |
| "A picture is worth 256 thousand words"  {utzoo, utgpu}!chp!intacc!mann  |

nelson@sun.soe.clarkson.edu (Russ Nelson) (10/01/88)

In article <1988Sep29.153539.17169@intacc.uucp> mann@intacc.uucp (Jeff Mann) writes:

   In article <10770001@hpcupt1.HP.COM> vandys@hpcupt1.HP.COM (Andrew Valencia(Seattle)) writes in part:
   >	I received an offer from uPort just a bit ago for a TCP/IP package.
   >I'm not saying it worked, but I am saying I saw one for sale.  For that
   >matter, I believe KA9Q's TCP/IP package works fine over 802.3, so there
							    ^^^^^
   But there are no drivers available for this, as far as I know... are there?

I'm working on a driver for the 3c501 (ack, but that's what I've got).
Having never written a Unix device driver, it's taking a little while.
I really, really like the way that device drivers get linked into the kernel,
and the fact that devices are accessed through file names that are otherwise
ordinary names.  I'd like comments on this plan:

ecinit()
{ initialize the device }

ecopen()
{ reserves the device for the first opener. }

ecclose()
{ free the device }

ecwrite()
{ send a packet. }

ecread()
{ return a packet from a buffer. ( not sure if I should malloc or use
  system buffers.)  Do not block -- return zero if no packet available. }

ecintr()
{ if transmit, restart transmit.  If receive, stuff packet into buffer. }

ecioctl()
{ ECGETA: get the ethernet address.
  ECSETA: set the ethernet address.
  ECSTAT: return error statistics.
  ECZERO: zero statistics.
}

--
--russ (nelson@clutx [.bitnet | .clarkson.edu])
To surrender is to remain in the hands of barbarians for the rest of my life.
To fight is to leave my bones exposed in the desert waste.

neighorn@qiclab.UUCP (Steve Neighorn) (10/02/88)

In article <1988Sep23.152733.6249@intacc.uucp> mann@intacc.UUCP (Jeff Mann) writes:
>
>I want a tape backup.  I'm *real* sick of swapping 50+ floppies around, but
>I sure as hell can't afford $2000.00 (Canadian) for the ONLY drive supported.
>That's more than my whole system is worth!   And what about SCSI??????

We are all sick of 50+ floppies, which is why lots of people have tape
drives. Microport sells an Everex tape driver package for $100. You
purchase this package, get a Wangtek 60meg tape drive, an Everex
EVA831 tape controller card, and a cable. Total cost from mailorder
houses, about $500 for the hardware. So for $600 total, you can have
an external tape drive. Power can be run from the inside of the box via
an extender power cable. This is the cheap route. You could also get a
complete Everex tape drive system which includes all of the above hard-
ware in a nice box with a power supply for about $850.

I don't know the exact exchange rate between US and Canadian dollars,
but I doubt it is 3:1.

Good luck with your backup quest.
-- 
Steven C. Neighorn            !tektronix!{psu-cs,reed,ogcvax}!qiclab!neighorn
Intel Corporation            "Where we BUILD the Star Fighters that defend the
Development Tools Operation      frontier against Xur and the Ko-dan Armada"
80960 Language Group            work: (503) 696-7264 / home: (503) 645-7015