I have a multi processor using the SC/MP both the first 8A500 pmos then later 8073.
I have a board I made up with the Nibble basic rom version too.??Their basic issue
was paged the unique 12bit addressing and they were slow compared to 8085 and z80.

When you put cores together on the same silicon you likely have access to the
internal machine and syncing them gets a lot easier.? The yabut is the software
still has to know it has resources and how best to use them to advantage. So
getting multiple cpus to work and share data was easy but Multiplan and Editors
had not one clue?they had additional CPU(s) to work with so no advantage.?

The system software I did take advantage of that but it was far from portable
as it had to know about the resource to use it being CP/M BIOS was the first
place?for devices? to appear.? THe second was doorbell and system control blocks
(SCBs) that were ways to transfer messages, commands, and status.??The
first multitask/multi-user boxes built around MPM and CP/M-3 new of networking
but the number of thing it could do were again not limited by hardware but
applications software so must software at best to print remotely but even then
the software was dumb.

At the time I was also using VAX and PDP-11 systems with multiuser and multitasking?
OSs at their core so as a z80 (z180 and Z280 as well) user it seemed we never got
to a more complete and effective OS with applications in the 8080/8085/z80 realm
of 8bitters.? For the moment Chromix and UZIunix are left out as both had their
own hardware needs and also space limitations to be useful.

Then again when you can have a cpu to yourself neither multitask or multi-user was
high value save for foreground background multitasking as applied to printing
(print spoolers or printer managers).? ?The obvious reason for that is why sit and
wait for printer which had value.

Oddly the 1802 is one of the early cpus where foreground background and simple?
multi task management were both sensible and useful in real time control
applications.

Allison

Hi Allison:
Thanks for taking the time to explain some of the past design issues and your involvement with them.?

Yea, lots of timing and bottle-neck issues for access to resources. Integrated circuit fabrication was still cutting its teeth back then, so things were slow but getting faster.

I think NATIONAL SEMI was one of the 1st to design a mutli-processor CPU arrangement using their INS8060 and INS8070 series SC/MP CPUs. I was lucky enough to have a friend (who used to work for QXI, a NATIONAL SEMI rep.) sample me an INS8073 with NS tiny BASIC built into RAM. It was literally my very first CPU board design (?.?

In the past decade, I've seen other projects with multiple 8051 cores blazing along at some 400 MHz as well (but not with 12 clock cycles per instr. cycle).

To think what you were doing 40 years ago has ALL been stuffed into single-chip "SoC's" (System on a Chip) with quad ARM cores, many megabytes of RAM and FLASH memory and lots of other peripherals happily running at 1200+ MHz (instead of 6 MHz!). Linux to the rescue for multiprocessing as the OS yielded us the (GOOGLE) "ANDROID" virtual JAVA machine user interface.

For certain, its been an interesting journey, the past few decades. In many ways, I miss the simplistic nature of computing from back then. :)

Peace and blessings,
Scott

Scott,

Like many of the time performance a was a thing...
Exity was only one and that was later.? IN 1980 I setup to make the meanest
z80 system I could using 6mhz Z80s that were appearing.? Basic design was?
remarkably simple, 4 z80 CPUs each its own board with 64K of ram and a bus
access protocol (S100 BUS) to share data or resources? (disks, Serial IO and
printer).? ?All of the resources were smart as in each had local Z80, 8085, 8749
(printer buffer) to perform tasks.? The goal was to share IO and offload tasks like
find a file or spool to printer.? That meant doing surgery to CP/M to split out parts
so that the master cpu (server) did end up with all the work (Bios stuff or BDOS)
or running a full copy of CP/M on each or the four cpus.? Lots of interprocess
communications, interrupts, plus a little foreground background on each.
Basically the box was a bunch of networked (via parallel bus) cpus and service
providers.

In the end while I ran it for years and proved useful a single faster cpu was
not burdened with as much overhead.? It also proved CP/M based software
while unaware of the multiple cpus was the actual problem as it was unaware
of resources that could be useful.? The hardware was doable and all but software
had to catch up.? Multiprocessing and time slice was the realm of PDP-11 and VAX
or other heavy iron, the Z80 could but the OS was not yet there.

CP/net and MPM made some things easier? but the the existing applications
software was still built for 64K (or smaller) spaces and had no idea of extensions
or how to use them.? That required something Unix or later and all of the seriously
needed larger memory space.

Allison

Hi Bill,

Yes it all fits together. There is a problem with the FP on top of the SIO though that I haven't tracked down. The FP works fine but there is a segment 'e' I think it is, that is not lit on all characters. I think its a shorted bit on the SIO board, although they all checked out on with an ohm meter when I was testing before plugging it. The CPU and SIO board work fine, as does the FP even when using the bit banging I/O port on the FP with the CPM.

Paul

If speed is an issue, there is always this route...?


Peace and blessings,
Scott

Yes, that is the case.? My Kaypro4/84 has a 1mb ramdisk, very fast.
The compupro also has a 512K ram disk and hard disks with smart
controller, faster even at only 4mhz z80.

AT class machines even with then hard disks that the same bottleneck.

The 8080, well, it was just slow, though it beat the tar out of the 8008
by an easy factor of 10x or more.? Still for the first system I ever?
built it was pretty amazing, then I got to see the 8080 instruction set.

Back then the PDP-8 (e or i) was the benchmark used, later PDP-11.

Generally at each moment I was pushing for best performance or
greatest simplicity.? the Z80 or 8085 generally did that well for
most cases.? For less 8049 or 8051.

Allison

I've collected and build CP/M systems for decades (over 4)
so I have system that run from sludge slow to current idea of fast.
Z80s at 10mhz, Z180s at 12mhz and A Z280 at 10mhz.
That and no shortage of S100, Multibus, and SBC systems
to run.

Biggest bottleneck is mass storage IO, 5.25 floppy generally
is slow, many hard disks were not fast.? ?Of all CF is usually
fastest (best if 8bit mode) and SD slower as the SD has
more protocol overhead and the serializing if not done in
hardware is slow.? Of course there are implementation
dependencies but the faster?systems usually farm out
the disk/storage IO to a slave cpu in some form.??

Speed is good if it can be had simply.? A Z80 running at? 6, 8
or 10mhz can be simple.?

All that said why speed is nice to have, I use multiplan, Dbase,
smallC, BDS-C, Pascal and a lot of other software like Vedit
and speed helps.

Allison

开云体育

Considering the DEC Robin ( I have two) is a 4mhz Z80 with 64K of ram?
with banked boot/terminal roms and slow floppies (TM100s) is the
comparison?that's terrible.

A long time back I got a dozen of the Z80 board used to make up
the Robin, They are 4mhz Z80, 64k of Dram, two 24 pin eprom
sockets (2732), FDC and 4 serial ports, basis for a good system.

So I took one and put on revised roms and it uses a CF
interfaced to the old FDC? socket (no floppies) and its
fast ss any 4mhz Z80.??

Having multiple real systems to compare at 4mhz is nice:

Allison

开云体育

Thanks all,

Working fine now. I downloaded the CPM files from the website and simply copied the contents of the ZIP to the SDcard and then booted the Z80MC with the SIO board but not the FP. Some activity and? then hey presto CPM prompt.?

Works fine. Bit slower than I recall my old DEC Robin, but fine. Will have some fun putting Small C and Pascal on to the card.?

I've got plans for a Z80 PIO and CTC on my expansion board.

Meant to mention that I reversed the pins and sockets on the boards, so that I could plug directly into the CPU board with single core bare wires or male pin back to back breadboard wires, rather than having to use DIP sockets. I also found some 20 way versions of the white connectors from EBay and used those with one 10 way supplied to make the inter board connectors.?

Here's a photo of the completed boards.

Best Regards

Pau l

With care cpm explorer runs in wine.

Either case CP/M knows nothing of fat so what many do is create a
contiguous 8mb file as a container file and then find out what block
address it starts at.? Then CP/M addresses the block directly
without fat.

So CP/M will not read fat files, the file structure is very incompatible.
Its possible to run a utility that does BIOS direct DISK IO and read
FAT using a program that knows fat structure but that's only useful
to get from CP/M to FAT or FAT to CP/M.? I have that for AmproLB+
and Kaypro4/84 its called DOS.? Obvious the media must be compatible
(720K 3.5" floppy was the standard here for that) as back then SD and CF
were far in the future.? The program called DOS is on the SC CD but it
would need a great amount of work for FAT devices larger than 500mb.

SD cards do not care what file format you use.? ?If you partition it you
can have multiple CP/M partitions.? Oh CP/M does not use partition
table on that media either.? ? Use a freshly formated or new SD
(1gb is plenty large) and write a bunch of FAT files of 8MB each then use?
a program to get the starting bock address of each.? The CP/M BIOS
needs to know [when you code it, or as a default that can be changed
for B:/C:/D:... drives] what the block starting address is for a given Logical Drive.

Using linux DD is both your friend and feind.? It can write blocks to specified
addresses easily but only using great care as it can write blocks to specified?
addresses easily (oops wrong blocks or device).

Allison

Paul Bigwood wrote:

I've finally finished my SIO Card - long story!, and it seems to be
working fine, but I can't boot CPM.
I'm pretty sure the SD Card I have used (SanDisk 2GB Black) is the one
supplied with the kit.
Checking it on my W10 PC shows no files but around 100K of data used?
Booting the Z80MC and typing S shows a sub menu L and W. L prompts the
same sub menu and W and prompts to enter a file name? Entering a file
name like TEST.TXT just comes back with a ?. Have I overwritten the
system somehow?

I haven't been able to find any instructions on line. Is there a FAQ I
should be looking at? Can I download the files from somewhere?

Hi Paul,

I format the SD-card in FAT16; but do not pre-load the CP/M files onto it. The computer I'm using is running linux, and it's problematic to load the CP/M files on it because Josh's CP/M explorer is Windows. (I have to get another computer with a working Windows setup to do it).

If you look on my website, you'll find the files to load, and the CP/M explorer program to do it. Once you get them loaded, you should be all set!

Once CP/M is on the Z80MC, then it has the XMODEM program and can transfer any files or disk images with any PC "terminal" program that supports XMODEM.

Lee

--
A designer knows he has achieved perfection not when there is
nothing left to add, but when there is nothing left to take away.
-- Antoine de Saint Exupery
--
Lee Hart, 814 8th Ave N, Sartell MN 56377, www.sunrise-ev.com

Hello Josh,

I've got some excellent news! I figured out what the problem was, turns out the UART chip was the issue. I ordered a replacement because I wanted to test the socketed stuff before I started desoldering, and when I replaced it the board started right up! So your initial suspect was the culprit. Thanks a ton for the help, this has been an absolute blast (not a joke, I have been really enjoying working on this).

Thanks for your help,
Morgan

Hello Josh,

Unfortunately I don't have one of those to test with. I'm mostly limited to just a multimeter at the moment.

Thanks,?
Morgan