Coco/Atari OS9 new hardware

[Iggy]

I know this is far off our usual topics.
But I think I've previously mentioned my fondness for the Tandy Color Computer, the Coco3, the 6809 processor (and its 6309 Hitachi variant), and the OS-9 operating system.

I have some friends who still market hardware for the Color computer (memory expansions, hard drive interfaces, 6309 conversions, stc), Cloud-9.

http://www.cloud9tech.com/

They've just introduced a new adapter board.
It allows you to install a 68B09E (or a Hitachi 63X03E) into an Atari 800XL, 1200XL, or 130XE.
This allows you to run the multi-tasking, multi-user operating system NitrosOS-9 (which is a variant of Microware's OS-9 operating system which powered CD-i video disc players and was available for the Color Computer).

While lacking the polish of AmigaOS, OS9 is a very powerful operating system still used today in a lot of real time process control applications.
With the adaption of Atari hardware, many of the features which the Color Computer lacks will now be available.

I think this will make one neat 8-bit system.

[Iggy]

The problem with CD-i was the crappy resolution and the compression necessary to fit a video CD onto a disc with only the capacity of an audio CD.
And the OS underlying the system was virtually transparent.
Anyone that's seen Level II OS-9 on a Color Computer 3 (or the 68K version) would have to admit that its an unusually powerful OS (if somewhat intimidating since it can do some scary things via CLI).

The thing I've always been amazed by is the fact that an 8 bit OS could support a real time multi-tasking, multi-user operating system.

I've seen a Color Computer running this driving three VT100 terminals and operating a Point of Sale system.

[Iggy]

I don't think it's all that much of a shocker; people think preemptive multi-tasking and multi-user capability are more advanced than they really are because it took so long for Mac and Windows machines to get them, but even by the late '60s mainframes could run OSes implementing these features. The 6809 (like the 6502) also has the advantage that there aren't so many registers to save as modern CPUs have, so context switches are fairly quick.

True, I've even seen multi-user OS' on a Z80.
Several years ago some one gave me an MPM system to play with.
With all the similarities between CPM and MS-Dos it wasn't very hard to figure out the command syntax.

However, multi-tasking on early Intel processor was difficult, operated poorly, and wasn't very useful. Until the i386 came out, it didn't really make sense.
That's one of the things that's always bugged me about the Amiga.
Early on, it had so many advantages over the PC.
Pity it didn't develop as quickly.

[Iggy]

Blame the segment-register kluge - without an MMU, programs are either stuck with 64KB code + 64KB data, or they play around with the segment registers themselves and imperil system stability, or the OS has to do any segment-switching for the program, which would be cumbersome as all hell. (The 286 had an MMU, but bone-headedly it wasn't compatible with ordinary 8086 code such as any given PC BIOS and went essentially unused - I think Xenix might've supported it, but that's it.) The 386 was the first to combine the ideas that "memory-mapping would be useful" and "but it should also work with the software and hardware everybody's already using" and come up with something useful.

Weird, isn't it?
the 68000 (and even the 6809 before it) could support multi-tasking quite well (even without an MMU), but it took Intel three revisions before they could get it right.
How'd we blow that advantage?

[Iggy]

Software.
 
The reason that you need a 386 to multitask is because everyone wanted to run their old dos software, which assumed it was the only program running.
 
It would be trivial to write a multitasking operating system for an 8086, however you'd have no software to run on it. There was no reason to ditch all the software and start again, because the 8086 wasn't quick enough to warrant it.
 
I don't think anyone realised just how big and important the PC would become, or it would have been designed better.

Yeah, software, I almost forgot that.
That's why I left Delmar Co in the early '80s.
you can have a better system, but without sottware it doesn't matter.

[Iggy]

It's worth noting that the Data General NOVA and later the seminal Xerox Alto both ran on bit-slice ALUs, not even what we would think of as proper "CPUs".
 
Minix runs passably well on an 8086, although one needs to keep in mind memory issues (max 1mb addressable). While I don't think you could run a GUI on top of that, you never know. There's a guy building a Mac-like OS for 8-bit Atari systems and it looks slick. If that'd hit back in the 80s I'd probably still have one

I've ordered one of Cloud-9s 6809 adapters and I'm going to install a Hitachi 63B09E in it.
One of the first things I'd like to do is explore porting a GUI to run on top of NitosOS9.
 
BTW - Does anyone know if its possible to overclock an 8 bit Atari?
I can get 63C09E processor that can easily be clocked to 3.58 MHz.

[Iggy]

I've seen a pseudo-GUI running OS/9 - it's like an early DOS window manager/tree type interface (uses text in discrete boxes separated by ASCII borders versus blitting, etc.)

In the late '80s a company I worked for (Delmar Co.) ported a GUI called GWindows to our 68K based OS9 systems, but that was X Windows/Motif based and by today's standards it would pretty dated. Also it was running on a much faster machine.
 
Tandy had a product called Multi-Vue for its variant of OS9, but it was pretty dreadful. If I remember correctly you couldn't overlap windows.
 

That is incumbent on how reliant the Atari is on specific opcodes, local bus speeds. etc. You mean for running OS/9 though, not the Atari's basic interpreter and DOS, right?

From what I can tell, 8 bit Atari computers are a lot like the Color Computer in that a lot of their timig is derived from a common clock.
 
This make overclocking impractical.

[Iggy]

I don't think it's all that much of a shocker; people think preemptive multi-tasking and multi-user capability are more advanced than they really are because it took so long for Mac and Windows machines to get them, but even by the late '60s mainframes could run OSes implementing these features. The 6809 (like the 6502) also has the advantage that there aren't so many registers to save as modern CPUs have, so context switches are fairly quick.

There are a couple things I failed to point out about OS9 and NitroOS9.
Not only are they both real time multi-tasking, multi-user OS', but they both use position independent re-entrant code.  
While this can make programming more complicated as you can't hard code locations, it allows multiple processes to call the same module (requiring only one copy of a program module for multiple uses).
With the limited amount of memory in an 8 bit system this allows for my efficient coding.
And I've seen 6809 systems with as many as 255 active program modules.

I'm looking forward to using this with the Atari 130XEs list driven graphics processors.

[Iggy]

So does this OS work with the VBXE?


Either way it looks like I'm going to have to order one of these. My Atari's love to be upgraded!

Actually, xdude, I'd hold off buying one of these unless you're really into coding.
Because as near as I can see, there no software for it at all.
My first moves will be to study the hardware memory maps for Atari 8 bit systems and write some simple routines i can use for graphics and sound.

I also hope to get Pascal running fairly quickly (as it doesn't have a native graphics library) and then hope to manage to get Basic09 to run (without its Color computer specific GFX library).

Enhancing an assembler with some routines specific to Atari hardware would also be useful.

In some ways a 16 bit 6502C replacement would make more sense as the 6809 isn't 6502 compatible (so no existing software is going to run).

I am going to start compiling a list of OS9 titles that work or can be modified to work.

And I really look forward to using Antic and the GTIA.

[Iggy]

When I was a lad, I wanted to buy a coco2.  I ended up buying a c64 because of the games.  

OS9 and the 6809 as stated in the thread are a very nice combination.  I learned how to machine language and assembly on a multi-68020 OS 9 cluster.  This is part of the reason I wanted an Amiga, that and a production job.  Assembly on the 680x0 is so easy as compared to x86.

Ah, someone else with OS9 68K experience.
I was the retail manager of Delmar Company when we introduced the System IV and System V computers (68000 and 68020 based respectively).
I even encouraged our owners to invest in a port of Gwindows (a GUI written by Steve Adams for Gespac) for our systems.

In fact, I just exchanged a message today with a guy who wrote a basic text editor we used a lot (VED) Bob van der Poel.

But Steven is right, the Coco3 actually has some advantages over the 8 bit Ataris. Its display generator doesn't have Sprites, but it does support up to 640 by 192 resolutions and 80 column text displays. And unlike earlier Cocos, its got a interrupt generator that eliminated the need for software timing loops (and it can be expanded to 512K).
Plus it will run older Color computer software, Coco3 specific titles, and OS9 Level 2.

Right now an Atari modified with this board will only boot NitrosOS9.

I just like the 6809 processor more then the 6502 (almost as much as the 68000).

[Iggy]

I've also decide to buy a TerASIC DE-1 Board and run Gary Becker's FPGA implementation of the Coco3.

http://groups.yahoo.com/group/CoCo3FPGA/

Apparently it overclocks quite well.

[Iggy]

It would, except the 65C816 is kind of a compromised design. It's good for providing expanded capability, higher clock speeds, and near-perfect backwards compatibility, but it's nothing you'd really want to build an OS on top of.

Probably not, but they're still available in 4 and 5 MHz versions with both Plastic DIP 40 pin and PE —PLCC 44 pin packages.
16MB addressing and good backward compatibility.

As the company that used to produce these was bought by ON Semiconductor (which doesn't appear to support this anymore), does anyone know where to get in depth technical data?

I'm interested in some of the added addressing modes in the 16bit operating mode.

[Iggy]

Good stuff CJ.
At least this will get me started.
I already have a design that would allow replacement of a 6502 with one of these (some Commodore fanatic got one to run in a VIC20).

[Iggy]

My nostalgia here is pretty darn strong, and those refurbished Coco3's arent too bad a price. Wrestling with myself here as whether or not to dip into my savings to buy one. My biggest problem is my lack of room. Even if I do buy one Im going to have to make up a makeshift place to put it, which'd make using it less friendly than it otherwise could be.

Slight sidebar, but has anyone seen the fpga coco3 systems?

Gary Becker's stuff?
Yeah, I've seen one. It works pretty good, although it needs another computer to store files and simulate a disk drive. According to Gary you should be able to set this up to start immediately as a Coco3 and load software from flash.

I'm purchasing an FPGA board soon since one of the boards this will run on is only $150 and it will simulate up to a 24MHz Color Computer.

Speaking about OS-9, I found out there was an  Amiga port done by Microware back in the old days (It was priced about  400 to 800 USD). The port was being updated on each OS-9 release, and if  I am not mistaken, it is still supported by the company that took over  Microware OS-9 development.

Has anyone ever had some experience with it?

If I'm not mistaken, that was a third party port by an Australian company.
I don't know if it was updated or if its still available.
There was also an AtariST port that used to be available on the internet.

As a side note, the 68K version of OS9 was just updated by Microware/Radisys last year.

[Iggy]

Gary Becker's stuff?
Yeah, I've seen one. It works pretty good, although it needs another computer to store files and simulate a disk drive. According to Gary you should be able to set this up to start immediately as a Coco3 and load software from flash.

I'd buy one if it had the classic joystick ports, a built in floppy controller and simulated orchestra and sound speech card. Otherwise, may as well just use a pc set up nicely with any of the coco 3 emulators, or an actual coco3.

Steven

Personally, as I still have some program paks I'd like to have a cartridge port.
I don't really need floppies, but I'd like a hard disk interface (flash doesn't hold enough - although for this application it might be adequate, my original Coco3 hard drive was 40MB).
Since the sound and speech pak isn't supported by much software its not an issue for me (might be if I'd ever owned one).

Lack of joystick ports is a bigger issue.

I understand that they're working on an i/o board.
And I've talked to one developer that would like to incorporate a real processor (although I like the superior speed of the emulated CPU).