Holy Crap! IMSAI's weren't this expensive when new!

From: ajp166 <ajp166_at_bellatlantic.net>
Date: Sat Feb 10 21:46:19 2001

>You need to get your timing straight, Allison. If you take 1976 as a
>reference, there wasn't any sort of 'C' compiler on the horizon for CP/M

Right C wasn't 1976 then again I never said it was. The assertion was
in 1976 you could do better. Could you?

I asserted that using the 1976 design for many years later I did run C,
pascal, Cbasic, Mbasic, multiplan and even dbase. This still runs
to the idea that that the NS design could not do useful work. It's not
hacking the design, others did it with minimal effort on stock hardware.
It makes little difference that those may not have existed in 1976 (very
little did) it's the idea that the design did work and a 56k TPA was

>for that matter, for NDOS. I don't think 5-1/4" disk drives were cheap
>enough for NorthStar in '76 either. There weren't many of them around.

November 1976, $599, Mastercharge Card, LI Computer in Manhassett Li.
For that I got a MDS-A controller, 1 SA400 and their DOS and BASIC.
At that time another SA400 was about 400$ and 8" drive was $699
plus the then typical 8" drive required a bigger more complex power

The NS Horizon machine (complete S100 crate) was later. I did over
the years run every thing under the CP/M sun on it.

>fact, I don't think anybody I knew at the time had a CP/M version beyond
>1.4. The NorthStar stuff was pretty common by late '77, but the
>compiler-generated software I'm referring to was released in '77-'78.

Not news to me.

>However, much more basic is the absolute fact that you can't have a 56K
>if the available addressable memory space below the FDC is limited to
>because the FDC sits there. Even if the BIOS were of zero length, the
>isn't, and even if the BDOS were of zero length as well, you'd still
have to
>allow for the bottom 256 bytes, which don't figure into the TPA.

The FDC resides at 0xE800!!!!!

Faulty math! 64k minus the 4k (f000h) is 60k and then minus 2k
(e800 to efff) is 58k. Single density bios of 2k is was no big deal.
so with 8k in use or unavailable your down to 56k... or is 64-8 not 56k?

>You're right about the cost of memory in 1976, and, in fact, I'd say it
>quite a bit more than what you suggest in '76, since the 4116's cost
>$80 per chip in '76. Getting eight of them wasn't easy either since
>weren't in full production until late '77.

I know I'm right, I was there buying it. I had 56k of seals and PT
static (2102)
8kx8 ram cards in the crate. My first 64k dynamic card was 1980 paid
$199 for the board sans ram (I had a source).

>Several competitors did! Now, you picked this 1976 timeline, so please
>stick to it! What you did in '78 isn't relevant because it was a wholly
>different world then.

The 76 timeline is the design date of the FDC. The point to repeat it
can you using 1976 technology do better than 600$ for a minifloppy SYSTEM
with dos and basic?

No it was I was doing by 1978 based on my 1976 hardware.

>The TRS-80 Model 1 came out in late '77 didn't it?

First significant quantities were september 1977.

>> in 48k. The only one that wanted a larger TPA then 48k was SMALLC V2
>> compiler as the optimizer wanted more space.
>SmallC didn't exist then ...

The date is not important as I ran it later on the same hardware.
Your trying to weasel your way to a different place.

You have asserted that NS MDS-A system was a dog. I assert
your full of faulty data and that by 1976 hardware standards it
was it was a good design that would held up well for many
years. By then NS* had replaced it with a double
density version that interoperated with older version well yet
extended the useable storage.

I'd also assert that the ICOM FDOS system I have (same timeframe)
using the venerable 1771 was a true dog as it was slow, with rom at
B800 and even less software (barely a dos). It was also only good
for 71kb of storage on a SA400. The NS MDS-A did much better at
82k of space on the same drive. Anyhow I still have he ICOM junker
in the collection to remind me of a real dog.

>in '81, 64K of 4116's cost much less than $100.

Mostly useless as they were loose parts and not a board or kit.
I can pull down catalogs, Byte, Interface age and all for prices.
S100 cards complete, of reliable design were not $100 for 64k
until the sell out days of the mid to late 80s.

>> Psystem with Pascal for $50 (bargan in my eyes).
>UCSD Pascal, while a decent system, wouldn't run standard CP/M

Unimportant. At the time it represented a non CP/M system and a
real HLL for peanuts that did run on small hardware. At the time that
was pretty radical. Also there was a version of Psystem that ram using
CP/M (not NS*).

>The NorthStar controllers I saw were all shipped at 0xE000 with a boot
>that didn't find them anywhere else.

Nope that was nonstandard then. Standard was E800h I've never seen one
even nonstandard that low most of the custom roms version were
F000 or F800. I'd attribute that to error or bad memory.

>That's not exactly what I'm saying, but, in 1976, you couldn't run
>TurboPascal because it didn't exist, nor did BDS-C.

Well certainly your opinion. Like I said earlier I ran them when they
became available on that old useless, as you say, hardware with a real
56k TPA I ram them. I STILL RUN IT, I ahve the docs, I dont have to
rely on memory

>That's not relevant,
>though, because those compilers and interpreters don't represent useful

Again your opinion. For many those did represent useful tools for
business apps.

>People who bought a computer in 1976 generally justified it on the basis
>that they would use it to perform useful work. Programming was overhead
>everyone except the .001% of the folks out there who were engaged in
>creating software.

Save for in 1976 there was very little software at the applications
so many apps were home grown. What year was Visicalc introduced?

>Useful work involved running Word Processing, Accounting, Payroll, and
>Inventory Control software, Order Entry, printing reports, checks,
>cost tracking, shipping label/mailing list processing, etc. That sort
>work required you have some memory available since you had to sort and
>search quite a bit.

I know, it's how I make a living. I was using my altair to run a
payroll and all back then. It's where my appreciation for reliable
came from.

>> Prior to about 1982 that was not only not an issue it was irrelevent.
>> Back in that era running softsector was often not much help as
>> everyone ran something different format wise.
>How about an example, referenced to your 1976 timeline? I don't
believe I
>even saw a 5-1/4" drive until '77. They were probably out there, but
not of
>any particular interest. On the other hand, 8" drives, priced around
>were everywhere, and controllers that used them were not expensive.

They were of no interest as they were bare. Without a controller to make
it work it was not an attraction. However NS*, ICOM, and a few others
by were by fall of 1976 selling product FDC and Drives.

Then again finding a 8" drive in 1976 for $350 would ahve been a windfall
and non of my Bytes or old catalogs support it.

>> support.
>I don't see how you did that when the FDC lived at 0xE000.

Simple, it doesnt. It's standard address, E800h.

>> would have been more convenient but E800 was not a significant
>> Actually if you wanted to burn your own roms it was easy to do.
>That certainly wasn't an option in 1976.

Yes it was. The parts used were not mask parts, they are AIM and fuse
low density stuff. The most complex part on the board was a 256x4
prom. The address was set by a one 74S287 prom as an address decoder.

>> An SA801 and controller was easily $400 more expensive in late 76!
>That's why you built the Northstar in '78. Try sticking with your
>You picked it after all ...

I bought the MDS-A and SA400 in 76!, NS* Z80 board in march 1977. I
used it in the flaky Altair till the NS* Horizon chassis was available to
'78. Your trying to duck out of the issue.

>> as big as the S100 crate above it. I also have a HZ-207 box with two
>> height 8" 2sided DC motor drives and it's still big and weighs a ton.
>My Integrand box has the two drives in it, together with the carcdage
>power supply. Most of the noise is from the fans.

I gave one of those away. The AC motors of the SA801s tend to be
loud too.

>What you did or were able to do in '78 had no effect on what was going
on in
>'76, yet you keep making these forward references. The fact that the N*

What I did with the 1976 reference was to say this... could
you do a better fdc design using 1976 technology then they did.
After that you adotped the date to justify anything that doesnt
suit your assertion of facts that have no support. I have the hardware
multiple examples plus original docs and my notebooks to fall on.

Oh yes, I'm hardcore engineer. That translates too if you didn't
write it down it never existed. So as a result even my hobby hacking
is documented in scores of quad ruled notbooks.

What I was doing in 1978 was using 1976 hardware, it is the point.
Your trying to find a way to justify your arguement that something
was not good at the time because you didn't understand it or didn't
bother to. Your comment that in the early 80s the Boy Scouts
couldnt use them runs counter to my comment that in that exact
same time frame I did {as did others} run them with loads of
heavy apps.

>couldn't use them because they were shipped in an inadequate
>in '76. If you'll draw yourself a timeline of the various events that
>involved here, you'll see why the 56K memory map was inadequate for the
>"serious" business software of the mid-late 70's. That was ultimately
>with a compiler release that suppported smaller

MID 70s is 1975, Altair (introduction was january 1975 Pop tronics) was
very new and disks were limited to the PDP-8 and PDP-11 folks that were
paying lots more than Altair prices.

As to 56k not being enough in the 70s... even the average PDP-11 then was
huge at 128k. Most people I know were quite happy to have 56k in the
late 70s on any hardware. What was truely important to business users
was adaquate relaible mass storage and floppies were barely it. Serious
users wanted and paid dearly for hard disks. Now if you wanted to be
pandantic in 1976 you likely paid $295 (that would be mid year) per
8k of ram, 56k represented ~$2100!!!

late 70s, say 1978-79 was when the microcomputer folks were discovering
disks and trying to make them work for a living. Floppies were seen as
important tot his but any business user will tellyou that a real hard
disk was
the killer hardware. Just like Visicalc was a killer app.

Received on Sat Feb 10 2001 - 21:46:19 GMT

This archive was generated by hypermail 2.3.0 : Fri Oct 10 2014 - 23:33:44 BST