Was anyone else disappointed with the A1200?

[freehand]

oh no bruce lee has a new word to try and make the PC look bad chunky graphics lol

[Daytona675x]

No, I was not disappointed with my A1200.
The contrary: I loved it. Expanded it to the max over time and it was my primary machine until I required a PC for work (IIRC that was 1998, main reason was that I wanted to use Borland C++ Builder).
Of course the machine was not better anymore than some decent PC, but until then it still had a good price-performance ratio IMO.

Looking back now the hardware compromises become more obvious and painful sometimes.
But that's just because my Amiga motivations changed: back then the first thing I did was putting in some Blizzard and went for good, now I'm trying to squeeze eveything out of a stock A1200 and sometimes find myself swear "why didnt they put some hundred kB of Fast RAM into it" or "how many bugs has this AGA chipset" or "I hate 24bit color register changes in 2 steps" or "still 16bit blitter and 4 channel Paula, WTF"
But on the other hand: that's what makes it fun for me now.

So even now I still love it and am by now way disappointed.

[AestheticDebris]

Originally Posted by Bruce Abbott

But what's the point of chunky graphics if no software uses it? The only 'essential' reason for chunky graphics was textured-mapped 3D games. There weren't many of those around in 1992.

Once you get past 16 colours the scales for how best to design the hardware tip rapidly in favour of just using "chunky" 256 colour representation. It may have "enabled" the creation of 3D texture mapped games, but that certainly wasn't the only advantage to it.

[bluescrn]

Originally Posted by AestheticDebris

Once you get past 16 colours the scales for how best to design the hardware tip rapidly in favour of just using "chunky" 256 colour representation. It may have "enabled" the creation of 3D texture mapped games, but that certainly wasn't the only advantage to it.

A 256-color chunky pixel mode isn't anything magical, it's not really any use without the CPU to update it. And non-upgraded Amigas just weren't fast enough to make good use of it even if it had one. They were much better off with hardware scrolling and fewer colors (for faster blitting).

It would be like a 286 PC with VGA. My parents had one of those, an IBM PS/1, when I was growing up. It was hopeless as a games machine, especially with no soundcard and lacking even a standard slot in which to install one.

Yes, it could run Wolf3D, but it wasn't fast, you had to reduce the window size a lot for it to be playable. And it just couldn't scroll the screen or move sprites around anywhere near as well as an A500.

Kind of a shame that so much Amiga dev talent went into the efforts to imitate Doom on the underpowered machine, rather than make the most of AGA for 2D gaming in those dying years of the Amiga.

[Bruce Abbott]

Originally Posted by AestheticDebris

Once you get past 16 colours the scales for how best to design the hardware tip rapidly in favour of just using "chunky" 256 colour representation. It may have "enabled" the creation of 3D texture mapped games, but that certainly wasn't the only advantage to it.

A 32 color 320x200 screen uses 40kB of RAM. 256 colors needs 64k, 60% more. With bitplanes you can store images with fewer colors in fewer bitplanes to save space. Using bitplanes means you can do dual playfields with independent scrolling etc. Using bitplanes also gives more flexibility in where they can be placed in memory. VGA did its thing by having dedicated video memory that couldn't be used for anything else. The minimum was 256k - half the total RAM of an A500!

From Jay Miner's point of view the best thing about bitplanes was that you could design a nice simple board with 1 bitplane, then just duplicate it to add colors. 2, 4, 8, 16, 32, 64... all without needing special circuitry to suit each color depth. The Amiga already had 6 bitplanes with HAM in 1985 and EHB in 1986, and the OS was designed to handle up to 8 bitplanes so the obvious thing to do was extend the hardware to 8 bitplanes too.

An 8 bit chunky mode could have been added but it would be 'special' and difficult to integrate with the bitplane system. If I was in charge I probably would have asked for that though. "Can't you just throw in a VGA chip with dedicated RAM?", I would ask, "Who cares if the the OS can't use it - we only need it for games". And Amiga fans would rightly howl in disgust. "The new C128!", they would say.

But of course that wouldn't be enough. A faster CPU would also be needed to make the best of it, since the blitter, sprites, and copper wouldn't work with it. We would also need a high density floppy to store the larger image data, which would make it more PC compatible too. We could kill 3 birds with one stone by putting in an Intel CPU and floppy controller. Of course the machine wouldn't be 100% with the A500, so why bother making it compatible at all? We'll make it IBM compatible instead - problem solved! Now the customer can run any PC software on their 'Amiga' and they are happy.

"The all new 'next generation' Amiga DX2-66, with Intel inside. Runs Wolfenstein 3D, Doom and Microsoft Word out of the box! No more problems exchanging files with PCs, because it is a PC!"

[Karlos]

The Amiga ultimately implemented Chunky in the best possible way already. Via aftermarket graphics cards. You get your framebuffer, blitter, and even scalable video overlays.

Too much is made of the lack of chunky pixels. It didn't save the Falcon or the Archimedes.

I don't disagree that they would have been advantageous, but only had they have had proper support for 2D operations too.

A bigger disappointment was the Akiko solution to the problem. The fact you have to write spans of 32 chunky pixels to it, then read them back, all via the CPU, before writing them back to planar memory again with the CPU. What a waste of bus bandwidth.

[Thomas Richter]

Originally Posted by Bruce Abbott

Using bitplanes also gives more flexibility in where they can be placed in memory. VGA did its thing by having dedicated video memory that couldn't be used for anything else. The minimum was 256k - half the total RAM of an A500!

Err, that is due to the minimum video apperture size, and you would not need much less memory on an A500 if you would want to display 640x480 in 16 colors, because this is what VGA could do.

Originally Posted by Bruce Abbott

From Jay Miner's point of view the best thing about bitplanes was that you could design a nice simple board with 1 bitplane, then just duplicate it to add colors. 2, 4, 8, 16, 32, 64... all without needing special circuitry to suit each color depth.

That was surely a good argument in 1989, but it was no longer a good argument when memory became cheaper. The PC evolved, the Amiga stuck, and that was also its demise. The problem with bitplanes is that 8 bitplanes requires 8 blits (in most cases, at least), whereas chunky displays require only a single blit. It is thus inheritly slower.

Originally Posted by Bruce Abbott

An 8 bit chunky mode could have been added but it would be 'special' and difficult to integrate with the bitplane system.

Sure, and CBM failed to make the investment early enough.

Originally Posted by Bruce Abbott

If I was in charge I probably would have asked for that though. "Can't you just throw in a VGA chip with dedicated RAM?", I would ask, "Who cares if the the OS can't use it - we only need it for games".

If CBM would have made the investment, the Os would be able to do it. That is of course exactly what happened. Os extensions can do it today. Re-interpreting 8 bitplanes as chunky instead of planar by only making changes to Denise rather than to the DMA controller is not rocket science either.

Originally Posted by Bruce Abbott

But of course that wouldn't be enough...

That's really childish. By this argument, you would rule out any type of extension right away.

[TEG]

Originally Posted by CCCP alert

C128 was another joke-tastic idea from Commodore. You had to spend £800 (C128+1571+1902) on a machine to run crusty old pathetic CP/M programs on the Commodore 190x monitors OR spend £500 on an actual 512k IBM PC compatible PC from Amstrad including monitor. Erm.....

It was a bad idea on every front (no 2mhz CPU for 40 column regular Joe TV mode AKA C64 style games and zero access to the Z80 if you choose not to boot into CP/M but you paid for a complex motherboard with twin incompatible processors inside it). In fact it is no more useful in real life than a Commodore 64 and a 64kb RAM expansion to be brutally honest. CP/M was a duff idea even in 1986, if it wasn't then Amstrad would have conquered the small business/professional market with a CP/M not PC compatible

No, this is the words of someone who hasn't owned a C128.

The C128 7.0 basic was vastly superior to the C64 2.0 basic. And at the time, the basic was a thing. It was what made the machine accessible without having to buy a tierce software. Among that it had a sprites editor and an assembler monitor built in ROM. It was just incredible for the time.

The other very important point was its nice design. A Porsche compared to the Soviet union look of the C64. And to stay in the car analogy, the transmission was now effective with the debugged fast floppy drive connexion. It was a machine worth the price, especially of course due to the 99,999% perfect C64 compatibility.

[TEG]

Originally Posted by Bruce Abbott

PCs did indeed take off because of the name IBM, but that was back in the early 80's. By the time the Amiga arrived the PC was firmly established as the preferred platform. After the Amiga was released the PC was attracting just as many games and many more apps - even before it had a powerful 386 or 486 CPU, 3.5" HD floppy, SVGA graphics and a sound card. The PC had won before the Amiga had even left the starting gate.

I tend to agree with this one. Only tend because I think a trend, especially in the tech industry, can be changed and quickly.

I think that, at Amiga time, it was still possible because of the superiority of the machine but with some ajustements and very good marketing. The Amiga should have the ChatGPT of the time. If you show people how they will make money and have a clear advantage owning your product, everyone wanted it. But you have to imagine the futur and Commodore management was clearly not good at that.

Clearly the PC started to widespread during the 8 bits area. I remember going to fairs and their was already those big ugly boxes and it was at C64/Spectrum time.

When you were a professional there was really no alternative. Either you had at disposal the big serious ugly boxes running on green screen or the nice personal toys computers running on the television but with the advantage of music and colours.

On Wikipedia we can read:

"IBM decided in 1980 to market a low-cost single-user computer as quickly as possible. On August 12, 1981, the first IBM PC went on sale. There were three operating systems (OS) available for it. The least expensive and most popular was PC DOS made by Microsoft. In a crucial concession, IBM's agreement allowed Microsoft to sell its own version, MS-DOS, for non-IBM computers. The only component of the original PC architecture exclusive to IBM was the BIOS (Basic Input/Output System)."

I think it made a real difference yeah. Did another company which was building computers proposed its OS to others constructors? For example the PET is from 1977. Perhaps if Commodore would have licence and polish, adapt and market the PET "OS" it would have widespread on a lot of platforms.

I think IBM, due to its history, had the vision of how the market could be built and it made the difference to dominate.

[abu_the_monkey]

Originally Posted by Karlos

A bigger disappointment was the Akiko solution to the problem. The fact you have to write spans of 32 chunky pixels to it, then read them back, all via the CPU, before writing them back to planar memory again with the CPU. What a waste of bus bandwidth.

while I agree it is not the most efficient way to do c2p (DMA would be better), it was not the primary function of the Akiko (which people forget) it is a glue logic chip that handles many functions and reduced the number of ic's required. the c2p part was thrown in as an extra at the last minute as there was just enough space on the die, or so I read in an interview with Dave Haynie back in the day.

[Bruce Abbott]

Originally Posted by Thomas Richter

Err, that is due to the minimum video apperture size, and you would not need much less memory on an A500 if you would want to display 640x480 in 16 colors, because this is what VGA could do.

256k was needed for chunky pixels. The width of a single lores VGA pixel is ~80ns, but the original VGA card only had 120ns DRAM so it wasn't possible to read the pixels fast enough. The method used to get around this was to have 4 banks of 8 bits in parallel. Though from a programmer's perspective the memory is 64k x 8 bits, internally it is read 32 bits at a time and then the 4 pixels (one per byte) output one after the other.

Using 64kx4 DRAM chips you need 8 chips, which adds up to 256k bytes. Using 16kx4 chips (if you could still get them) gave you 64k, just enough for one 320x200 256 color screen but not enough for 320x240. It would be MCGA, not VGA.

Later VGA chips used fast page mode to read the RAM faster, as well as faster chips. So the Trident TVGA9000i for example could work with just two 256kx4 DRAM chips.

Commodore could have used a customized VGA controller that needed less RAM, but you would still want at least 128k for double buffering (more for PAL resolutions). 256k was the standard though so it made sense to have it. Either way that was extra RAM on top of the Amiga's ChipRAM, raising the price of the machine.

Of course there are other ways to get chunky pixels. Superhires has 160 bytes per line. Instead of serializing the individual bits, each byte could be a single pixel. This would work well for 3D games like Wolf3D and Doom in 2x1 mode (since the lower resolution is needed anyway to get a good frame rate). For 1x1 mode you could use 2 bitplanes and interleave the pixels.

However as the general computing requirements go up a faster CPU is required anyway, to the point where the overhead of C2P is the same as direct copy to ChipRAM. To get even more speed you need a latch which the CPU can write to quickly and then do other stuff while the pixels are being written to ChipRAM. This what Akiko C2P should have been (CPU writes to it once, it takes care of the rest).

That was surely a good argument in 1989, but it was no longer a good argument when memory became cheaper.

Memory still wasn't that cheap in 1992. In fact for a time the price actually went up, partly due to the use of SIMM modules (which cost more than bare RAM chips).

The PC evolved, the Amiga stuck, and that was also its demise.

No, the Amiga's demise was due to Commodore's demise. And that was largely due to underestimating the worth of the A500 while overestimating the ability of the engineers to come up with a replacement. The Amiga wasn't exactly 'stuck', just not rolling as smoothly as it should. In the book 'Commodore The Final Years', Brian Bagnall explains how they stopped production of the A500 because the new AGA models were supposed to be imminent. But that was a mistake. they should have kept pushing the A500 until sales dropped to nothing, like they did with the C64. They also should have shown us what they were doing with AGA, so customers would line up for it instead of thinking nothing was coming.

The problem with bitplanes is that 8 bitplanes requires 8 blits (in most cases, at least), whereas chunky displays require only a single blit. It is thus inheritly slower.

You never heard of interleaved bitplanes? The advantage of chunky is that there is less horizontal 'overhang' on non-aligned blits, which reduces the number of cycles needed. However the overhead gets proportionally smaller as the size increases, so it doesn't make much difference for large blits (the ones that generally use up the most time).

If CBM would have made the investment, the Os would be able to do it. That is of course exactly what happened. Os extensions can do it today. Re-interpreting 8 bitplanes as chunky instead of planar by only making changes to Denise rather than to the DMA controller is not rocket science either.

I agree. I would like to see somebody do that, and since it isn't rocket science it should be here any day now... Imagine being able to pull the Denise chip out of your OCS/ECS Amiga and pop in a replacement that did 256 color chunky. I would buy that! Make it an FPGA that I can reprogram and I'll pay double!

You say 'If Commodore had made the investment'. Well they did make investments but unfortunately not all in the right places. The A3000 and CDTV were money losers even though they provided a base for more successful models. Had they been more financial to start with this might have been OK, but since Jack left they were always struggling. So they couldn't afford to try breaking into new markets with risky prospects. Unix bombed - interactive TV bombed, only the A500 carried them through. AAA bombed too. If they had set their sights lower and looked after the 'low-end' market (where most fans were) better, they might have had more ability to invest (and pay off that XOR troll!).

That's really childish. By this argument, you would rule out any type of extension right away.

Sooner or later the Amiga was going to hit the wall of IBM compatibility, so why wait? When Windows 95 came out I knew it was game over. The one advantage the Amiga still had (apart from price) over PCs - a properly multitasking OS - and then it didn't. Of course the rest of the World didn't even notice since they had always been IBM anyway.

I am glad Commodore never managed to get next generation 'Amigas' out, because they wouldn't be Amigas. All Amigas are pretty much compatible with each other if you don't count PPC and OS4 (which I don't). This make life easier for the retro community and makes owning an Amiga (any Amiga) more worthwhile. What disappointed some people about the A1200 - that it wasn't radically different from previous models - is a good thing IMO.

30 years later surprising things have happened. Turns out an A1200 with FastRAM runs Wolf3D fine, and with an 030 it runs Doom fine. With a fast 060 it can handle pretty much anything that a Pentium system of the day could. 2.5" hard drives got cheaper and larger. The PCMCIA slot became quite useful. Most of the complaints about the A1200 were addressed a long time ago.

[grond]

Originally Posted by TEG

Did another company which was building computers proposed its OS to others constructors? For example the PET is from 1977. Perhaps if Commodore would have licence and polish, adapt and market the PET "OS" it would have widespread on a lot of platforms.

Isn't that what actually happened with at least many of the Commodore 8-bit computers? The BASIC interpreter was the OS of those computers and it was a Microsoft product (which is why e.g. the C64 BASIC was hardly adapted to the underlying hardware and you had to use lots of POKEy stuff to make the hardware do what you wanted).

[Thomas Richter]

Originally Posted by Bruce Abbott

256k was needed for chunky pixels.

No, and VGA does not even have chunky pixels, originally. You need 150K for 640x480x4, do the math yourself. The next power of 2 that fits is 256.


VGA has four bitplanes, each 640x480 max, only one of them visible at a time to the CPU through the "graphics controller", which allowed very simple operations on the plane. Chunky came later with the "chain 4" trick that just re-interpreted the bitplanes.

Originally Posted by Bruce Abbott

The width of a single lores VGA pixel is ~80ns, but the original VGA card only had 120ns DRAM so it wasn't possible to read the pixels fast enough. The method used to get around this was to have 4 banks of 8 bits in parallel.

VGA *is* planar, and from the programmer's perspective, it's one bitplane in a 64K window. Chain-4 came later, and that's the same trick CBM could have been using. It's just taking planar data, and through addressing tricks and linkage of two nibbles from a set of bitlanes, you get the illusion of chunky.

Instead CBM carried on with stupid planar with AGA.

Originally Posted by Bruce Abbott

However as the general computing requirements go up a faster CPU is required anyway, to the point where the overhead of C2P is the same as direct copy to ChipRAM.

Hardly. With C2P, you need to do the work twice. Render chunky, push into chip. With chunky, you do the work once. Write to chip. Done. You can only win by dropping one operation.

Originally Posted by Bruce Abbott

In the book 'Commodore The Final Years', Brian Bagnall explains how they stopped production of the A500 because the new AGA models were supposed to be imminent. But that was a mistake.

How to continue selling an outdated machine in an innovative market could be a mistake is a miracle to me.

Originally Posted by Bruce Abbott

You never heard of interleaved bitplanes?

That does not help you for off-screen to screen blits and with the line drawer. There are only few places where you can take advantage of it - basically, screen to screen blits without a mask. Otherwise, the modulo does not add up.

Originally Posted by Bruce Abbott

You say 'If Commodore had made the investment'. Well they did make investments but unfortunately not all in the right places. The A3000 and CDTV were money losers even though they provided a base for more successful models.

The result of stupid decisions, and the "read my lips" mandate not to put in new chips. The A3000 with AGA would be a substantial improvement. When I saw the A3000 at the CeBit in Germany, I wondered "why would anyone want to buy that? Same shit, more expensive."

Originally Posted by Bruce Abbott

Sooner or later the Amiga was going to hit the wall of IBM compatibility, so why wait?

Wait? No, not "wait", but innovate. Instead, CBM management blocked innovation. See the A3000.

Originally Posted by Bruce Abbott

When Windows 95 came out I knew it was game over. The one advantage the Amiga still had (apart from price) over PCs - a properly multitasking OS - and then it didn't. Of course the rest of the World didn't even notice since they had always been IBM anyway.

At that time, the Amiga still had a standing as multimedia/TV processing station. I've still seen Amigas around at the trade shows at that time, genlock, video overlay, all things Amiga could do nicely. CBM didn't care.

Originally Posted by Bruce Abbott

What disappointed some people about the A1200 - that it wasn't radically different from previous models - is a good thing IMO.

The bad thing is that it was two years too late, at least.

[grond]

Originally Posted by Bruce Abbott

Sooner or later the Amiga was going to hit the wall of IBM compatibility, so why wait? When Windows 95 came out I knew it was game over. The one advantage the Amiga still had (apart from price) over PCs - a properly multitasking OS - and then it didn't. Of course the rest of the World didn't even notice since they had always been IBM anyway.

The rest of the world simply didn't understand this at all. I remember arguing with a few PC guys I met when I did civil service. They said that they couldn't use more than one program at a time anyway, so what good was there in having multitasking. Then Windows 95 came out and they understood. But then it didn't matter any more since they finally got multitasking, too.

[aros-sg]

Originally Posted by Thomas Richter

That does not help you for off-screen to screen blits

If both screen and off-screen bitmaps are interleaved, it of course helps/works-in-a-single blit there too.

screen to screen blits without a mask.

It works also with the mask. Many many hardware banging games do that. But downside is of course that 1 single plane for mask is no longer enough. It works also in the OS if you do your own multi-step-mask-blits based on BltBitMapRastPort() with multi-plane mask bitmap (instead of BltMaskBitMapRastPort()).

[Bruce Abbott]

Originally Posted by Thomas Richter

No, and VGA does not even have chunky pixels, originally. You need 150K for 640x480x4, do the math yourself. The next power of 2 that fits is 256.

Yes, it does.

VGA has four bitplanes, each 640x480 max, only one of them visible at a time to the CPU through the "graphics controller", which allowed very simple operations on the plane. Chunky came later with the "chain 4" trick that just re-interpreted the bitplanes.

That's SVGA. Standard VGA (and MCGA) has chunky in mode 13h, at 320x200 256 colors. The vast majority of PC VGA games used this mode.

While 64k RAM is enough for a single 320x200 256 color screen, to get smooth animation you must double buffer which requires 128k. If you want 'square' pixels then you must run 320x240 which needs more than 64k per screen, for a total of over 128k with double buffering. The next size up is 256k!

VGA *is* planar

Not in 320x200 with 256 colors it isn't. You should know that!

Instead CBM carried on with stupid planar with AGA.

Like I said, extending from 6 to 8 bitplanes was the natural and obvious thing to do. Chunky would cause a lot of problems (as you probably know from working on RTG drivers) and the hardware design would be tricky too.

IBM had a much easier job with VGA because they didn't have to worry about maintaining compatibility with an existing blitter, sprites, copper, dual playfields, disk and audio DMA, as well as executing CPU instructions from video RAM. With dedicated memory fragmentation wasn't an issue either, so it didn't matter if there were large gaps between bitmaps. Furthermore since the PC wasn't multitasking and programs didn't have dedicated 'custom' screens constantly in video memory which could be 'pulled down' over each other, it didn't have to swap screen modes and palettes on the fly. Nor did it have to do overscan and genlocking.

I suspect the complications of supporting all this stuff was the main reason AAA never got finished. AA (renamed AGA on launch) was designed as a less ambitious extension to the existing chipset. It was developed in parallel with AAA when it looked like AAA might not be ready soon enough. Unfortunately after the AA design was completed the engineers decided they could do more with it, which further delayed its introduction. This was good technically, but bad for marketing. It might have been better to release a machine with the earlier version first, even if just to developers so the (Amiga) world would know about it.

How to continue selling an outdated machine in an innovative market could be a mistake is a miracle to me.

The C64 was released in 1982 and still selling in 1994. The Sega Mega Drive was released in 1989 and still selling in 1997. There was a 6 year gap between introduction of the Sony PlayStation 1 and 2.

What you call an 'innovative market', I call forcing customers to constantly upgrade. That was great for hardware vendors, not so great for users. But it wouldn't work in a smaller market where people would just hop to the vastly more popular PC instead. The strength of the Amiga was in consolidating what it already had, not what it could get by 'innovating'.

The result of stupid decisions, and the "read my lips" mandate not to put in new chips.

I'm sick of hearing that meme. The truth was a lot more nuanced than it implies.

The A3000 with AGA would be a substantial improvement.

Sure it would, in a perfect world that didn't exist.

When I saw the A3000 at the CeBit in Germany, I wondered "why would anyone want to buy that? Same shit, more expensive."

If you already had an A2000 then you could just plug in an accelerator card, hard drive controller, flicker fixer etc. for less. However those of us who didn't have an A2000 (and wanted a smaller form factor) could do the sums and see that it wasn't that much more expensive. Furthermore you got 2MB ChipRAM with 32 bit CPU access, a 32 bit Zorro-III expansion bus, and a 32 bit CPU slot for future expansion, as well as a high performance DMA SCSI hard drive controller (the official Commodore controller at the time wasn't great, and most 3rd party controllers were PIO). And you got the latest OS. You also got a machine ready to go out of the box, without any of the issues you might having upgrading an A2000.

It wasn't 'the same shit'. Sure it didn't have AGA, but so what? No other Amiga had it either, while cards providing similar functionality were already appearing for the A2000. Why would you want AGA when you could have a much higher performance Zorro-III graphics card? IMO the only 'shit' thing about the A3000 was ECS productivity mode, which was redundant because the A3000 had a flicker fixer built in (a much better one than the AGA-2000 too!).

At that time, the Amiga still had a standing as multimedia/TV processing station. I've still seen Amigas around at the trade shows at that time, genlock, video overlay, all things Amiga could do nicely. CBM didn't care.

A niche market. And CBM did care - otherwise they would have dropped genlocking and the video slot, just like VGA cards dropped the feature connector.

The A4000 was very popular for video production, especially when Newtek bundled it with their Video Toaster. The A1200 was also useful, with AGA providing the graphics needed for high quality video titling etc. A local TV station here used one for many years.

The bad thing is that it was two years too late, at least.

Yes, being late was a bad thing. What do you know the real world isn't perfect.

But this doesn't mean the A1200 itself was bad. My A1200 hasn't suffered at all just because I couldn't buy one in 1990. In fact it's better, because in 1990 they would have produced something else. The A1200 was based the A600's wedge-shaped design (which I love). Without the A600 we wouldn't have gotten the A1200. What would they done in 1990? Without full smd the motherboard would be much larger, requiring a bigger case than the A500 (which was already a bit big for my tastes) or a 'big box' design like the unwieldy A2000 or tightly packed A3000. This would make it a lot more expensive and harder to sell.

[Promilus]

256-Color Graphics Mode (Mode Hex 13)
This mode provides graphics with the capability of displaying 256
colors at one time.
The display buffer is sequential, starts at address hex A0000, and
is 64,000 bytes long. The first byte contains the color information
for the upper-left PEL. The second byte contains the second PEL,
and so on, for 64,000 PELs (320 x 200). The bit image data is
stored in all four memory maps and comprises four bit planes.
The four bit planes are sampled twice to produce eight bit-plane
values that address the video DAC.

Hmmm....
http://bitsavers.trailing-edge.com/p...nual_May92.pdf

[Thomas Richter]

Originally Posted by Bruce Abbott

Yes, it does.

That's SVGA. Standard VGA (and MCGA) has chunky in mode 13h, at 320x200 256 colors. The vast majority of PC VGA games used this mode.

No, again, VGA, *not* SVGA. The chain-4 mode is a VGA invention to turn the planar VGA system into a chunky system. And please, excuse me, I'm the one writing drivers for these chips.


SVGA added hi-color and true-color extensions.

Originally Posted by Bruce Abbott

Like I said, extending from 6 to 8 bitplanes was the natural and obvious thing to do. Chunky would cause a lot of problems (as you probably know from working on RTG drivers) and the hardware design would be tricky too.

The *hardware* is the easier part - re-interpreting planar as chunky is something VGA already did. Line drawing with the blitter does not work, but the Os integration would be a larger investment.

Originally Posted by Bruce Abbott

IBM had a much easier job with VGA because they didn't have to worry about maintaining compatibility with an existing blitter, sprites, copper, dual playfields, disk and audio DMA, as well as executing CPU instructions from video RAM.

The copper and sprites are not issues you have with a chunky extension either. But more important, IBM did not have to bother with Os compatibility because there was none. Back then, applications only selected the mode through the video bios and then banged the screen directly.

Originally Posted by Bruce Abbott

I suspect the complications of supporting all this stuff was the main reason AAA never got finished. AA (renamed AGA on launch) was designed as a less ambitious extension to the existing chipset.

The problem was (and still is) that CBM had to fight with a legacy Os, and an ill-designed graphics subsystem that was more or less bolted on the chipset design.

Originally Posted by Bruce Abbott

What you call an 'innovative market', I call forcing customers to constantly upgrade. That was great for hardware vendors, not so great for users.

Sure. And yet, it helped a lot to develop the market.

Originally Posted by Bruce Abbott

The strength of the Amiga was in consolidating what it already had, not what it could get by 'innovating'.

Only if you call "standstil" "consolidation". That's really an abuse of words.

Originally Posted by Bruce Abbott

If you already had an A2000 then you could just plug in an accelerator card, hard drive controller, flicker fixer etc. for less.

Precisely, and this is precisely why I never bought an A3000 or an A4000 because at the time CBM had these machines in the market, my A2000 could already do better with a graphics card, and an accelerator card. No wonder CBM could not sell these machines - they were not innovative. it was just the same shit, mildly recycled. Cleary, CBM run out of money because there was no need to buy their "consolidated" machines.

Originally Posted by Bruce Abbott

It wasn't 'the same shit'. Sure it didn't have AGA, but so what?

Why would anyone buy a new machine from CBM if it did not deliver more? It delivered less than what you got by third-party extensions.

Originally Posted by Bruce Abbott

No other Amiga had it either, while cards providing similar functionality were already appearing for the A2000. Why would you want AGA when you could have a much higher performance Zorro-III graphics card?

Yes, why would you? AGA did not deliver anything competative at the time it appeared. If it would have appeared with the A3000, it would have delivered something, and had created some demand for the machine. Just management made a stupid decision not investing into the machine.

Originally Posted by Bruce Abbott

But this doesn't mean the A1200 itself was bad.

It wasn't good enough either. it delivered too little, too late. By that time, I had my GVP 030 board and a graphics card. Why buy anything from CBM for that which had less RAM, and less powerful graphics.

[Turrican_3]

Commodore chairman Irving Gould saw a better opportunity when I got fired, and we met for lunch during the summer of 1990 at the Four Seasons restaurant in New York City. I was Interested because, among legacy products such as the PET, VIC-20 and C64, Commodore had a more modern machine called the Amiga which deserved the kind of treatment we applied to the 1984 Mac. In that meeting Gould helpfully confirmed everything negative I had heard about him and made my decision easy. But as we'll see later at Be I remembered the Amiga.

[...]

In our ensuing conversations, the Commodore Amiga became our "reference platform," an example of the good, the bad, and the promising. While the Amiga provided interesting multimedia features, we were less impressed by its overly complicated architecture and cobbled-together implementation. But the most important aspect to us was that it sold in attractively large numbers, proving there was a market for a third way.

Quotes from the Jean Louis Gassee book citing Commodore and Amiga. Perhaps there are even more references but I haven't found them yet.

[TEG]

Originally Posted by grond

Isn't that what actually happened with at least many of the Commodore 8-bit computers? The BASIC interpreter was the OS of those computers and it was a Microsoft product (which is why e.g. the C64 BASIC was hardly adapted to the underlying hardware and you had to use lots of POKEy stuff to make the hardware do what you wanted).

Well, BASIC was a software not an OS but it's true that on 8 bits computers the boundary was blurry. Anyway it was not meant to be a disk operative system. There was CP/M (1974) but it was for Intel 8080/8085 and Z80/Z8000 (derivative from the 8080).

The PET and the Apple II are 1977. Apple DOS was released in 1978 and Commodore DOS in 1979. MS-DOS and the PC are 1981 so they appeared 3 years later.

So why the PET or the Apple II did not flooded the market after all, like it will be the case with the IBM PC those 3 years later? Especially the Apple II, having in 1978, disk drive, expansion slots, colours and an OS.

Although the Apple II sold well from the launch, the initial market was to hobbyists and computer enthusiasts. Sales expanded exponentially into the business and professional market, when the spreadsheet program VisiCalc was launched in mid-1979. VisiCalc is credited as the defining killer app in the microcomputer industry.
https://en.wikipedia.org/wiki/Apple_II

I guess the marketing capacity played a crucial role and the size of the company behind the machine too. Apple was a startup after all. Or there was others factors? I don't know what was the reputation of the CBM brand at the time.

YEAR PC           Apple II
1977 0             600     
1978 0             7,600     
1979 0             35,000     
1980 0             78,000     
1981 35,000        210,000 
1982 240,000       279,000 
1983 1,300,000     420,000

We can see it's only in 1983 that the PC sales overtook the Apple II ones. So 6 years after the Apple machine was launched. Was it impossible to keep the lead in front of IBM or there was some errors made too? It's interesting to compare this part of the history in regards of the Amiga one.