What Makes The Amiga CD32 The Best Amiga?

[S0ulA55a551n]

Originally Posted by mcgeezer

Here's a case point about the CD32 not being great.

Most people know that I'm porting Super Sprint to AGA Amigas. I last touched the CD32 back when I was doing Bomb Jack so I had the code there for saving to NVRAM. Bomb Jack has 10 hi-score names, Super Sprint has 30.

Just storing the 30*8 bytes brings it to 240 bytes and nearly a quarter of the NVRAM just for this game. Was it worth the effort to support CD32? Not really. Why did I do it? Because as I said earlier, I like all Amiga's.

Should I adjust it from saving 30 names from the arcade?

(Except them new fangled blood sucking things... don't like them).

Just wondering what you have against vampires ? If you just use it as an accelerator then its the same as any amiga accelerator except a lot lot lot faster. Is it the product you dont like, or the team behind it. Just interested. I have two vampires. But I can find some of the team behind the product a tad abrasive

[roondar]

Originally Posted by mcgeezer

Yeah probably.

Probably only worth storing the changed scores too.

Might as well do it, there 36 chars. 1 long word for the score but it doesn't go above 999,999.

Edit: Looks like I can get it down to under 100 bytes - 10 units.
Certainly not a priority though.

Was driving to work today (don't ask, my employer doesn't believe in the whole working from home business even though it's perfectly possible) and remembered this thread, prior to your edit.

One additional trick would be to use some form of other pre-encoded compression technique (f.ex. Huffman coding with a static pre-calculated Huffman table and drastically slimmed down headers). In this case the amount of space you'll need in VRAM will be variable but normally much lower.

Or if the last digit of the score never changes or only allows a couple of values you might be able to switch to one word per score

[mcgeezer]

Originally Posted by S0ulA55a551n

Just wondering what you have against vampires ? If you just use it as an accelerator then its the same as any amiga accelerator except a lot lot lot faster. Is it the product you dont like, or the team behind it. Just interested. I have two vampires. But I can find some of the team behind the product a tad abrasive

To be fair, it's not that I dislike them - it was a bit of tongue and cheek.
However as a programmer the whole supporting of accelerators is a royal pain in the arse, it's a real headache because there's a bit of an expectation to support all hardware but it's a hell of a lot of work. While I'll try to get limited testing done with accelerators I will always aim to get stuff working on stock hardware.

Originally Posted by roondar

Was driving to work today (don't ask, my employer doesn't believe in the whole working from home business even though it's perfectly possible) and remembered this thread, prior to your edit.

One additional trick would be to use some form of other pre-encoded compression technique (f.ex. Huffman coding with a static pre-calculated Huffman table and drastically slimmed down headers). In this case the amount of space you'll need in VRAM will be variable but normally much lower.

Or if the last digit of the score never changes or only allows a couple of values you might be able to switch to one word per score

Yeah, loads of stuff I could do. My point is that for a games console of that time 1kb of NVRAM was pathetic. I really shouldnt have to employ compression techniques to store a simple hiscore table - the hardware should have more NVRAM.

I'll spend limited time doing it for sure... if the CD32 players don't have enough NVRAM space for the hi-score then well so be it.

[roondar]

Originally Posted by mcgeezer

Yeah, loads of stuff I could do. My point is that for a games console of that time 1kb of NVRAM was pathetic.

I absolutely agree!

The 3do has 32KB built in as standard and people complained about that being far too limited all the time (quite literally the second google hit I found about it talked about it being a pitiful amount).

Now, had Commodore provided a memory card slot and memory cards from day one this wouldn't be a problem. But they never did anything of the sort, nor required developers to offer options to store data externally. It's basically a giant mess.

Anyway, I was just considering the puzzle for fun

[S0ulA55a551n]

Originally Posted by roondar

I absolutely agree!

The 3do has 32KB built in as standard and people complained about that being far too limited all the time (quite literally the second google hit I found about it talked about it being a pitiful amount).

Now, had Commodore provided a memory card slot and memory cards from day one this wouldn't be a problem. But they never did anything of the sort, nor required developers to offer options to store data externally. It's basically a giant mess.

Anyway, I was just considering the puzzle for fun

For me its the single biggest issue with the CD32, who wants a console where you have to choose what gems you can save, even if they offered it in the first place. Passwords etc were always such a cope out and hated it when game makers were too cheap on the MD/SNES to include game save on the cartridge.

I cant think of single comparable console that does not allow game saves of decent size, or a way of providing it

Every other amiga model provided a way of doing it. So this alone arguably makes the CD32 the worst amiga.

[str0m]

To be fair the Saturn was also bad with only a small amount, I forget how much but some games can take up half although that's rare. The Dreamcast and it's '200 blocks' also rubbish. You could get larger ones but split into banks so you had to flip between them making it a right faff to find your save sometimes.

Larger memory cards like in the PS1 etc were a massive improvement

Always like how the Neo Geo arcade cabs had their memory card slot so you could take your save home

edit: With the Saturn you could plug in a backup ram cart and transfer the internal memory to it. Some games may have supported saving direct, not sure.

[roondar]

As I understand it, the Saturn/3do/PS1/Dreamcast/PS2 all supported external memory cards from day one. Not optimal, but clearly a better solution than a tiny fixed amount.

Edit: didn't see your edit...

[Thorham]

Originally Posted by Gilbert

My point is purely that CD32 crushes the Megadrive and SNES in terms of hardware power.

Obviously. The problem is that it beats those 16 bit consoles as a computer, not a console.

[roondar]

Originally Posted by Thorham

Obviously. The problem is that it beats those 16 bit consoles as a computer, not a console.

That's actually a great way to put it. It also shows why the A1200 was usually the better choice than the CD32 as that actually is a full computer from the get-go.

[Gilbert]

Originally Posted by roondar

Note: I had to shorten this post quite a bit as I originally made it much too long (some things I didn't answer as a result). Sometimes me doing this makes posts seem less friendly, but that is not the intention.

No, they really didn't understate the abilities.

Rather, they overstated the CPU's performance by a factor of two, fudged the sprite abilities by not pointing out the Blitter is not going to be able to compensate for the limited number of sprites per scanline, etc.
It's worth pointing out that Amiga graphics and sound take up much more space than MD graphics for various technical reasons. You really can't compare the size of VRAM on consoles directly to the size of Amiga chip ram.

The MD can only display 20 sprites per line!? That's the equivalent of up to 640 pixels of sprites per line - that's actually a lot. Meanwhile, the Blitter can display at most around 20 objects of the same size as MD sprites in total. Not per line, just in total.
What you're doing here is called cherry picking: you pick a single (or a few) example(s) that suit your argument best and ignore they're not representative of the whole.

Case in point: there are much better playing & looking vertical shooters on the MD than Battle Squadron and Battle Squadron happens to be a port from the Amiga version, which means it was not originally designed with the MD's greater sprite abilities in mind. Anyway, even so the MD version visibly runs much smoother than the Amiga version in your video. Which is really kind of important for shooters
You may not know this, but I actually program the Amiga. I've done Blitter, Copper and Sprite programming and made several examples available that show interesting ways to use them. I actually know reasonably well what is and isn't possible.

--
And now for something a bit more serious:

Other, quite talented, Amiga programmers in this thread (such as mcgeezer & DanScott) have also pointed out to you the CD32 is not as powerful as you think. By saying the CD32 can really do much more than what we've seen (and more than what the coders here are saying), what you're essentially and hopefully unintentionally doing is saying that we all don't know what we're talking about. It further implies that all the programmers for the system in the past were also either incompetent or lazy because well, real world CD32 games are not actually more impressive than 16 bit console games.

This is the key reason I'm still replying because I don't think this would be at all a fair assessment of the abilities of the coders in this thread/the Amiga community as a whole. Constantly overstating the abilities of the Amiga makes the people that code on it, make music/art on it seem to be really poor at what they do. And I really, really dislike that as we have some stellar individuals on this forum/in this community doing great work.

I thought you said they understated how many colours the Amiga could disiplay while playing back video in your earlier post. And the max number of colours they stated onscreen in Ham 8 mode was lower too according to the specs I googled online. So that's 2 specs they understated.

As for 640 pixel sprite width - if you look at a lot of Megadrive schmups they flash sprites on and off on alternate frames so something is up with the horizontal limit e.g. Musha does this. I guess if you are using a lot of sprites for small objects you are in trouble. Also Rainbow Islands - The megadrive clearly runs out of horizontal sprites while drawing the rainbows.... masive flicker. The Amiga 500(!) can just easily draw them to the background. The CD32 could do a perfect version of Rainbow Islands easy but the Megadrive probably couldn't

My negative points (I have lot of positive points too) about Megadrive games is that they lack onscreen colors and the sound is weak (in general) and a lot have that annoying status panel (if the Megadrive has all these sprites - why use that panel? just overlay the score on the screen)

Here is some interesting discussion about the realities if using sprites and VRAM on the Megadrive and SNES. Is nowhere near as simple as on the Amiga. There is a lot of decompression of sprites that goes on to to save cart space.

http://forums.nesdev.com/viewtopic.php?f=23&t=17601

The point was - that is the Amiga 500 version of Battle Squadron based on 1985 tech. The CD32 one would be twice as good. I read the interview on the Codetapper website and the dude said that Hybris alone has "100s of colours" onscreen on the A500 version

Well if you look at my other contributions on the forum you will see me giving praise for new games when I like them. So that's not fair saying I am insulting Amiga developers by praising the CD32. I also bought Reboot R and gave it good praise. So I am supporting the community. I even seriously offered McGeezer £50 to remake Ghouls n Ghosts. But sadly the rest of the CD32 community didn't step in and help. Action not words guys!

I feel in the CPS1 thread we established the CD32 could do close versions of CPS1 games (using dual playfield and its big sprites) so I don't think saying it can beat the 16 bit Megadrive is any way controversial. Especially since it could fit the Megadrive Ghouls n Ghosts cart into it's 2Mb memory twice over.

No one back in the day made decent CD32 games - even something like Banshee has no music. I was told CD32 can stream in new data maybe 250k+ a second, no one has used that ability. There was no big userbase for them to bother making a huge effort for back then. So the CD32 has a lot of untapped potential. Those rotation effects they used on Brain the Lion - I remember reading took like 6 years after the Amiga came out to figure those out. 32 bit + AGA + CD is all new hardware to find tricks for.

Since people are getting upset though I will post less and also write more upbeat posts and make more posts about the positive points of the CD32 instead.

EDIT : Hey roondar you remember us discussing the opposite of this (me championing the Megadrive) a long time ago? Funny how things change. I didn't even like CD32 much back then. You guys helped change my mind about it!

Here s what the big man has to say about the CD32 and Megadrive

[rabidgerry]

I don't really like the CD32, or perhaps should I say I have no interest in it, if you like it fine.

But this guys video.....................well what an asshole.....that's all I can really say on the matter.

[ Show youtube player ]

[DanScott]

Originally Posted by rabidgerry

I don't really like the CD32, or perhaps should I say I have no interest in it, if you like it fine.

But this guys video.....................well what an asshole.....that's all I can really say on the matter.

[ Show youtube player ]

actually... he's pretty funny, and everything he says (bar 1 or 2 historical inaccuracies) is true

glad he didn't slate Bubba n Stix!

[d4rk3lf]

@Gilbert
Now sure how that interview is relevant.
His job was to praise new C= console, and I guess we can easily find very similar interviews from Sega representatives.

What makes me a bitter taste is that we are comparing Sega from 1989 (with CD Rom) to new C= console (CD 32) that should be released in 1993.
So, 4 years later.... 4 years in computer life is very long period.

No wonder why Playstation destroyed everything on the market, when it was released.
Competition was one (computer) century behind.

[d4rk3lf]

----------
P.S.
Long story short:
I just can't grasp, how they were not ashamed to compare new console to 4 year old console that was already on the market? It's crazy.
This interview proves exactly opposite to what Gilbert want.

[Gilbert]

Originally Posted by DanScott

actually... he's pretty funny, and everything he says (bar 1 or 2 historical inaccuracies) is true

glad he didn't slate Bubba n Stix!

It is funny in places but trouble with these videos is they never have the best games. You could do a review for the SNES with some pretty terrible games if you wanted - but no-one would ever do that. Let alone do one with a broken SNES.

I think the GameSack CD32 episode is more balanced - but again could have done with better games

[ Show youtube player ]

Originally Posted by mcgeezer

Just storing the 30*8 bytes brings it to 240 bytes and nearly a quarter of the NVRAM just for this game. Was it worth the effort to support CD32? Not really. Why did I do it? Because as I said earlier, I like all Amiga's.

Should I adjust it from saving 30 names from the arcade?

(Except them new fangled blood sucking things... don't like them).

I think Vampires have their place but real Amiga users just use the standard models and maybe add extra memory.

I had an idea about compression. In the future will it possible to use some kind of mathematical formula (or more than one combined with unique sections) to produce the game data from a few initial variables? A bit like how the formula for pi (22/7) produces that long non-repeating number that could fill a whole hard drive. I read a while back that AI coded a version of Pacman after watching the game run many times. AIs can figure out things the human mind can't - like they do already with 'big data'. They could figure out some kind of amazing compression maybe? Would this be possible? (i am being serious but I know this might seem a bit dumb)

[rabidgerry]

Originally Posted by DanScott

actually... he's pretty funny, and everything he says (bar 1 or 2 historical inaccuracies) is true

glad he didn't slate Bubba n Stix!

I've watched his movie videos before and the are a lot more informative and less childish than that bullshit. I got 2-3 mins into that before I wanted to punch his face in. Such a contrast to my previous experience with his videos.

I watched a video just before that on basically the same thing and despite the guy sounding like he had a shoe in his mouth it was much more informative and a lot less childish.

Originally Posted by Gilbert

It is funny in places but trouble with these videos is they never have the best games. You could do a review for the SNES with some pretty terrible games if you wanted - but no-one would ever do that. Let alone do one with a broken SNES.

Yes it was like he went out of his way to make it look like shit(ter). One thing blew my mind, he didn't know which button was fire. My logic defaults to the red button as fire, he nit picked the life out of the button layout simply for the sake of it.

[mcgeezer]

Originally Posted by Gilbert

I had an idea about compression. In the future will it possible to use some kind of mathematical formula (or more than one combined with unique sections) to produce the game data from a few initial variables? A bit like how the formula for pi (22/7) produces that long non-repeating number that could fill a whole hard drive. I read a while back that AI coded a version of Pacman after watching the game run many times. AIs can figure out things the human mind can't - like they do already with 'big data'. They could figure out some kind of amazing compression maybe? Would this be possible? (i am being serious but I know this might seem a bit dumb)

In the future, definitely.

[Gilbert]

Originally Posted by mcgeezer

In the future, definitely.

I guess maybe the base size would have to be quite large still - a bit like human DNA information to grow a human or acorn to grow an oak


I thought of some cool compression ideas for CD32

Highscore table - store difference between successive scores (in the table) where this saves space (compression routine can decide if this is best way and code for in saved bytes)

Football game with lots of editable player names. Store as differences from large list of preset names stored on CD
You could have a thousand first names and a thousand surnames (or much less than this might work better)

I mean this is not quite right, the way I am stating it - but I think the large CD space can be used to store predefined values that can be used as a base to build decompressed values from

[S0ulA55a551n]

Originally Posted by Gilbert

I guess maybe the base size would have to be quite large still - a bit like human DNA information to grow a human or acorn to grow an oak


I thought of some cool compression ideas for CD32

Highscore table - store difference between successive scores (in the table) where this saves space (compression routine can decide if this is best way and code for in saved bytes)

Football game with lots of editable player names. Store as differences from large list of preset names stored on CD
You could have a thousand first names and a thousand surnames (or much less than this might work better)

I mean this is not quite right, the way I am stating it - but I think the large CD space can be used to store predefined values that can be used as a base to build decompressed values from

I like that you are trying to think around the low save space problem, and necessity does breed invention. I mean you could store all sorts of combinations of data on cd and used a relatively small set of indexes to build up larger data sets.

[mcgeezer]

Originally Posted by Gilbert

I guess maybe the base size would have to be quite large still - a bit like human DNA information to grow a human or acorn to grow an oak


I thought of some cool compression ideas for CD32

Highscore table - store difference between successive scores (in the table) where this saves space (compression routine can decide if this is best way and code for in saved bytes)

Football game with lots of editable player names. Store as differences from large list of preset names stored on CD
You could have a thousand first names and a thousand surnames (or much less than this might work better)

I mean this is not quite right, the way I am stating it - but I think the large CD space can be used to store predefined values that can be used as a base to build decompressed values from

Nice.

Alternatively in the future when we can buy flux capacitors from the shop, we burn rubber up to 88, kidnap some Commodore engineers and have them read this thread from the future. This way i’ll not have to code the Worlds greatest compression routine because commodore could just put an extra 1kb of nvram in their next gen console.