Talk:Amstrad PCW

Games
Were enough games released for the PCW to justify a List of Amstrad PCW games page? 2fort5r (talk) 17:41, 24 August 2009 (UTC)
 * Several dozen were produced, according to, and . I'll create the list. 2fort5r (talk) 19:10, 24 August 2009 (UTC)
 * Seeing as I've ended up here looking for more reason after seeing a youtube video of "over 50 Amstrad PCW games in under 30 minutes", I'd say there's more than enough. Heck, more seem to have been made for it than certain actual games consoles...
 * (link: https://www.youtube.com/watch?v=eQ_d3m-DBcc ) 80.189.129.200 (talk) 18:58, 19 March 2018 (UTC)

CPM app memory use...
I wonder if some of them - particularly the "serious" apps where an occasional split second delay wouldn't be noticable - were smart enough to swap out data to the "RAM disk" so as to effectively use >61kb of memory? Not quite virtual memory, more silently splitting the working file into several smaller subfiles, then auto-saving the bits which aren't currently being displayed and re-using that memory space... Might even be how Locoscript did it. 77.102.101.220 (talk) 14:36, 8 December 2009 (UTC)


 * Locoscript didn't run under CP/M. CP/M and Locoscript both came on separate disks. Locoscript was written especially for the PCW, so of course they were free to use all the features it had, including the RAM. 94.197.127.63 (talk) 23:27, 28 March 2013 (UTC)


 * It's how enough programs on other machines like the ST used RAMdisks, or at least demand-loading what code and data they needed from the disc and later discarding it completely when no longer needed in order to save memory at the expense of speed (which a system with more RAM could accelerate with a virtual drive)... so I wouldn't be at all surprised to find a program written specifically for a Z80 based machine with 256kb or more RAM to use that kind of technique pretty implicitly - maybe not so much saving anything to the extra memory, but certainly keeping all but the most essential code and program data in there and only copying it into working RAM when it was absolutely needed. It's essentially a different kind of memory paging, at the end of the day... 80.189.129.200 (talk) 19:02, 19 March 2018 (UTC)

There is no real need or point in speculating here. The article notes that PCW used CP/M version 3 (aka CP/M Plus). The Wikipedia page for CP/M https://en.wikipedia.org/wiki/CP/M#CP/M_Plus notes that one of the improvements in v3 was support in the API for memory paging. Amstrad liked telling people to do things the correct and official way. So that API and Amstrad's virtual 'M' drive filesystem device define the two ways well-behaved third party programs were expected to use the machines full memory. Sure, there was no need in the PCW design for memory protection, so some PCW programs will have rolled their own schemes separate from the official API too. I agree with 94.197.127.63 that Locoscript was special. It seems likely that the bestselling program-switcher "flipper" also rolled their own. 150.143.30.31 (talk) 16:53, 6 April 2018 (UTC)

3" drive problems
The drive would sometimes fail due to a glorified rubber band stretching or getting slick. I replaced over a dozen of them to get the drives back up and running.

Also the drive used a light to see if the disc was write-locked or not. Sunlight or other bright lights shining into the drive could cause it to not work correctly. —Preceding unsigned comment added by Whittlej (talk • contribs) 19:53, 23 December 2009 (UTC)

info on 3.5" disk history is completely untrue
I'm not quite sure how to reword the paragraph nicely, but

"Rumours that Amstrad opted for the 3-inch disk format because it bought a large quantity of these drives cheaply from Matsushita when makers of IBM-compatible PCs standardised on the 3½-inch format.[26] cannot possibly be true because the PCW was designed in 1984 and the first IBM PC with a 3.5" drive was the IBM_Personal_System/2 launched in 1987."

is flat out false. The IBM_PC_Convertible, introduced in 1986 used 3.5" floppies. Prior to even that, PC-Compatible DOS laptops had used 3.5" floppy drives since at least 1985, such as the Toshiba T1100.

I can't say anything about the reasons that Amstrad did go with the 3.0" drive (I'm going to imagine it was economics), but nearly all of the statements WRT 3.5" use in the PC world are just incorrect.

Perhaps they *did* get a good deal on them since the rest of the world (not just PCs, but Apple, Commodore, Atari, etc.) *did* standardize on 3.5", but that would need some citing.

--Oregonerik (talk) 21:42, 10 February 2010 (UTC)
 * Can we cite by linking to other WP articles? Because really, about the ONLY systems using 5.25" or any other disc size other than 3.5" after 1984 were the IBM PCs and compatibles (whose market share didn't truly skyrocket until the 90s, when 1.4mb 3.5" floppies had become the norm for them also) - including the Tandys, Apple's by-then-obsolescent "II" range (except for the IIgs) and stillborn Lisa, the Sharp X68000 and NEC PC88s/early PC98s which barely made it out of Japan (all 5.25), and the Amstrads/Sinclairs plus a limited few MSX computers that paralleled them in using 3.0 and even smaller sizes (2.5, 2.0...). Newer PC types including the PS/2, plus the Macintosh, Amiga, Atari ST, Acorn Archimedes, and pretty much anything else you could think of including the Sam Coupe (a would-be Spectrum successor) and Amstrad's own portable word processor devices latched on to the 3.5" format like a highly ironic magnet. Perhaps Sugar and his crew thought that miniaturisation would continue even further and they were getting ahead of the curve, but given his usual commercial shrewdness, and similar acknowledged practices by other OEMs (e.g. Atari getting a cheap deal on a load of single-sided 3.5" drives which had the head on the "wrong" side vs the normal setup, and "20mb" hard drives that were actually 30 or 40mb's with one or two faulty heads, not to mention Sinclair's well known practice of using twice the number of half-defective memory chips bought at knock down price and merely installed with the MSB either disconnected or tied permanently active...) I would suggest that Occam's Razor points towards it simply being the cheapest of all possible options that didn't require lumping a big ol' 5.25" drive on the side on the monitor (and there then not being enough space for a dual-drive design given the typical monitor tube size), in the style of older CP/M workstations. 80.189.129.200 (talk) 18:56, 19 March 2018 (UTC)

Daisy Wheel Printer
Article claims

"The daisy-wheel printer could not produce graphics."

If true, then it is the only daisy wheel printer on the planet that can't. Every daisy wheel printer can print graphics by printing full stops (or no full stops) to make up the graphic image. It may not print graphics particularly fast, but it does print them. 109.156.49.202 (talk) 17:12, 6 October 2011 (UTC)

Unless you're aware of software that could do this on the PCW, then the statement is true. I'm pretty sure there was no such software.Markbarnes (talk) 22:27, 30 December 2012 (UTC)


 * Even then, to make graphics out of full-stops would require the print head to be able to move fractional distances horizontally, which is possible since I think it was a proportional font. But it would also need to be able to scroll the paper in tiny distances vertically. Which I'm pretty sure it couldn't. And even if one were able to torture the printer in this way, it'd look AWFUL! Whoever bought a 1-pin dot matrix printer? Where you had to wait for the "." to come round on the wheel before it could print each pixel? It'd take half a day to print a page! And wear out the "." spoke.


 * The claim is nonsense. Daisywheel printers can't do graphics. They do do plenty of sound though!


 * 94.197.127.63 (talk) 23:25, 28 March 2013 (UTC)


 * Au contraire. Daisy wheel printers can do graphics.  All daisy wheel printers use stepper motors to advance the paper, move the carriage and rotate the daisy wheel.  The gearing is such that multiple steps are required for a normal line feed and character width.  I am fairly certain that the PCW as a system cannot do graphics, but that will be a limitation of the firmware (which I believe is actually in the monitor), not the printer mechanism.  I do not understand your point about waiting for the '.' character to "come round ... before it could print each pixel".  Since the '.' is the only character being printed, the daisy wheel itself does not need to rotate at all while such graphics are printed.  No one can claim that the graphics were printed in high resolution or at any great speed, but they were printable.


 * Several (but certainly not all) 'office' type packages than ran under MS-DOS or similar supported graphics on daisy wheel printers, (and some early windows daisy wheel and golf ball printer drivers did as well). I can tell you that Ability Plus certainly did support such graphics for printing out line graphs; bar charts and pie diagrams.  You had time to make yourself a cup of cofee (and drink it) while waiting for the print run.  DieSwartzPunkt (talk) 16:38, 21 August 2014 (UTC)

"Roseanne" or "Rosanne" OS?
The article says that Amstrad PcW16 had "Roseanne" operating system; actually, this OS called "Rosanne". http://toastytech.com/guis/pcwsetup.png 37.112.239.141 (talk) 14:47, 15 June 2013 (UTC)

Booting
The article claims "The ROM-based code cannot display text, being too small to support character generation; instead it displays a bright screen which is progressively filled by black stripes as the code is loaded from floppy."

This is wrong. The boot code only loads the first sector. It knows how to turn the display on and off and a few other cues to stuff a disk in. The first disc sector then loads more and that can do whatever it likes, including implementing the stripes.

82.17.37.186 (talk) 16:03, 20 May 2015 (UTC)]

External links modified
Hello fellow Wikipedians,

I have just added archive links to 1 one external link on Amstrad PCW. Please take a moment to review my edit. If necessary, add after the link to keep me from modifying it. Alternatively, you can add to keep me off the page altogether. I made the following changes:
 * Added archive https://web.archive.org/20110605134222/http://myweb.tiscali.co.uk/rwcrouch/dtp.pdf to http://myweb.tiscali.co.uk/rwcrouch/dtp.pdf

When you have finished reviewing my changes, please set the checked parameter below to true to let others know.

Cheers. —cyberbot II  Talk to my owner :Online 02:40, 30 August 2015 (UTC)

Joyce
The machine was called a Joyce, because that was the name of Sinclairs secretary, and it became the internal name foir it before it was branded PC8256 81.171.59.186 (talk) 23:04, 7 October 2015 (UTC).

Reasonably sure you intended to say "Sugar's secretary" there. Sinclair was a totally separate company/guy at that point in time. — Preceding unsigned comment added by 150.143.20.128 (talk) 23:52, 12 November 2018 (UTC)

External links modified
Hello fellow Wikipedians,

I have just added archive links to 2 one external links on Amstrad PCW. Please take a moment to review my edit. If necessary, add after the link to keep me from modifying it. Alternatively, you can add to keep me off the page altogether. I made the following changes:
 * Added archive http://web.archive.org/web/20110716134644/http://www.micromart.co.uk/features/article/default.aspx?id=22652 to http://www.micromart.co.uk/features/article/default.aspx?id=22652
 * Added archive http://web.archive.org/web/20110718191930/http://www.ansible.co.uk/ai/pcwpluspcwp1996.html to http://www.ansible.co.uk/ai/pcwpluspcwp1996.html#mar

When you have finished reviewing my changes, please set the checked parameter below to true to let others know.

Cheers.—cyberbot II  Talk to my owner :Online 23:47, 27 February 2016 (UTC)

External links modified
Hello fellow Wikipedians,

I have just modified one external link on Amstrad PCW. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:
 * Added archive https://web.archive.org/web/20080705195916/http://labs.pcw.co.uk/2007/07/a-little-bit-of.html to http://labs.pcw.co.uk/2007/07/a-little-bit-of.html

When you have finished reviewing my changes, you may follow the instructions on the template below to fix any issues with the URLs.

Cheers.— InternetArchiveBot  (Report bug) 07:46, 4 July 2017 (UTC)

External links modified
Hello fellow Wikipedians,

I have just modified 2 external links on Amstrad PCW. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:
 * Added archive https://web.archive.org/web/20101117090401/http://www.hud.ac.uk/ITsec/roger%20professional/ownership.pdf to http://www.hud.ac.uk/ITsec/roger%20professional/ownership.pdf
 * Added archive https://web.archive.org/web/20081120002508/http://www.virtuallearning.org.uk/changemanage/digital_kids/Cyborg%20kids.pdf to http://www.virtuallearning.org.uk/changemanage/digital_kids/Cyborg%20kids.pdf

When you have finished reviewing my changes, you may follow the instructions on the template below to fix any issues with the URLs.

Cheers.— InternetArchiveBot  (Report bug) 17:44, 27 July 2017 (UTC)

Price claims read almost like an anachronistic advert for the system...
Which is totally unnecessary. I mean, can we not?

Sure, a complete word processing system for £399 was pretty revolutionary in 1986. But saying that the only other computer capable of the task under £1000, much less £2000, was the Tandy 1000, is a complete barrel of horse feathers. If nothing else, Amstrad's own CPCs could do a reasonable job of word processing, with their own version of CP/M, 80-column screen mode and low-eyestrain green-phosphor monitor option, as well as the exact same CPU and about as much memory directly available (certainly in the 464) and I bet you could put together a CPC based home system with computer, monitor, printer and software for far less than a grand even in 1986. Notwithstanding that, Atari had launched the ST the previous year and one of their grandstand claims was that of selling the first 1mb RAM computer for under £1000 - OK, a monitor was extra, but only about £200 for the PCW-rivalling hi-rez mono model, or nothing if you had a compatible TV or RGB monitor already that could plug right in, and that model came with a double-sided 3.5" disc drive built in. The 512kb model (plus an external single sided drive, rather essential as they canned the idea of selling 128 and 256k models running software from cartridge) was no more than £799, and both of them tended to come with a few essential bundled applications including a basic but useful word processor (not quite Locoscript, but decent enough), art package, etc. Or if you preferred Commodore, you could have a 256k Amiga for under £1500, or a C128 (offering similar or better functionality to the CPC) for rather less than a grand. Not to mention similar quality offerings from Acorn (both the BBC (A, B, Master) on the CPC/C128 level, and original Archimedes on the ST/Amiga level) and a few other marques that were easily under £2k and often under £1k.

Of course, those all would need a separate printer adding on, but a basic 9-pin dot matrix of PCW quality was only a couple hundred pounds extra, so you could, with e.g. an ST, a SCART cable, the bundled software and a base model printer, put together something with PCW-like functionality for less than the cost of a Tandy 1000. Which is probably why the ST and Amiga came to dominate the UK and european computer scene for several years afterwards, with the Mac and PC being rather less popular in the home than at the office, Amstrad and Sinclair's fortunes steadily waning, and Tandy's attempts getting absolutely nowhere.

The PCW (and the later Amstrad portables) remaining as your cheapest dedicated word processing option for quite some time, don't get me wrong, but the gap didn't start out anywhere near as wide as is being made out here, and the introduction of the PCW16 was essentially forced by it having closed up entirely, with the last of the 16-bits selling even more cheaply than the extremely outdated-looking dedicated machines could manage, and the neither the 16 nor any of the short-lived forays into the market by other typewriter and WP manufacturers around that time (particularly from Canon and Brother, who offered such things as an electric typewriter which used line-at-a-time bubblejet rather than impact printing, and both a built-in 1-line LCD editing display plus the option of plugging in a colour VGA monitor and mouse for full-screen word processing and inserting full colour clipart... either readymade, or your own creations from the Paint-like drawing applet... needless to say it died on its ass) being anything like good enough to offer a challenge to a mono-screen ST which by then could be yours for about £200 all up (or less than £150 if you just had the main unit that, thanks to a built in modulator, could now plug into any TV), let alone the growing contingent of budget SVGA home PCs.

I feel tempted to edit that section For Great Truth And Justice, but honestly I think it more likely needs nuked from orbit and completely done over. 80.189.129.200 (talk) 19:21, 19 March 2018 (UTC)

--- Claims in PCW article seem fine to me. Plenty of contemporary references to support them. Amstrad clearly put a lot of stock in the fact that, with one purchase you got what you needed. Sure if you buy a different machine and a separate printer and fished for a compatible monitor and software; you could probably cobble together a system with similar capacity for a lower price. Or if you hunted around for a re-purposed bundle deal you could maybe do the same; if you put up with some glaring disadvantages. But, it is worth noting, if you brought some discrete logic chips, a Z80 and a box, you could get a system for an even cheaper price (and people did). That is not the claim. The claim is that for this price you got a complete system dedicated for the task. Seems true and honest enough to me.

That is it really. But trailing off while picking up on a few of your supporting notes.. Amstrad's offerring was so low-priced and different to other's approaches that they actually refused to sell the monitors or keyboards separately. Plus I am reasonably sure the early contemporary ST and Amigas required an external tuner at extra cost before they could be connected to a TV. Yes, Amstrad had bad ideas after their original PCW. IBM compatible PC's other than those made and priced by Amstrad didn't really matter in the relevant low-price target market in the early 80s UK. The IBM PC landscape was several years behind America in the UK back then. You'd be seriously unlikely to see IBM compatible machines with the sorts of resolutions you quote actually on the ground here, regardless of whether the internet claims they existed or not. And I'd suggest that STs and Amigas mostly did well because they matched the British public's idea of what a computer should look like (before PC dominance). 150.143.30.31 (talk) 10:14, 6 April 2018 (UTC)

Truth of screen mode details?
OK, the single screenshot given here is way too low quality to be able to draw any proper value judgements, but... the in-text claim is of a 720x256 resolution (assuming an 8x8 font, to get a 90x32 result...), at 1bpp (and ??hz/khz). That seems a little low rez when things like the PC had already been using 720x350 to give better readability at 80x25 characters, but, eh, it's possible I suppose - and the narrow pixel aspect wouldn't be unfamiliar vs many other machines that tended to work in the (512x192/560x192/) 640x200 to 640x256 range.

However from footage I've seen of emulated software, particularly games, it certainly seems more like they had an IBM MDA / Hercules / EGA Mono affair going on... IE, not 1bpp, but 2bpp with three actual brightness levels... black, regular green, and high-intensity (and maybe the fourth state representing a hardware generated cursor blink or similar). Though this doesn't seem to be employed with high resolution graphics, which seem to stick to just black and the less intense green. Given that 720x256 at 2bpp would use up a majority of the available memory (45kb instead of 22.5kb), is it possible that there was a lower-resolution 2bpp graphics mode (360x256 still being pretty good by 1986 standards), and maybe an attributed character cell textmode to allow high intensity hilighting at 90x32? Most converted games still seem to occupy a limited size window with static decorative graphics around it, so I don't think there's anything with fewer lines or anything other than 720/360 pixels (unless the ones which look to have large black side borders are actually 640/320 or 512/256 pixels wide of course).

Of course, it could be that all the green is high intensity, and the low intensity is formed simply from a simple 2x1 chequerboard (or even just vertical stripe) dither, and was blurred into a solid colour by compression, but the resolution of the video capture was sufficiently high (1280x720, almost twice the width and almost three times the height, reducing the likelihood of complete chroma blurring - and in any case, green is essentially recorded with full resolution as it's the red and blue difference from luma that are recorded with lower resolution...) and high enough bitrate that this should have become obvious with at least the more static images, plus larger-scale dithering between the three levels was decidedly obvious within the videos...

Also, even 22.5kb is a lot of memory to take out of the total available to the CPU (and moreover, to CP/M), so is it possible that the screen could consume that much OR 45kb, but only be accessed through 16kb banks swapped in to the main 64kb space? In high rez 1bpp or low rez 2bpp, the top (bottom? left, right?) two thirds could be addressed in one go (so, 720/360x170, or 480/240x256, or even a corner measuring 512/256x256 under an exotic addressing scheme), and in high rez 2bpp, one third (720x85, 512x120, 240x256... etc), which would probably be considered sufficient for reasonably smooth updating of a large enough chunk of the screen in a single vertical refresh by the standards of the time...

Have we got a good enough source for the stated mode that we're sure it's 720x256 1bpp only, and nothing more than that, or accompanied by any different variants deliberately provided e.g. for graphing with Logo or a spreadsheet app? I mean, I see one of the well used ones is Old-Computers.com, which is a reliably interesting read, but often a bit wayward when it comes to accurate specs; they list the screen mode as being "720x200 (NTSC)" alongside "720x256 (PAL)", which seems nonsensical for a machine that only uses its built-in screen and has no option to connect either outside sources to it, or to plug in additional external displays (the 50/60hz difference wouldn't be important either - IBM's MDA and the Hercules card both ran at 50hz themselves, even in NTSC areas), and the CPU as "the processor runs at 4mhz, but is slowed to 3.4mhz by the internal clock", whatever THAT'S supposed to mean.

(I mention that mainly because of a query over how much RAM Locoscript actually used, and how much it laid free for a RAMdisc - OC claims 154 + 102, which equals 256 of course... however, if we subtract 22.5 from 154, we get 131.5k... which suggests maybe 128kb is used as working memory - 48kb fixed plus a further 5 banks of 16kb each, which fits nicely with the original intent of building a 128kb machine possibly with a small and entirely separate video memory for a purely character based display a la MDA, or reserving one 16kb bank in the base model for a 640x200 bitmap display or maybe 320x400 portrait - and then 24kb is reserved for the screen (closest quasi-"round" number to 22.5; maybe the characterset has to be included too? The 192 printable ASCII characters, at 8x8, 1bpp, make up a perfectly neat 1.5kb...), with 104kb given over to the RAMdisc, of which 2kb is held for the filesystem and the rest is usable by files...? The "154kb" figure is otherwise a bit hard to explain given the PCW's memory structure... And, well, if that's inaccurate, or at least poorly described, then what else needs to be questioned?) 80.189.129.200 (talk) 19:45, 19 March 2018 (UTC)

-- Wow, that is a big wall of text... When you could have just read the most accurate info we have: i.e. the web site of the person who studied all the published info and reverse engineered the machine to create the emulator. Consider: http://www.seasip.info/Unix/Joyce/ The first link on the first page seems to answer your questions: http://www.seasip.info/Unix/Joyce/hardware.pdf Just Sayin'

You also seem to be working on the assumption that the screen data needs to be paged into the Z80's 64k addressing range while using CP/M. I have no real idea why you'd think this. It's the OS's job to update the screen data, not a CP/M userspace program's job. That's the main point of a real OS (such as CP/M) -to provide an abstraction to run the same program on any machine or display hardware without needing to change the program binary. The PCW actually had the largest amount of free addressing space available for CP/M of any machine I have ever heard of. Much more than the CPC or the popular contemporary commercial CP/M business machines did.

You seem heavy with IBM PC ideas that are not particularly relevant or helpful to you. I am guessing it is hindering your understanding today. "the processor runs at 4mhz, but is slowed to 3.4mhz" means that "the processor runs at 4mhz, but is slowed to 3.4mhz". The PCW's CPU executes instructions slower than the raw clock would suggest because of memory contention. A helpful thing for the article to point out: but quite normal on 80's computers. Amstrad used what we call "M1" timings. Just like the MSX machines did. Both mhz numbers are true. M1 effectively rounds up execution times to multiples of 4 T-States. As an example: a 6 cycle instruction would take 8 cycles in practice, a 7 cycle instruction would take 8 cycles, while a 12 cycle instruction would take 12 cycles, etc. The scheme is actually straight out of Zilog's contemporary documentation. It is far from the best system, the ZX Spectrum did much much better, but it was easier and cheaper for Amstrad to just copy Zilog's reference designs and do what almost everyone else was doing (outside America).

I don't think any changes are necessary to the article. You are just guessing how things might be working under the hood and guessing wrong due to personal bias/experience. But if you think people from your country/background need more context to understand the PCW: I have no issue with it being added. It's not like I own the article. Just trying to be helpful. Perhaps there should be a wikipedia entry on M1 timings since several important Z80 machines used them. But in reallity, CP/M and Z80 stuff always gets poor coverage on the web due to the timing of it's mainstream death in that awkward window just before the web. — Preceding unsigned comment added by 150.143.30.31 (talk) 23:23, 5 April 2018 (UTC)