8-bit OS sought

One MiB of core memory required at the end of core era a toll rack of memory boxes. Previously, several racks would have been required.

Paul

You guys are making me feel old. I used full-featured operating systems (except for the eye candy) that ran in less than 16K words (32Kbytes) of RAM (no ROM -- you booted using front panel switches) off a 2MB 14-inch hard disk for years before the microprocessor became truly useful as a "computer", as opposed to a "programmable controller".

HP's RTE series DEC's RTOS series Data General's ?? XDS General Automation

These all had hierarchical filesystems, multi-user terminal ports, line printers, a range of hard disks, FORTRAN compilers, BASIC interpreters, etc., etc.

And no, I wouldn't like to go back to those days. I like what I'm doing now.

John Perry AS&M

When I started, the computers all ran on steam. We lost many a good coder to boiler explosions in those days. We didn't have any fancy "gooeys" either; just ones and zeros. I remember the great zero shortage of ought-nine - I had to write an entire real-time operating system using only ones. Then I had to port it to COBOL.

The programmers always whine about the boiler explosions... Back when we hired hardware engineers, we'd ask them to hold up their hands. Anyone missing a couple of fingers was an experienced flipflop designer...

You had ones?!?

When I started out, we had nothing but zeros...

Some people might forget that AutoCAD (yes, AutoCAD) started life on a CP/M system. I was running version 1.2 on an S-100 based box. It was sold with an 8087 math co-processor card and a 512K "ramdisk" board. I still have an eight inch floppy somewhere with design files.

I think your memory is probably hazy, ( understandable given your great age...:) - Civilisations generally had ONE first, and discovered Zero later...

See

"The discovery of Zero was made only trice in history: by Babylonian scholars, the Hindus and the Mayans."

The hooked an 8087 to a Z80?

Data Genera's RDOS ran on Nova's and the early model Eclipse's. AOS was introduced for the Eclipse, and AOS/VS followed for the MV line.

And yep, I used to boot using the front panel switches. I think the first removable disk packs that I used were 5 Mbyte ones.

Me neither. It was fun back then though.

Casey

The idea is for a usable standalone 8-bit system with an OS to be created. Possibly something like the commodore64, on which I used to program basic, but I had simpler unix types in mind like minix, elks, qnx, at least the interface. I'd take a DOS for 8-bit systems with enough commands to make the system usable for loading/saving programs, transferring data to another system, maybe with kermit if not tcpip, making simple scripts like basic...

So I'd give up TCPIP if it comes with a good kermit type program, and if the interface allows programs to be loaded/saved from another storage. A very ancient UNIX could be compiled for it, but I'm testing waters for something in common modern use for embedded systems.

news:...

Thats exactly what I was looking for. Ive been aiming for PICs and AVRs. This I'll try.

Thanks all for the help.

This all makes light of things, of course. But there was a lot of reality in John Perry's comments. We really *did* toggle in our bootstrap problems, memorized from frequent use, to load from disk or paper tape. We really *did* use 7 switches and a push button to enter ASCII text. We really *could* read a paper tape directly by just looking at it. We really *did* write sophisticated and fast time-sharing systems with both BASIC *and* assembler support (I did, anyway) using a system with only 16k word of core memory.

Sometimes, in the spate of Windows VxDs, DLLs, COM, ATL, .NET, and the constant propaganda about how all the C++ compilers are just as good at generating code as any assembler programmer is... well, one may very well think that old war stories about commercial grade compilers and operating systems and so on running in 8k or 16k machines are like old-timer stories of having to walk 10 miles to school in the snow, uphill both ways. ;)

But quite satisfying and sophisticated systems really did happen in limited memory. The system I worked on in the early 1970's, for example, ran in 16k of memory, provided time-shared BASIC and assembler, used a 10k swap area for each user that was logged in, managed a complete file system with user accounts and directories, supported up to 32 timeshared users via serial communications over modem, and provided extremely good execution times and at the same time quite satisfying response times when a user entered a command or line (1/4 sec waits were typical.) Of course, it was written entirely in assembly, from scratch.

Jon

Bit of a Freudian slip, there. I meant, "We really *did* toggle in our bootstrap __programs__,"

hehe.

Jon

Put me in front of a PDP-8 and I bet I can still boot it from the front panel switches. I used a PDP=8 to test the original 30-30 Winchester drives and vacuum column tape drives for Mr. Wang.

My memory has since failed me, I suspect. I still have my "8" manuals and I used to program on 8's, as well. Also, the IBM-1130; the PDP-11's of course, PDP-10, IBM-360's, and eventually I was lucky enough to actually build my own Altair 8800 -- which started with a wonderful 256 bytes of memory! (Could not afford to fill the other three sockets with the static ram chips that would have brought this up to 1k.) I often hand-entered bootstraps for the PDP-8, PDP-11, and of course, my Altair.

Jon

CP/M was available for the 8086.

What graphics card(s) did it support?

Bob

Ah yes. CP/M-86. You didn't say you were using that newfangled modern stuff. :)

Let me guess... Concurrent-CP/M-86?

CP/M didn't support any graphics cards.

AutoCAD, however, did. Graphics was strictly an application thing under CP/M. When you built a graphics card you didn't write a CP/M driver for it (except to make it look like a dumb ASCII terminal), you wrote an AutoCAD driver for it. That said, I don't remember any specific S-100 graphics cards, but I remember they were _expensive_ ($thousands).

Here's what google found:

Here's an interesting blurb from '85:

More on Quadram S-100 Boards: Set consists of two IEEE /696 boards, 1) single board computer and 2) graphics board--Valiant is name to be given board-pair! Very appropriate, "courageous" and "valuable." SBC contains new AMD 9580 combined hard and floppy (auto-select between 5.25" and 8") disk controller chip along with AMD 9581 data separator, and of course, Hitachi HD64180 microprocessor. From what we've seen, HD64180 makes Intel 80186 chip look weak. Furthermore, we understand HD64180 has built-in (hidden) circuits to on-chip manage RAM to one megabyte, same as 8088, 8086 and 80186! Such circuits will likely be made operative if demand requires it.

Get ready for real graphics quality from this Quadram duo. Initially, resolution offered is up to 2,048 horizontal (columns) by 1,024 vertical (rows, scan lines) pixels, with 256-color palette out of possible 1,024 palettes: 262,144 total color possibilities! Graphics board uses Motorola 68020 as co-processor, Hitachi HD63484 CRT graphics controller, and up to 2 megabytes of RAM (using 8419 DRAM controller), all dedicated to graphics. NEC 7220 GDC chip simply can't compare in performance to later-design HD63484. Time-of-day stamp is built into SBC hardware and BIOS. OS runs near top- of-RAM using less than 20k-bytes, with applications allowed remainder of 512k- bytes. Z-News 204 has more about these S-100 boards.