PcW16...

litwr · 10:33, 17 February 19

Hi all
Is there any information about how many the PcW16 were sold/produced? It was the fastest the Z80-based computer. I am curious why did Alan Sugar miss a more advanced CPU for it like the z280 or z380?
I am also interested to run my pi-calculator with the PcW16 (http://litwr2.atspace.eu/pi/pi-spigot-benchmark.html). Could anybody provide me some help for this? Is there a better www-location where I can get more chances to meet somebody with a real PcW16?
Thanks

robcfg · 11:24, 17 February 19

I have a PcW16, @GUNHED has one too if I remember right, and I have 2 or 3 friends that have one.

It may take some time for me to get it out from my storage place, but I'll be glad to help.

GUNHED · 12:14, 18 February 19

It has an advanced CPU (look at the PCB), IIRC it's some kind of custom Z80 (I hope not to mix things up though). The Z280 was buggy (a little bit) and the Z380 was different again.

I prefer the Z80 itself, because it provides all opcodes. The Z180 f.e. does miss some of the important non-documented (8 bit parts of IX and IY f.e.).

If the PcW16 would have a color screen, it would have been very successful. It's OS is really great, better than apple imho.

Bryce · 13:49, 18 February 19

What is a Z180 f.e. ? I only know of the Z180 P and Z180 V which are the only ones that made it to mass production as far as I know. Did Amstrad have some other custom version made?

Bryce.

JonB · 17:30, 18 February 19

I have one, too.

Its Z80 is not a discrete device, it is incorporated into one of the custom chips (the ASIC?). In use it is pretty sluggish and the "cabinet" (a flash memory used to store files) is small. The floppy drive is 1.44 MB but seems to have been intended as a backup drive only. There is a CP/M Plus implementation for it (John Elliot wrote it) which is nice, but it runs as an emulation under the PCW16's OS and has to do lots of page switching to work at all (according to JE). So as a result it isn't much faster (subjectively) than a PCW. Impressive achievement, though.

The main board appears to have an IDE interface (or support for one at any rate) and I believe that it is colour capable. Check out this blog post by @Habi : http://www.habisoft.com/irmia/comentarios.asp?id=335

robcfg · 18:05, 18 February 19

The PcW16 was meant to have VGA output, which is what Habi did to his board.

Also, it was supposed to have an IDE interface, as evidenced by the unused sockets. It's possible to add the missing chips, but you'll be having to write your support for the IDE part.

litwr · 21:25, 18 February 19

Thanks! It looks like the PcW16 is not so rare as I thought.

Is there any emulator for it? I have only a CP/M version of the pi-calculator. It is a PI-GENERIC3.COM file for CP/M from my archive. It uses a CP/M+ standard call to get a value of a timer but the accuracy of this call is very low, only seconds. So it is worth to run the program for 100 (or even 1000) digits several times and get the average. If there is an emulator for the PcW16 somewhere then I could try to write a PcW16 specific version of the program with more accurate time measurement. However it could take a time, I need to find and read the documentation and to write the code.

Quote from: GUNHED on 12:14, 18 February 19
It has an advanced CPU (look at the PCB), IIRC it's some kind of custom Z80 (I hope not to mix things up though). The Z280 was buggy (a little bit) and the Z380 was different again.

I prefer the Z80 itself, because it provides all opcodes. The Z180 f.e. does miss some of the important non-documented (8 bit parts of IX and IY f.e.).

If the PcW16 would have a color screen, it would have been very successful. It's OS is really great, better than apple imho.

Does anybody know something specific about the z280's bugs? Are there a list of them somewhere?

Indeed the Z180 and Z380 missed several useful, albeit undocumented instructions but they had a lot of new instructions and ability to work with large amount of memory without paging. The Z380 could be much faster than the Z80.

IMHO it would have been interesting if PcW16 had used the ARM and software emulation for the Z80. The ARM at 16 MHz could beat 80486 at the same frequency. It could be at least 4-6 faster than the Z80.

Quote from: JonB on 17:30, 18 February 19
I have one, too.

Its Z80 is not a discrete device, it is incorporated into one of the custom chips (the ASIC?). In use it is pretty sluggish and the "cabinet" (a flash memory used to store files) is small. The floppy drive is 1.44 MB but seems to have been intended as a backup drive only. There is a CP/M Plus implementation for it (John Elliot wrote it) which is nice, but it runs as an emulation under the PCW16's OS and has to do lots of page switching to work at all (according to JE). So as a result it isn't much faster (subjectively) than a PCW. Impressive achievement, though.

It is a bit odd that there is no direct way to run CP/M on the Z80-based system. I can't even imagine what prevents just to boot CP/M.

IMHO 1 MB of flash was too small, it should have been at least 4 MB.

remax · 21:38, 18 February 19

Last time i checked, PCW16 was working under Mame. I'm the one who added the softlist.

Also, i don't remember the name but there is a part of the Joyce emulator (Anne if my memory is good), that emulate the PCW16 quite well, and there is another emulator called AnnaRosa, but i never tested it.

JohnElliott · 23:33, 18 February 19

Quote from: litwr on 21:25, 18 February 19
It is a bit odd that there is no direct way to run CP/M on the Z80-based system. I can't even imagine what prevents just to boot CP/M.

It would have meant I'd have had to write a hardware-level driver for the floppy controller, rather than just call the Rosanne functions.
If anyone wants to do their own port (or go one better, implement virtual consoles and bring up MP/M) feel free

JonB · 07:55, 19 February 19

I see no good picture of its main board anywhere, but there is a WinBond chip in the middle of the board. I'll wager this provides floppy, serial, parallel and IDE ports. Take a look at this picture: http://www.habisoft.com/pcwwiki/lib/exe/detail.php?id=en%3Ahardware%3Aplacas&media=hardware:placas:pcw16:PcW16_3501-001P-3_PCB_Top.jpg. You can see the Winbond chip with traces leading off to all the ports. There is a 40 way connector pad too (is this IDE?) as well as what looks like space to expand the flash memory and RAM. The board also hints at colour capabilities (see the connector with power and video - has blue and red lines unconnected) and has a provision for a VGA socket.

It's intriguing, but to John's point, I'm guessing that the Winbond chip is a commodity piece and so it should be reasonably easy to write drivers for it. Not that I have time right now.. but it is even easier to write IDE drivers, if the hardware could be reverse engineered so as to derive the port addresses.

So, a native CP/M BIOS is required.. http://www.gaby.de/cpm/manuals/archive/cpm22htm/ch6.htm

JohnElliott · 08:51, 19 February 19

The port addresses are given in ANNEASIC.RTF:

Code Select

Z-80 Address    S-IO Address    Comment 
00-07           000-007         Nothing selected        
08-0F           1F0-1F7         IDE CS0 
10-17           200-207         Games port at 201       
18-1F           3F0-3F7         FDC and IDE CS1 
20-27           3E8-3EF         Serial Port 2 (IRQ3)    
28-2F           3F8-3FF         Serial Port 1 (IRQ4)    
30-37           368-36F         Nothing selected        
38-3F           378-37F         Parallel Printer 378-37A

ANNE emulates the colour capability and fully-populated memory (2Mb flash, 2Mb RAM) but not IDE.

GUNHED · 15:13, 19 February 19

Quote from: JohnElliott on 23:33, 18 February 19
It would have meant I'd have had to write a hardware-level driver for the floppy controller, rather than just call the Rosanne functions.
If anyone wants to do their own port (or go one better, implement virtual consoles and bring up MP/M) feel free

Honestly, therefore we do need more documentation.

(maybe it changed, because I took a very close look about few years ago). I wish it will get more support.

litwr · 18:50, 20 February 19

Quote from: JohnElliott on 23:33, 18 February 19
It would have meant I'd have had to write a hardware-level driver for the floppy controller, rather than just call the Rosanne functions.
If anyone wants to do their own port (or go one better, implement virtual consoles and bring up MP/M) feel free

Thanks! So a program without OS calls, such as a pi-calculator, will not slow down?

JohnElliott · 20:40, 20 February 19

I never found it particularly slow in the first place - except in very early versions that used Rosanne calls to scroll the screen.

robcfg · 22:34, 20 February 19

On top of that, there's some patched CP/M, if I remember right.

News:

PcW16...