28 Jun
2007
28 Jun
'07
11:59 p.m.
any thoughts? And although he claimed the 2000 never
generated a NMI, won't a parity error always do this?
Irrelevant maybe cuz that's a catastrophic fault
thing, not something typically encountered...
Maybe he provided more details, but I'm tired...
> I was working on a mod, never completed, for the
T2K
> that would make it
> after the O/S was loaded, 100% PC-compatible.
> Required 1 hardware chip,
> and a customized IO.SYS. Whereupon any DOS
> application would run on the T2K,
> even stuff doing serial-port manipulation, direct
> video-memory writes, DMA,
> etc.
The one-chip mod was a PAL that monitored the address
lines from the
CPU,
when It saw an I/O instruction in 'low' address-space,
it generated an
NMI (not used at all by the T2K), then the software
service routine
for NMI unwound the stack to find the offending I/O
instruction, and
re-mapped the "PC" functionality to the T2K hardware.
Coupled with a
timer-tick 'refresh' routine that copied data from "PC
video memory" to
the T2K video memory (remapping attributes, etc. as
required.)
The chip actually had 2 modes of operation -- NMI
active, as described
above, and 'NMI inactive', where it pretty much did
nothing -- except
listen for the 'magic words' that made it go active,
that is. :) This
enable/disable mode switching was necessary, to allow
"T2K DOS"
internals
(and/or the 'sofware service routine) to access the
T2K hardware
directly.
____________________________________________________________________________________
Moody friends. Drama queens. Your life? Nope! - their life, your story. Play Sims Stories
at Yahoo! Games.
http://sims.yahoo.com/
29 Jun
29 Jun
12:37 a.m.
On 28 Jun 2007 at 20:59, Chris M wrote:
The one-chip mod was a PAL that monitored the address
lines from the
CPU, when It saw an I/O instruction in 'low' address-space, it generated
an NMI (not used at all by the T2K), then the software service routine
for NMI unwound the stack to find the offending I/O instruction, and
re-mapped the "PC" functionality to the T2K hardware. Coupled with a
timer-tick 'refresh' routine that copied data from "PC video memory"
to
the T2K video memory (remapping attributes, etc. as
required.)
I've done a bit of that myself--and it's extremely difficult to get
such emulation perfect. There's always some idiot programmer who
used a feature or mode that you never thought *anyone* would use.
F'rinstance--how about port 61H with the "refresh" bit toggling every
50 or so microseconds? It's pretty difficult to emulate that
accurately with a software trap scheme.
Cheers,
Chuck
8:28 p.m.
On Jun 29, 2007, at 7:43 PM, Fred Cisin wrote:
*snicker* I see I'm not the only one who's drinking tonight. ;)
-Dave
--
Dave McGuire
Port Charlotte, FL
any thoughts?
And although he claimed the 2000 never
generated a NMI, won't a parity error always do this?
If running 8 bit memory (without a parity bit), parity errors "NEVER
HAPPEN"! 
30 Jun
30 Jun
12:11 a.m.
Dave McGuire wrote:
Ummm that gives you a *Party* error instead.
Did any of the small computers use error correcting codes
or was that only left to a few large computers?
If running 8
bit memory (without a parity bit), parity errors "NEVER
HAPPEN"!
*snicker* I see I'm not the only one who's drinking tonight. ;) -Dave
12:36 a.m.
Fred Cisin wrote:
On Fri, 29 Jun 2007, woodelf wrote:
Could you disable the PC's memory and just run the ECC
board?
Ummm that gives you a *Party* error instead.
Did any of the small computers use error correcting codes
or was that only left to a few large computers?
In late 1982?, I had a 192K ECC memory board from Boulder Creek
Systems in a 5150 
12:49 a.m.
> In late 1982?, I had a 192K ECC memory board from
Boulder Creek
> Systems in a 5150
On Fri, 29 Jun 2007, woodelf wrote:
Could you disable the PC's memory and just run the
ECC board?
NO.
I could set the starting address, but no documentation on how to set it to
start at 0, and no obvious way to disable the soldered in first row of
RAM.
2:08 a.m.
New subject: Tandy 2000 hack -ECC memory
I was the designer of the Boulder Creek Systems board. It would have
been nice to be able
to disable the PC's base memory to allow all ECC memory but there was no
easy way without
hacking up the main board. At a West Cost Computer Show in 1982, some
IBM employees
came up to my booth and pulled out memory chips while the board was
running on an extender
card. They did not believe it was real ECC memory. After seeing it was
still running, IBM bought
50 boards. That was our biggest sale. No one else really cared about the
ECC, they just wanted
more memory for cheaper. The board had single bit correction, and
multi-bit detection which would
then generate the parity error. ECC codes were generated in a small
bipolar PROM. Somewhere
I still have my main board that I hacked to disable the RAM and run all
ECC cards.
Bob
Fred Cisin wrote:
> In late
1982?, I had a 192K ECC memory board from Boulder Creek
> Systems in a 5150
>
On Fri, 29 Jun 2007, woodelf wrote:
Could you disable the PC's memory and just
run the ECC board?
NO.
I could set the starting address, but no documentation on how to set it to
start at 0, and no obvious way to disable the soldered in first row of
RAM.
3:49 p.m.
New subject: Tandy 2000 hack -ECC memory
On Fri, 29 Jun 2007, Bob Rosenbloom wrote:
I was the designer of the Boulder Creek Systems board.
Thank you!
It would have been nice to be able to disable the
PC's base memory to
allow all ECC memory but there was no easy way without hacking up the
main board.
At a West Cost Computer Show in 1982,
PC Faire?
(Fall 1982, offshoot of West Coast Computer Faire, you were in
the balcony of Cicic Auditorium?)
some IBM employees came up to my booth and pulled out
memory chips while
the board was running on an extender card. They did not believe it was
real ECC memory. After seeing it was still running, IBM bought 50
boards. That was our biggest sale.
No one else really cared about the ECC, they just
wanted more memory for
cheaper.
I cared.
Although I was/am cheap, I'm disgusted by "Macintosh doesn't have a parity
bit, and it NEVER has PARITY ERRORS".
--
Grumpy Ol' Fred cisin at xenosoft.com
7:08 p.m.
New subject: Tandy 2000 hack -ECC memory
On Jun 30, 2007, at 2:08 AM, Bob Rosenbloom wrote:
I was the designer of the Boulder Creek Systems board.
It would
have been nice to be able
to disable the PC's base memory to allow all ECC memory but there
was no easy way without
hacking up the main board. At a West Cost Computer Show in 1982,
some IBM employees
came up to my booth and pulled out memory chips while the board was
running on an extender
card. They did not believe it was real ECC memory. After seeing it
was still running, IBM bought
50 boards. That was our biggest sale. No one else really cared
about the ECC, they just wanted
more memory for cheaper.
...Could that have been the beginning of the divergence of the PC
world from the rest of the computing world? ;)
The board had single bit correction, and multi-bit
detection which
would
then generate the parity error. ECC codes were generated in a small
bipolar PROM.
Most impressive!
Somewhere
I still have my main board that I hacked to disable the RAM and run
all ECC cards.
Very cool! That needs to be documented on a web page somewhere!
-Dave
--
Dave McGuire
Port Charlotte, FL
1 Jul
1 Jul
6:38 p.m.
New subject: Items available for trade
I have been sorting through my storage buildings trying to make room for
more stuff coming in and
have the following items for trade -- for smaller & lighter items! I
have not tested or powered
on any of these so don't know if they work, but they are complete.
1. Control Data, CPI, 9 track reel to reel tape drive, with cables.
2. Hewlett Packard 7979 front loading 9 track tape drive.
3. Hewlett Packard 7980S front loading 9 track tape drive.
4. Hewlett Packard 7937 disc drive.
5. DEC RA81 disc drive, with cables.
6. DEC RL02 cartridge drive, this one may be for parts, it's missing the
door latch parts,
but otherwise looks complete.
7. DEC PDP/11-44 computer. This has the following boards and came with
the RA81 drive:
M7090
M7903 THROUGH M7908
M8743 (four of these memory boards)
M7762
M7258
M7800
TC131 TAPE CONTROLLER
M7527
M7585
M7587
M9302
If anyone's interested, just email me.
Bob
30 Jun
30 Jun
5:34 a.m.
On Fri, 2007-06-29 at 22:11 -0600, woodelf wrote:
Dave McGuire wrote:
Ummm that gives you a *Party* error instead.
Did any of the small computers use error correcting codes
or was that only left to a few large computers?
I had a talking bird that kept saying "Pieces of nine! Pieces of twelve!
Pieces of zero!".
I sent it back though.
Too many parroty errors.
Gordon
If
running 8 bit memory (without a parity bit), parity errors "NEVER
HAPPEN"!
*snicker* I see I'm not the only one who's drinking tonight. ;)
6306
days inactive
6308
days old
12 comments
7 participants
participants (7)
-
bfranchuk＠jetnet.ab.ca -
bobalan＠sbcglobal.net -
cclist＠sydex.com -
chrism3667＠yahoo.com -
cisin＠xenosoft.com -
gordon＠gjcp.net -
mcguire＠neurotica.com