Tandy Silver

[Bill Vermillion]

Bill Campbell, the prominent pundit, on Wed, Aug 25 11:29  while half 
mumbling half-witicized: 

> On Wed, Aug 25, 2004, Bill Vermillion wrote:
> >On Wed, Aug 25 08:34 , Bill Campbell, showing utter disregard for 
> >spell-checkers gave us this: 

> >> On Wed, Aug 25, 2004, GCC Consulting wrote:
> >> ...
> >> >> Which chips were these?   There certainly weren't many standards in
> >> >> the early days.  I remember some of the best serial cards for the
> >> >> PC didn't use the brain-dead 8250 Intel chips.   Those were such a
> >> >> pain, particularly for those who integrated them into 
> >> >> multiport serial cards, as they could get into a state where 
> >> >> only a complete power cycle would clear those.

> >> >I'm having a "senior moment here" but I think instead of using
> >> >the 286 processor, they use a 186. It was something akin to
> >> >this. Tandy used a slightly different version of the processor
> >> >everyone else was using.

> >> If I remeber correctly the Tandy 2000 used an 80186, and had
> >> enhanced video capabilities that were (a) very nice, and (b)
> >> incompatible with anything non-Tandy. The Tandy 3000 was the
> >> first non-IBM Microchannel box (and 8MB of RAM for the Tandy
> >> 4000 cost close to 10 grand :-).

> >Now who's having a senior moment.  The 3000s were 80268 machine.
> >Standard AT style bus.  The 4000's were 80386 machines also
> >with the standard AT architecture.  The 5000's were 80386 with
> >MCA architecture.

> I can explain my senior moment in that the only Tandy x86 boxes
> I ever ran were the Tandy 4000s. I wouldn't touch anything
> Intel less than an 80386, and didn't do MCA as it was basically
> IBM proprietary.

Ah. 'splains that one.  I had one site that I wound up working for
part-time for 10 years.  At one time there were 11 Xenix systems
on model 3000s. 

As things changed I replaced five of those with one IBM model 80
[they got a good deal on those as they were on a state contract]
and put in the Specialix RIO cards with those very long links for
serial portts - so one machine wound up replacing five Xenix
systems.  Then they finally migrated onto Netware.

I once wrote a complex [well at least it was complex for me] that
analyzed pay to part-time employees for the past 10 years.   It was
to determine if they were wrongly classes as part-time and I was
calculating penalites based on the Federal penalites for income
tax, which meant those sometimes changed each quarter, and I had to
calculate daily interest within that quarter.

The first run was 12 hours, so the head 'borrowed' the only
'386 machine from one department and we got that run down to
about 5 hours.   After trying at least a dozen different ways of
selecting employess they settled on one.

I made a final run, exported it to the mainframe, and it processed
everthing in five minutes.  The IT head was one of the best
computer programmers I've ever met, and he'd never seen FP code
before and I explained what I was doing and he said "you do good
work".

The reason the runs were long was that these were part-time
employees and there were typically 500 on the payroll at any one
time and we processed 10 years worth of employees.

I was working/processing this every day and on weekends for about
2.5 weeks, while normally I was there for two 6 hour days a week.

I think they finally settled on payment/penalties for just
over $1 million.

I was glad when we got off those slow machines.  The 80386 machine
came from the first department every to have FP there and that was
on an old Tandy 6000.  It's 5 hour processing would have been more
like 5 days on the 6K.

Bill

-- 
Bill Vermillion - bv @ wjv . com