What if the CD32 had 1.5MB chip + 0.5MB Fast RAM

[eXeler0]

Would the CD32 have been a better console if it had 1.5MB ChipRAM + 0.5MB FastRAM instead of the 2+0 config it got?


Drawbacks?
Benefits?

[alpine9000]

Originally Posted by eXeler0

Would the CD32 have been a better console if it had 1.5MB ChipRAM + 0.5MB FastRAM instead of the 2+0 config it got?


Drawbacks?
Benefits?

I would say YES!

Especially the CD32 where games could have been written specifically for that configuration.

[-Acid-]

What they should have had was 2mb chip and a simm slot on the board, empty if needed but the option to put some fast ram in should have been there (same on the A1200)... it would have cost next to nothing but added so many more possibilities.

[Amigajay]

Better as in would make some games run faster, but not any better because at heart it still had the behind the times 020 and AA chipsets, so no different or amazing games would have come out of it. plus Commodore wasnt in a position to add more cost to a machine, along with annoying A1200 owners in the process which they would have never have done, which is why i guess the Akiko idea was born, a little extra performance for next to no cost.

[jotd]

Better because it's very likely that game code takes around 300 to 400k memory.

however, if the executable takes more than that in 1 segment, you have to split the segments or it won't fit in the 512k ram.

Would have been an issue for CD32load adapted games from AGA 2MB chip games. also, what would have been the fastmem base address? $180000 ? something else?

[rare_j]

A while back a programmer posted that as little as 64kb dedicated to the cpu could have made a big difference to the A1200 and CD32.

[Seiya]

a CD32 with 2 MB Fast, maybe some more graphics and animations for sure.

[Leffmann]

It would definitely be a nice push for time-critical code, but 512K of chipmem is a lot to lose.

In practice I don't think they could've arbitrated the memory like that because of how the architecture works. It would've been 1M+1M, or 2M+256K or more.

Originally Posted by rare_j

A while back a programmer posted that as little as 64kb dedicated to the cpu could have made a big difference to the A1200 and CD32.

I think so too. The Gameboy Advance has the equivalent of 32K of fastmem to help it out, even with a much faster CPU to begin with.

[meynaf]

The faster the cpu, the more effect the fastmem has. Thus I'm afraid 14mhz 68020 is too low and that converting 512k from chip to fast wouldn't have made much of a difference ; better yes, but nothing spectacular.

[Bruce Abbott]

Upsides:
1. slightly faster code

Downsides:
1. less memory for graphics and sound

2. Increased hardware costs

3. Lower compatibiliy with stock A1200

Code running from true Fast RAM can be up to 70% faster. That's the same as an 8088 running at 8MHz vs. 4.77MHz (ie. not much better). In practice the actual speedup might be less because the CPU must still slow down when accessing ChipRAM and custom chip registers, and some programmers would probably just piss away the extra speed anyway (hey, we've got faster RAM - so now we can code in C and port over this shovelware!).

Splitting the 2MB RAM space into 1.5MB Chip and 0.5MB Fast makes no sense. You would need more chips and more board space than simply adding 2MB of FastRAM, while still needing a separate controller and bus for the 0.5MB.

2MB of true FastRAM would have been welcomed, but that would have raised the cost significantly. The console market was very price sensitive and Commodore didn't have a lot of money to spare. The CD32 was Commodore's last desperate attempt to regain market share without going even more into the red, so it had to be as cheap as possible to produce.

IMO they got the balance about right. AGA titles were starting to appear for the stock A1200 (2MB, floppy drive only) showing what the platform was capable of. The CD32 matched this hardware, but with the addition of a huge 'hard drive' at lower cost. It also had the added value of playing hifi music CDs (no mp3 players back then!) and you could plug in a mouse and keyboard to almost make it into a full computer - the only essential thing missing was a floppy drive (which is why I designed a DIY board to give it one). If all you wanted to do was play AGA games then it was better value than the A1200 - and you had an upgrade path to full computer status via the expansion port.

Not having a floppy drive port on the stock CD32 may seem like an oversight, but it was a deliberate design decision intended to fix the problem most developers were concerned about - software piracy. Unfortunately what Commodore (and most developers) didn't appreciate was that piracy was a major sales driver. A machine that locked out pirates was bound to fail!

Even without FastRAM the CD32 was much more powerful than the Sega Megadrive (its nearest rival and likely inspiration). Only problem was by the time Commodore got it out the door the 2D console market was dying. Then the Sony Playstation arrived... to compete with that the CD32 would have needed a much more powerful processor and a completely redesigned chipset.

[Amigajay]

Originally Posted by Bruce Abbott

Only problem was by the time Commodore got it out the door the 2D console market was dying. Then the Sony Playstation arrived... to compete with that the CD32 would have needed a much more powerful processor and a completely redesigned chipset.

The 2D console market was thriving still in 1993 and would for a couple more years, heck the SNES had only been released the year before! Yes we had the expensive 3DO and Jaguar but they both failed because the 16bit market was only in its first full stride, i believe nothing released between 1993 and 1995 would have survived, Playstation did because the timing was right and they put hundreds of millions into it, Commodore couldnt have entered the console market even if they shoved a 060 and 8mb of ram into the CD32 and sold it for the same price!

[OmegaMax]

Super Nintendo was released in 1990,1991 in north America.

[Amigajay]

Originally Posted by OmegaMax

Super Nintendo was released in 1990,1991 in north America.

I was talking about Europe where the CD32 had its main release.

[alpine9000]

Originally Posted by Bruce Abbott

Upsides:
1. slightly faster code

Downsides:
1. less memory for graphics and sound

2. Increased hardware costs

3. Lower compatibiliy with stock A1200

It's not just that the code runs faster, it's that you can run code while the blitter is operating without the same bus contention as when running code out of chip ram. This can make a huge difference to what you can get done in a frame.

[Bruce Abbott]

Originally Posted by alpine9000

It's not just that the code runs faster, it's that you can run code while the blitter is operating without the same bus contention as when running code out of chip ram. This can make a huge difference to what you can get done in a frame.

It can make a huge difference in some circumstances, but I still say that in practice few CD32 titles would have been significantly improved.

Why not? First the technical reasons:-

1. In most games graphics manipulations are by far the biggest workload, which is best handled by the custom chips (this is still true today. Even the fastest Intel CPU sucks at software rendering when compared to a good 3D graphics card). The CPU could render some graphics in parallel to help the blitter, but any speedup would be modest.

2. You can minimise memory contention by making use of system routines that run from ROM, getting critical code to fit in the 68020's instruction cache, and using lower screen depth/resolution (the CD32 was practically limited to lower resolutions anyway because it had to work with a standard TV set).

Secondly there are the development conditions of the time to consider:-

1. AGA titles were beginning to appear for the A1200, which also had only 2MB Chip and no fast RAM. Since contemporary AGA games had to work in that environment anyway, developers were used to working within those limitions.

2. To gain sufficient market share and be profitable, CD32 titles would have to be produced rapidly with low development costs. That means quick ports from older Amiga models, rather than all-new code carefully crafted to wring the best out of the CD32.

3. The kinds of games that could really use the speed boost would probably still be too slow. To to be worth developing 'next generation' games for, the CD32 would have needed a much faster CPU and far more advanced graphics engine. Given Commodore's financial position that was simply not possible.

IMO the clamour for ever more powerful hardware missed the point. A great game is defined by originality and playability - not CPU power or pixel density. Many great games worked fine on a stock A500 with no Fast RAM and a miserable 16 bit CPU running at 7MHz. With 4 times the memory bandwidth and more than twice the CPU power AGA games should have knocked our socks off - but did they? Many AGA games were slick and colorful, but boring. The reason? Too much emphasis on technical prowess and not enough on gameplay and innovation.

I recently downloaded The Faery Tale Adventure (which I used to play a lot on my A1000) and am having a blast running it on an A500 I picked up from eBay. This game won't even run on my A12000/030/50MHz, but I don't care. It could be 'upgraded' to take advantage of the faster CPU, extra RAM and AGA chipset, but what would be the point? - it's just about perfect as it is. A faster frame rate, more colors and fancier graphics wouldn't make it any more playable or enjoyable.

[alpine9000]

Originally Posted by Bruce Abbott

It can make a huge difference in some circumstances, but I still say that in practice few CD32 titles would have been significantly improved.

Why not? First the technical reasons:-

1. In most games graphics manipulations are by far the biggest workload, which is best handled by the custom chips (this is still true today. Even the fastest Intel CPU sucks at software rendering when compared to a good 3D graphics card). The CPU could render some graphics in parallel to help the blitter, but any speedup would be modest.

2. You can minimise memory contention by making use of system routines that run from ROM, getting critical code to fit in the 68020's instruction cache, and using lower screen depth/resolution (the CD32 was practically limited to lower resolutions anyway because it had to work with a standard TV set).

Secondly there are the development conditions of the time to consider:-

1. AGA titles were beginning to appear for the A1200, which also had only 2MB Chip and no fast RAM. Since contemporary AGA games had to work in that environment anyway, developers were used to working within those limitions.

2. To gain sufficient market share and be profitable, CD32 titles would have to be produced rapidly with low development costs. That means quick ports from older Amiga models, rather than all-new code carefully crafted to wring the best out of the CD32.

3. The kinds of games that could really use the speed boost would probably still be too slow. To to be worth developing 'next generation' games for, the CD32 would have needed a much faster CPU and far more advanced graphics engine. Given Commodore's financial position that was simply not possible.

IMO the clamour for ever more powerful hardware missed the point. A great game is defined by originality and playability - not CPU power or pixel density. Many great games worked fine on a stock A500 with no Fast RAM and a miserable 16 bit CPU running at 7MHz. With 4 times the memory bandwidth and more than twice the CPU power AGA games should have knocked our socks off - but did they? Many AGA games were slick and colorful, but boring. The reason? Too much emphasis on technical prowess and not enough on gameplay and innovation.

I recently downloaded The Faery Tale Adventure (which I used to play a lot on my A1000) and am having a blast running it on an A500 I picked up from eBay. This game won't even run on my A12000/030/50MHz, but I don't care. It could be 'upgraded' to take advantage of the faster CPU, extra RAM and AGA chipset, but what would be the point? - it's just about perfect as it is. A faster frame rate, more colors and fancier graphics wouldn't make it any more playable or enjoyable.

You can't run game logic out of the ROM and a game with many intelligent on screen objects is unlikely to be able to run the logic routines without blowing the tiny 020 cache.

I just finished writing a game with "relatively" complex AI. It runs much faster and allows many more on-screen objects on my A1200 with fast ram compared to chip-ram only config. If the AI logic wasn't coded for an A500 with chip ram, it could have been much more interesting.

When you say what's the point ? I say the games would be richer and more fun.

As to if software houses would have taken advantage of it, you're probably right. Not much CD32 specific stuff out there.

[IanP]

Adding fast RAM to the CD32 design would add to it's complexity and cost. The DRAM is organised as 4 x 8 bit chips ganged together to provide the 32 bits for the data bus with multiplexed row and column style addressing. Alice converts the Amiga 23 direct address lines and other addressing signals to chip RAM address signals in conjunction with Akiko, there's also some buffering. So adding fast RAM would double the required DRAM interface circuitry and you can't use the existing Alice chip if it is to handle any of the extra work. If you assume they stick with 8 bit chips then for 512K of fast RAM you need 4 x 128K chips and for 1.5M chip RAM you need 12 x 128K chips. You then have all the complexity of laying out and routing all these chips on a small motherboard. So maybe it's clear why there is no fast RAM or SIMM slot on the CD32 for the marginal performance boost it would give.

[kolla]

And besides, there is an expansion slot, and a card to add 8MB RAM and IDE controller is literally something that can be designed and built over a weekend

[gjb1985]

Originally Posted by kolla

And besides, there is an expansion slot, and a card to add 8MB RAM and IDE controller is literally something that can be designed and built over a weekend

this is what I got and now my cd32 is the only relevant amiga hardware I need.

[Hiddenevil]

Love my CD32, I bought one when they first came out here in the UK. I remember after the charm faded, I began to realise how restricted the console really was. As a teenager, I think I bought in to the adverts, believing it would have "Incredible graphics". But it was much of the same really and after a few years sold it on. My 060 A1200 could run quake, doom, the games on my CD32 seemed some what lacklustre by comparison, especially next to my SNES.

Knowing what I do now about Commodore, the idea that they would actually invest money on developing a decent console, leaves me smiling. If there were two ways of doing something, Commodore would do neither and do it on the cheap instead :-P

I think compared to cartridge based consoles of the time, the CD32 wasn't badly spec'd, Commodore just didn't have the resources to properly support it. Like entering your car in the race, but neglecting to fill the tank with gas.

The extra memory couldn't have hurt, more save slots anyone?