View Single Post
Old 27 July 2019, 13:31   #677
Registered User

Join Date: Jul 2015
Location: The Netherlands
Posts: 1,301
Originally Posted by grond View Post
From my point of view all our discussion was about how in detail Commodore presented too little of an upgrade with the A1200. Looking at the details required looking at what they could have and should have done instead and what technical and ecnomical effects this would have had.
Well, that's not how I looked at our discussion about Doom. It does explains why you keep talking about machines that were not available. Anyway, my problem with such things is that it's impossible to know the effects of unreleased machines on the market. It's pretty clear you believe they would've changed Commodore's (and hence the Amiga's) fortunes for the better.

I personally don't think so.

However (and this is the important bit): there is no way to know for certain either way. Personally, I think this is part of why we keep going around in circles. As long as we keep talking about things that, at best, we can only guess about... I don't see how we'll ever reach any agreement. We both think very differently about this and are talking about stuff that's (sorry) hypothetical - not reality.
I'd be interested in examples just out of historical interest. The only 3D games on the Amiga I ever liked were Alien Breed 3D (the first part, the second was unplayable) and Ambermoon.
I'd guess the biggest one in terms of sales would be Microprose's Grand Prix and Elite 2: Frontier. But I'd personally also include the many 3D driving games, games such as Midwinter and Zeewolf, Flight Simulators, etc. Even Epic (though it was far too short) would've done much better on faster Amiga's. And of course there's, Wing Commander. Almost all of those games ran very poorly on a base A500 and much better on an accelerated Amiga's with Fast Ram.

Many of these were also (as far as I can tell) really popular at the time. Just look at the hype (in magazines) that some of these games received.

And yet, almost no one bought a faster Amiga to play such games better. Meanwhile, in DOS land it seems that people were buying faster PC's and better graphics cards long before Doom. Now, I can't be certain that this was because of those games, but I can remember the hype around Wing Commander and articles talking about what kind of PC to get to run it well. AFAIK the same just didn't happen with the Amiga.

In fact... IIRC, when Wing Commander came out for the A500, people complained about the frame rate. Reviews didn't point out it ran much better if you bought an accelerator, instead marking it down for poor performance.
Unless you are merely copying tiles together and only using sprites for moving objects, you will have to use quite a few cookie-cut BOBs for a game such as Warcraft. With interleaved bitmaps the mask needs to have the same size as all pixel data, with non-interleaved bitmaps you can reuse the mask for all bitplanes. Bandwidth remains the same, but memory impact is less for non-interleaved bitmaps.
Right, no bandwidth difference but you use more memory.
BTW, such cookie-cut BOBs also require reading the entire background which means a bandwidth penalty when compared to chunky. Chunky graphics can save memory bandwidth here because you can just skip writing to the graphics buffer if you have read a transparent (i.e. zero value) pixel and thus save bandwidth while writing and all bandwidth for reading the background image.
True, a chunky display can save up to 1/4th of the bandwidth for cookie cut bobs in the optimal case. That's not the whole story though. If you really want to optimise for bandwidth on the Amiga, you really only need to cookie cut parts that include transparent pixels. You can simply do regular blits for the rest. Now for small objects that doesn't make much sense, but for bigger ones it can. I think some Amiga games actually do this, but I can't remember of the top of my head which ones.

However, chunky still doesn't get all the advantages you might think here regardless. Skipping transparent pixels also loses some bandwidth, because skipping pixels means using the bus less efficiently. See, it requires either knowing what's in the background (so reading it first - which negates the advantage) or writing in smaller packets when transparency is shown (which still requires the mask). Depending what you're drawing, this can be as bad as requiring byte by byte transfers. This is much less efficient as you really want your transfers to be as big as possible to optimise bandwidth usage. Block based drawing might technically spend more bytes on the bus, but they do their transfers in much bigger chunks and that might negate some or all of the above advantage.

An interesting question then is: how did DOS games actually do their 2D drawing? I not 100% certain, but AFAIK there seem to be only two real world implementations that were used in large numbers: compiled sprites, or block transfers. The latter being more generic and more like the Blitter does it, the former allowing the programmer to be very specific with things such as transparency, but also less efficient in some ways as pointed out above (not to mention requiring sprite specific code for each animation frame - generated or hand written).

A long time ago I did speak to someone who coded some simple PC games on VGA and he said they used a simple dirty rectangle approach and drew everything in rectangles very much like the Blitter would've. I can't remember exactly why they did it like that though. Would be interesting to see how common this was.
All chunky is hypothetical when talking about Amigas, isn't it?
Real world advantages and disadvantages are clearly vastly more interesting than stuff that could've been done, but wasn't.

My problem here is that your way of arguing pro's and con's is that it borders on revisionism. The solution you called for does not actually exist. No one made VGA cards with a dual layer 16/16 colour mode. Real world chunky displays therefore did not have this advantage, which in turn makes me really uncomfortable accepting it as 'an advantage'.
Whenever we were discussing the pros and cons of chunky vs. planar it is only relevant for my point that AGA clearly is a job half done and that a typical engineering team would not have been satisfied with the result. They would have wanted to add chunky the moment they stepped up to 8 bit palette depth. It's a natural choice.
I disagree that 8 bit chunky would've been the natural choice for the Amiga. It never supported chunky before and at the time of AGA's design phase (1991), the first popular 3D chunky games were not yet out.

I fully agree that AGA was not great. But 8 bit planar made perfect sense for the Amiga. It already had six bitplanes, adding two more is the natural choice to get to 256 colours.
And I disagree. All the Amiga demo and coding scene was about from 1993 to 1996 was to implement Doom-type engines. Perhaps it was a mass psychosis but the effect Doom had on young people at that time was immense.
The market disagrees with you. Don't get me wrong, I understand where you are coming from - but memory and culture can be deceptive. 2D games vastly outsold 3D ones. As an example: the original Lemmings outsold Doom by almost 8x over their respective lifetimes. 2D consoles were at the height of their popularity when the A1200 launched. It wasn't until the PlayStation launched over here in late 1995 that 3D suddenly became really hot. Further cementing the fact that 2D did so well is how the first 3D game (chronologically) on the list of best ever selling games came out in 2001 (GTA III).
Yes, you are probably right. Looking at all the technical shortcomings of the PC it becomes clear that the technology itself wasn't the decisive factor. At least not until the PC technology was so powerful that it could do great games. Until then it was probably just about the software that a PC could run and that somehow had become standard in offices around the world.

However, in my time as a great hindsight-manager I would also have had Commodore invest in software, porting and developer support. And thus Commodore also would have been economically more powerful than Apple...
Hindsight is always 20/20. And that is probably the big problem with threads like these. It's so easy to conclude you/someone else would've done better and that things would be different.

Truth is that it is fun to consider such things. But, in reality, none of us know how we really would've done things if we were a Commodore manager back in the day. We only know what to do because we know where we ended up. Had you been a manager back then, you wouldn't know that and thus may have made very similar mistakes.
I'm not sure I understand. You can always leave bits unused in each byte but you would have to transmit the meaningless bits anyway during DMA.
It was a silly idea. I was thinking that you could have a display that operated like a planar one during display fetches (i.e. only fetch four to six planes worth of data), but like a chunky one during writes. In retrospect it doesn't make much sense.

Guess it's best I don't go building any hardware any time soon
Well, rearranging pixel data is surely a small effort. Of course, the meaning of "small" just depends on what you are comparing it to. If support for those pixel formats could be designed in the time it takes to write a forum comment, then they would have done it. If you compare to the effort for a hardware polygon rasterizer, the effort becomes invisible.
That makes sense.
I'm not sure about this. There were two peak years and then a long plateau phase at a very high level. The end of the peak seems to be rather related to the Xbox 360 eating away a chunk of the market. The next generation was due because Sony did not disregard the competition as Commodore did but not so much because they saw their current product become a shelf warmer starting next year.
The XBOX 360 was launched late 2005, PS2 sales started declining in two years earlier than that. Both next gen consoles were announced at almost exactly the same time.

I just don't agree here. The PS2 was clearly declining in sales when the PS3 was announced (and later launched). The XBOX 360 was clearly pushed forwards due to poor XBOX sales at the time. Neither of these companies launched their next machine while the previous one was raking in record sales. The same thing actually happened before: Sony launched the PS2 after the PS1 had started to decline. And AFAIK it also happened again with the current consoles.

As I pointed out before: in 1991, Commodore sold more Amiga's than ever before. This really is different from either Sony or Microsoft's time line, where both launched new products while the previous product was selling well below the figures of their best years.
BTW, I'm pretty sure that both Sony and Microsoft start developing their next generation of consoles the moment they have the recent one out of the door. If they didn't, they would have to lay off half of their engineering department for a few years (the other half would make cost-reduced slim versions of the current model). The Commodore R&D team was also always busy but unfortunately with a lot of braindead products (264, 128 and 65 as most prominent examples) when they should have concentrated on the Amiga.
They indeed did start designing the next PlayStation almost as soon as the first was released (assuming Wikipedia got it right ). Commodore actually did the same with the Amiga, they just made foolish decisions on how to go about it. There's the Ranger chipset that was never finalized, there's the many flavours of AAA, etc.

They should've just stuck with one direction. Jay Miner's ideas for the 1987 Amiga update were fascinating and would've been very nice to see. But instead they went all over the place. I don't think the 264 & 128 are that relevant in terms of Amiga development though. Both were designed (and IIRC released) before Commodore was even considering buying the Amiga Corporation.
Originally Posted by Galahad/FLT View Post
And where are the SIMM sockets going to go on the board? Yes they could have fitted them, but the only space is where the expansion boards go underneath, which would have removed further expansion.

Commodore should have at least have designed and released cheap expansion ram boards either complete or with simm sockets on them, but they seemed to figure (quite correctly) that after market companies would sort that.

Remember the 512k ram expansions for A500, how many had an actual official Commodore board? Probably very few because the after market alternatives were cheaper and did the same job.
I seem to remember that A1200 RAM only expansions were not that expensive. Though still clearly a lot more expensive than a SIMM was.
Originally Posted by rare_j View Post
Also, why was 4mb the most common RAM upgrade available? If a basic 2mb fast ram expansion option were available more cheaply this could have made a big difference.
Initially, there actually were 1, 2 and 4MB expansions. But the 1 and 2MB expansions soon died out. I guess they didn't sell?

Last edited by roondar; 27 July 2019 at 13:56.
roondar is offline  
Page generated in 0.06303 seconds with 11 queries