Zarchos wrote:I am surprised nobody (or only a few people) refer to the Acorn machines in this thread.
After all they are what everybody was looking for :
The Archimedes does interest me quite a lot! My mate grew up with a BBC Micro, and later, an Archimedes, and learned ARM assembly on the system. So, I've learned a lot about the machine's internals from him. Including the astounding memory bandwidth. I can hardly believe it could do 32MB/s in 1987, but he wrote a benchmark and confirmed it. Although the totals were a bit lower, and dependent on the screen mode in use (since it's essentially chip RAM). Compare that to the ST's 4MB/s, and the Amiga 500's peak of 8MB/s using the blitter with all bitplanes turned off.
The ARM instruction set looks quite interesting too, with a barrel shifter usable within certain instructions. Ideal for fixed point memory addressing like audio resampling and image scaling.
I was also very surprised to see it could drive 640x256 in 256 colours, and 640x512 non-interlaced in 16 colours (Is this correct? That's the best information I've been able to find so far). The A400 series could even handle 1152x896 in non-interlaced mono, making it one of the first consumer-level computers to handle a high resolution (3 years before the TT030). The IBM 8514 card could handle 1024x768 in 256 colours in the same year, which sounds impressive, until you realise it had to interlace the display to achieve that. The IBM card also had extremely little software support and probably cost about the same as a complete Archimedes by itself (let alone the house-priced PC it goes in).
The Archimedes did seem to have some technical shortcomings compared to the Amiga. The 8 sound channels unfortunately all share the same sample rate, so you need to resample in software - partly defeating the point of having 8 channels. I guess Acorn didn't really "get" sample based music. I imagine it would only take a couple of percent CPU to mix and resample four channels in software though. Another shortcoming was Acorn's lack of care about gaming, hence there is no hardware scrolling. Let alone parallax scrolling. But the memory bandwidth is so high you can just brute-force it in software. Apparently it gets a bit marginal with parallax scrolling, but it can be done.
I suppose in terms of potential memory bandwidth, a 68030 at 24MHz would provide the same memory performance as an ARM at 8MHz. But I think most 68030 systems at the time had severely under-specced memory and relied on the cache.
From what my mate has told me, almost no software ever tried to redefine the palette in 256 colour mode, because of the complex interactions that occurred. It doesn't really offer you 16 shades of each of the 16 colours as you might hope. Instead, redefining the palette seems to only offer a way to slightly re-shuffle the bit allocations in the 8 bit truecolour mode. For example, you could set up rrggbbtt (Archimedes default), or rrrgggbb (MSX2) by redefining the palette. It's basically an 8 bit truecolour mode, rather than an indexed mode. Not quite as good as 256 definable colours, but it's good enough. Of course they went to 256 definable colours later, but that was well after VGA AGA and Falcon came out.
I think for a typical home machine the Archimedes was probably still too expensive. According to some research I did a while ago, the Archimedes A3010 was the equivalent of $899, versus $799 for the already too-expensive Falcon, and $599 for the fairly-popular Amiga 1200. There is a threshold beyond which a typical home user won't pay, regardless of specs or software. I fear that the Archimedes was above that threshold.
A huge problem was "Little Britain Syndrome." Acorn seemed only interested in selling to the UK, and even worse, mainly only the schools. Since they not only invented a completely incompatible architecture, but also a completely incompatible CPU, that made porting far more difficult. Even more so considering the most preferred high level language on the platform seemed to be BASIC rather than C. Not selling the machine overseas pretty much guarantees eventual doom, since Britain alone isn't big enough to be self-sustaining for computing for very long.
This has a serious knock-on effect for the consumer, since fewer users means less software. The target market being education did mean there was some useful productivity software, but it was a niche nonetheless. It seems that my mate had to program a lot of his own software, while I never felt the same need on the ST or Amiga. That said, I'm surprised there was as much software as there is for the platform given the limited audience.
I can't ever remember seeing an Archimedes at any computer store in the UK, throughout the 80s and 90s. Tons of STs and Amigas, and by the early 90s, also a few PCs and I even saw a Mac clone once. But the only Archimedes I ever saw was during my final years at school, when the school got a single Archimedes for the whole school (amongst a sea of BBC micros). We also got a PC around the same time, and there was much more software on the PC than the Archimedes had available in our school. And even more on the BBCs.
As a result of that, I can't ever consider the Archimedes a competitor in the home computer market, even if it was technically impressive. It could have been huge, if Acorn had thought bigger.
Zarchos wrote:I won't even talk about PACE technology, who bought the rights of RISC OS, used it in its set top box, and made hundreds millions with it.
I don't really know much about set top boxes, but I believe Amstrad were/are a major player. I suppose Commodore made a crack at it with the CDTV. I think Gateway were going to do something with it too.
joska wrote:Here in Norway you'd mainly find two types of computers in people's homes in the late 80's and early 90's - PC's and Commodore 64.
That's quite a different situation to the UK, from what I recall. The ST and Amiga were seen as credible platforms for getting work done in the home, and it wasn't unheard of for businesses to use them. I believe in Germany, the ST was a mainstream business computer - and even used for industrial machine control instead of PCs. There was a hardened version of the ST for that purpose.
I suppose it didn't help that schools all used BBC micros, and later Archimedes systems. As a result, nobody received any training on the extremely difficult to use DOS software like Lotus and Wordperfect. So naturally, people chose the easy to use platforms with mice, with WYSIWYG. Not the hostile text-only green screen monster that costs more than their car, and requires you to memorize 50,000 keyboard shortcuts. Not to mention having to understand high memory areas, MDA/CGA/EGA/VGA, EMS cards, XMS, EMM386, MMUs, IRQs, DMAs, config.sys entries, alternative boot configurations, DOS commands, PKZIP command line options, and much more. That's not quality software. It's just a polished turd, leaking diarrhea all over your keyboard. It makes me projectile vomit just thinking about those days on the PC. *shudder*
By that measure, I would say software for PCs at the time was vastly, hugely inferior to ST/Amiga/Archimedes software. It didn't really matter if some PC software supported particular business feature X. That was moot, when it was far too difficult to use. The PC needed Windows 3.1 to make it accessible to a normal person. Not surprisingly, sales of PCs boomed here after 3.1 came out.
joska wrote:Some people here seem to think that hardware specs sells computers. It doesn't.
I think it does, but not directly. People didn't choose their machines based on reading the specs. They did choose based on price alone in many cases - which is why so many gamers ended up with STs instead of Amigas.
Where hardware specs do come into the question is with cost of manufacture, and with the software available for the platform. Many gamers could see that quite a few games looked better on the Amiga than the ST, and hence bought the Amiga instead (if they could afford it). Or if you were doing art or video, THE machine to have was the Amiga. Or music, the ST. Those choices were all based indirectly on hardware capabilities.
In terms of gaming, the killer app for the PC was Doom. Doom was only made possible directly by hardware specs - you needed a very fast 386 or 486 to get it to run well, and you needed a chunky display. Any computer with a planar display or a slower processor simply couldn't deliver what people were looking for. So people started buying computers with "doom hardware."
Balancing cost against killer hardware features is key for home computers. But businesses in the 80s and early 90s seemed to be the exception, for some bizarre reason. They would buy heap-of-crap IBMs, with very little power, for a huge sum of money. And then have to spend even more money on training to overcome the extremely hostile usability of the software. I'm glad to see that some countries like Germany could see beyond this madness.
But I feel that home computers are more interesting than discussing business computers, and it's a much higher volume market. Not necessarily more lucrative (due to smaller margins) but more units sold.
Atari had the chance with the STE to eat away at some of the market for the Amiga, and the new 16 bit consoles. But due to a series of mistakes, some of which were technical, that couldn't happen. If the STE had been pumping out graphics better than a SNES, I'm fairly sure it would have been a much bigger hit.