Re: [M100] OS9 for M100?

[Cobergland]

I can throw in a few $...but am limited budgeted. I do like seeing real stuff 
get done. Thanks all. 

Sent from my iPhone

On May 2, 2016, at 4:34 PM, "John R. Hogerhuis"  wrote:

> 
> 
> On Monday, May 2, 2016, Ken Pettit  wrote:
> Thanks Gary!   I didn't even think about something like Kickstarter 4 years 
> ago.  Might be an interesting effort to post something there and see what 
> type of support it receives.  :)
> 
> Ken
> 
> Have to make sure to publicize it in retrocomputing community if you went 
> that way. And I'm sure there are still some nostalgic reporters that loved 
> the model 100 that would throw some funds at it. There'd be interest beyond 
> our little sewing circle here just hard to say how much. 
> 
> Best case it actually gets fabricated in silicon and custom plastics if 
> there's real funding. 
> 
> For real success there would have to be applications outside retrocomputing. 
> 
> -- John.  

Re: [M100] OS9 for M100?

[Ken Pettit]

On 5/2/16 4:34 PM, John R. Hogerhuis wrote:


Best case it actually gets fabricated in silicon and custom plastics 
if there's real funding.


Yeah, my mind keeps asking "How can I slip this into the ASIC I'm 
working on at the office" :)


Ken

Re: [M100] OS9 for M100?

[John R. Hogerhuis]

On Monday, May 2, 2016, Ken Pettit  wrote:

> Thanks Gary!   I didn't even think about something like Kickstarter 4
> years ago.  Might be an interesting effort to post something there and see
> what type of support it receives.  :)
>
> Ken
>

Have to make sure to publicize it in retrocomputing community if you went
that way. And I'm sure there are still some nostalgic reporters that loved
the model 100 that would throw some funds at it. There'd be interest beyond
our little sewing circle here just hard to say how much.

Best case it actually gets fabricated in silicon and custom plastics if
there's real funding.

For real success there would have to be applications outside
retrocomputing.

-- John.

Re: [M100] OS9 for M100?

[Ken Pettit]

Thanks Gary!   I didn't even think about something like Kickstarter 4 
years ago.  Might be an interesting effort to post something there and 
see what type of support it receives.  :)


Ken

On 5/2/16 4:15 PM, Gary Weber wrote:
On Mon, May 2, 2016 at 4:08 PM, Ken Pettit > wrote:
> I had most (or maybe all) of this tested and working 4 years ago as 
I recall.


On 5/2/16 4:01 PM, John R. Hogerhuis wrote:
> So where do I buy one :-) Take my money, now!

I share this sentiment as well!

Brilliant idea, Ken.  Sure would be an interesting and quite useful 
hybrid laptop machine, indeed.  If you started a Kickstarter or a 
GoFundMe, I would be there to "encourage" you in this endeavor.  ;-)


Gary

Re: [M100] OS9 for M100?

[John R. Hogerhuis]

Oh I know another reason to use special NOPs. NOPs are opcode 00 so which
tends to be a special character. Probably some or all NOP-alternatives can
be embedded directly in BASIC string literals. And don't think BASIC file
I/O can be used for NULs.

So whatever. They are minor edge cases and they could be addressed as
necessary with patches.

-- John.

Re: [M100] OS9 for M100?

[John R. Hogerhuis]

On Mon, May 2, 2016 at 3:58 PM, Ken Pettit  wrote:

> Really?  Something uses "MOV A,A".  It's just a NOP.  Who would have
> thought.
>
> Ken
>
>
I was thinking maybe they were going for a specific cycle count but it
looks like a NOP should be the same cycle count as mov a,a (4)

I'd guess it was either a multiple entry instruction or someone just doing
it because they could. The mood/mov struck...

-- John.

Re: [M100] OS9 for M100?

[Ken Pettit]

I just looked at my files for the extended 8085.  They are all dated 
April 2012!  Has it really been 4 years since I worked on that stuff?  
Man time sure does fly by.


Anyway, I saw that I also remapped the RSTV opcode since it is basically 
useless in the M100 ROM, and I also have extended opcodes for LJMP and 
LCALL (24-bit JMP and CALL).


That project was built around an Actel IGLOO chip (sitting on a board 
with limited peripherals).  I should dust off that project and get it 
running on my Digilent NEXYS 4 or Zedboard board.


Ken

On 5/2/16 3:18 PM, John R. Hogerhuis wrote:



On Mon, May 2, 2016 at 3:01 PM, Ken Pettit > wrote:


I used the MOV A,A   MOV B,B, MOV C,C, etc. opcodes and remaped them. 



Clever! As you know though there's all kinds of strange code in the 
ROM used in "byte fighter" techniques where the programmer coded a 
jump into the middle of an instruction effectively creating multiple 
entry points into the same instruction with different outcomes 
depending on the entry point. Did Bill ever use those opcodes as 
"special no-ops"?


-- John.

Re: [M100] OS9 for M100?

[John R. Hogerhuis]

On Mon, May 2, 2016 at 3:01 PM, Ken Pettit  wrote:

> I used the MOV A,A   MOV B,B, MOV C,C, etc. opcodes and remaped them.


Clever! As you know though there's all kinds of strange code in the ROM
used in "byte fighter" techniques where the programmer coded a jump into
the middle of an instruction effectively creating multiple entry points
into the same instruction with different outcomes depending on the entry
point. Did Bill ever use those opcodes as "special no-ops"?

-- John.

Re: [M100] OS9 for M100?

[Ken Pettit]

No, a CPLD wouldn't do it.  The Complex Instruction architecture of the 
8085 causes the gate count to be way to high for a CPLD.  Even before 
adding any extended instructions / addressing modes.


Ken

On 5/2/16 3:00 PM, Stephen Adolph wrote:

I think there might be some better low power options now, than years
ago.  I have not looked into it at all.  A CPLD implementation does
not seem likely given the need for DLLs and timing circuits.

Ken might know?


On Mon, May 2, 2016 at 5:40 PM, John R. Hogerhuis  wrote:


On Mon, May 2, 2016 at 2:38 PM, Stephen Adolph  wrote:

I recall now... I knew I had none of the answers but I knew you would!
So, what did you do to create the new instructions?


And I'm interested to hear about the relative power consumption.

-- John.

Re: [M100] OS9 for M100?

[Ken Pettit]

I used the MOV A,A   MOV B,B, MOV C,C, etc. opcodes and remaped them.  
The new instructions I added are:


SBcc - Short conditional branch
==
   Branches +31 / -32 conditionally using 1 byte following the opcode 
(4 conditions) (2 byte opcode)


BR  - Branch
==
  Branches +127  -128 using 1 byte following the opcode  (2 byte opcode)

Bcc - Conditional branch
==
   Branches +4095 / - 4096 relative conditionally (8 conditions) (3 
byte opcode)


Rcall - Relative CALL
==
   CALL relative +32767 / -32768  (3 byte opcode)

STP - Set Page
==
   Set the Extended address page (bits 23-16) from the contents of A.  
Each 16-bit register has it's own extended address page (HL, DE, BC, SP, 
PC, plus one for indirect opcodes such as STA, LDA, SHLD, LHLD)


   The current page directs indirect (HL), (DE), (BC) accesses to occur 
in a different 64K region


GTP - Get Page
   Get the Extended address page (bits 23-16) into A

LPUSH
==
   Push 24 bit versions of HL, DE, BC, SP

LPOP
==
   Pop 24 bit versions of HL, DE, BC, SP

LRET
==
   Return from a subroutine using 3 byte address POPed into PC from stack

Ken

On 5/2/16 2:38 PM, Stephen Adolph wrote:

I recall now... I knew I had none of the answers but I knew you would!
So, what did you do to create the new instructions?

On Mon, May 2, 2016 at 5:29 PM, Ken Pettit  wrote:

I already have that.  It supports 16M address space plus I already patched
the ROM so the ISR functions properly jump to the the lower 64K region and
return to the upper regions if needed.

Ken


On 5/2/16 1:32 PM, Stephen Adolph wrote:

ok, here is a proposal.  Let's get a small FPGA and make an 80c85V2
that includes extra instructions...make it retro compatible.

On Mon, May 2, 2016 at 4:19 PM, John R. Hogerhuis 
wrote:


On Monday, May 2, 2016, Bruce H McIntosh  wrote:

On 2016-05-01 03:03, Hiraghm wrote:

I know awhile back folks were talking about a CP/M port for the Model
T.
Earlier today I was watching some Youtube videos on CP/M, which
reminded
me of OS/9, and that got me wondering.
OS9 is re-entrant and position independent, multi-user and
multi-tasking.
The original OS9 was designed to operate in 64k of ram.


OS/9 had a lot of stuff that relied utterly upon the 6809 CPU's
architectural quirks. It was eventually ported to the 68000 and 80386.
The
8085 might be a bit of a stretch.

https://en.wikipedia.org/wiki/OS-9


And it would be slow. You would have to use some expensive tricks to
simulate position independent code.

CP/M is way more doable and opens up access to a lot of compilers and
other
applications. The roadmap and hardware is all there for the cp/m project.
Just need the people and gumption to get it done.

-- John.

Re: [M100] OS9 for M100?

[Stephen Adolph]

I think there might be some better low power options now, than years
ago.  I have not looked into it at all.  A CPLD implementation does
not seem likely given the need for DLLs and timing circuits.

Ken might know?


On Mon, May 2, 2016 at 5:40 PM, John R. Hogerhuis  wrote:
>
>
> On Mon, May 2, 2016 at 2:38 PM, Stephen Adolph  wrote:
>>
>> I recall now... I knew I had none of the answers but I knew you would!
>> So, what did you do to create the new instructions?
>>
>
> And I'm interested to hear about the relative power consumption.
>
> -- John.

Re: [M100] OS9 for M100?

[John R. Hogerhuis]

On Mon, May 2, 2016 at 2:38 PM, Stephen Adolph  wrote:

> I recall now... I knew I had none of the answers but I knew you would!
> So, what did you do to create the new instructions?
>
>
And I'm interested to hear about the relative power consumption.

-- John.

Re: [M100] OS9 for M100?

[Stephen Adolph]

ok, here is a proposal.  Let's get a small FPGA and make an 80c85V2
that includes extra instructions...make it retro compatible.

On Mon, May 2, 2016 at 4:19 PM, John R. Hogerhuis  wrote:
>
>
> On Monday, May 2, 2016, Bruce H McIntosh  wrote:
>>
>> On 2016-05-01 03:03, Hiraghm wrote:
>>>
>>> I know awhile back folks were talking about a CP/M port for the Model T.
>>> Earlier today I was watching some Youtube videos on CP/M, which reminded
>>> me of OS/9, and that got me wondering.
>>> OS9 is re-entrant and position independent, multi-user and multi-tasking.
>>> The original OS9 was designed to operate in 64k of ram.
>>>
>> OS/9 had a lot of stuff that relied utterly upon the 6809 CPU's
>> architectural quirks. It was eventually ported to the 68000 and 80386. The
>> 8085 might be a bit of a stretch.
>>
>> https://en.wikipedia.org/wiki/OS-9
>>
>
> And it would be slow. You would have to use some expensive tricks to
> simulate position independent code.
>
> CP/M is way more doable and opens up access to a lot of compilers and other
> applications. The roadmap and hardware is all there for the cp/m project.
> Just need the people and gumption to get it done.
>
> -- John.

Re: [M100] OS9 for M100?

[John R. Hogerhuis]

On Monday, May 2, 2016, Bruce H McIntosh  wrote:

> On 2016-05-01 03:03, Hiraghm wrote:
>
>> I know awhile back folks were talking about a CP/M port for the Model T.
>> Earlier today I was watching some Youtube videos on CP/M, which reminded
>> me of OS/9, and that got me wondering.
>> OS9 is re-entrant and position independent, multi-user and multi-tasking.
>> The original OS9 was designed to operate in 64k of ram.
>>
>> OS/9 had a lot of stuff that relied utterly upon the 6809 CPU's
> architectural quirks. It was eventually ported to the 68000 and 80386. The
> 8085 might be a bit of a stretch.
>
> https://en.wikipedia.org/wiki/OS-9
>
>
And it would be slow. You would have to use some expensive tricks to
simulate position independent code.

CP/M is way more doable and opens up access to a lot of compilers and other
applications. The roadmap and hardware is all there for the cp/m project.
Just need the people and gumption to get it done.

-- John.

Re: [M100] OS9 for M100?

[Bruce H McIntosh]

On 2016-05-01 03:03, Hiraghm wrote:

I know awhile back folks were talking about a CP/M port for the Model T.
Earlier today I was watching some Youtube videos on CP/M, which reminded
me of OS/9, and that got me wondering.
OS9 is re-entrant and position independent, multi-user and multi-tasking.
The original OS9 was designed to operate in 64k of ram.

OS/9 had a lot of stuff that relied utterly upon the 6809 CPU's 
architectural quirks. It was eventually ported to the 68000 and 80386. 
The 8085 might be a bit of a stretch.


https://en.wikipedia.org/wiki/OS-9


--

Bruce H. McIntosh  scots...@afn.org  www.afn.org/~scotsman  WA4UF

 Network geek with a strong affinity for Telecasters

Re: [M100] OS9 for M100?

[John Gardner]

https://en.wikipedia.org/wiki/OS-9

That's amazing!  Thanks, Bill.

Re: [M100] OS9 for M100?

[Bill Nobel]

It would be a huge task.  I am one of the co-authors of Nitros9 which was a 
complete rewrite of the original Microware version that took advantage of the 
Hitachi 6309 CPU with enhancements.  Level 1 we have it running in 32k barely, 
not much room left for programs to run.  Level 2 would be a little more 
difficult.  Sure the expanded memory of Rex and such.  But Level 2 had a unique 
Vector page in hardware that would keep 512 bytes of RAM at the top of memory 
no matter what memory is mapped into the workspace would have this constant 512 
bytes always there.  It is used to do the task switching and memory mapping for 
the numerous tasks you had running.

Now if anybody is interested the entire Nitros9 source code is available freely 
at: https://sourceforge.net/projects/nitros9
It's completely free to download and look at.  There are many people working on 
it now, but at the beginning of the project back in the ‘90s it was just 3 of 
us that reverse engineered the code and re-wrote it.

Bill Nobel
b_nobel@ hotmail.com


> On May 1, 2016, at 1:03 AM, Hiraghm  wrote:
> 
> I know awhile back folks were talking about a CP/M port for the Model T.
> Earlier today I was watching some Youtube videos on CP/M, which reminded me 
> of OS/9, and that got me wondering.
> OS9 is re-entrant and position independent, multi-user and multi-tasking.
> The original OS9 was designed to operate in 64k of ram.
> 
> I understand that the 8085 doesn't have relative addressing? I think that 
> could be worked around, however. Maybe.
> But I was wondering if maybe a custom subset of OS9 could be developed for a 
> rom image, taking over the Model T, and be designed to work with the Rex, Rex 
> 2 and Quad add-ons?
> An OS that could "flatten" the memory banks of the Quad would be fantastic, 
> and working with the Rex 2's built-in storage would, imo, make the Model T 
> much more useful & powerful.
> 
> I don't know enough to know whether or not such a thing is feasible.