Theory thread: How would you update C64?

[ImmortalA1000]

Originally Posted by Havie

This is a wonderful tune - there's something special about chip music compared to samples and the SID does it best.

Amiga 500 etc can do reasonable, for your average TV/monitor speaker, SID emulation thanks to Hakan Sundell and Ron Birk's routines.
[ Show youtube player ]

Samples are a double edged sword, you're at the mercy of the musician's standards when it comes to instrument selection and sample frequency of instrument samples. Ditto for SFX, it got old fast hearing the same old 'Battlestar Galactica' explosion in games.

With the SID you have to be an expert programmer AND a musician which is a bit of a negative. If you can understand what the guys from BBC's Radiophonic Workshop are talking about in interviews then great, if not then you will have problems making cool sounding instruments. It's quite a technical process and back then I never found a decent package to replicate the sterling coding efforts of Rob and Whittaker that did use undocumented features of the 6581 at the time. Not that being a musician using something like a Roland Promars analogue synth is any different, exchange typing in HEX codes with many many wire patches etc to get the cool sounds that make all the difference.

Also helps if you accidentally discover 3/4 of the SID oscillators are not mutually exclusive, you can set the bit for any combination of three waveform types triangle/sawtooth/pulse and the SID will use more than one oscillator on one channel. So you can easily get the sound of 6-9 channel SID audio from a 3 channel SID in some situations.

Here is a video showing the effect on an 8580 SID in my 1991 64C
[ Show youtube player ]

6581 also does it but the bias/timbre of the resultant combined waveform sound may be slightly different, not tried it. Need a monitor/TV with really good quality speakers for this too if I'm honest.

IIRC David Whittaker used some short waveforms and used those for EPIC like a sort of virtual analgue synth but the samples are so short they have that sort of SNES 'short sample' sound to them which I never liked myself vs rich samples of Agony/Beast1 etc etc

[Promilus]

Originally Posted by Retro-Nerd

Yep, the SuperCPU. There are some SCPU fixed classic C64 games which benefits from this massive turbo/ram expansion.



Stunt Car Racer e.g. run in 50fps. Or faster screen switching for Last Ninja, very fast vector filling for the Leaderboard greens.


[ Show youtube player ]

SuperCPU, or actually CPU itself is neither expensive nor hard to get (super cpu add-on IS expensive and rare). You can still get 65C816 for fair price. TODAY! Going semi-16bit (because it's only 16bit internally) is fairly good but C64 base is still fairly limited. Actually to spread wings it requires both SuperCPU and REU (or at least faster 6502 AND DMA chip). And even then VIC II and PLA will severely limit memory performance in first 64K. But - with fast CPU having access to faster memory and with DMA C64 could really do much better. With 65C816 and few megs of ram and DMA C64 could match most of C65 specs (obviously no blitter and other advanced VIC III features).

[ImmortalA1000]

The difference between Super CPU only Metal Dust and Enforcer:Full Metal Blaster for a stock C64 is a good comparison of what can be done.

[oscar_ates]

I would update c64 by throwing it away and buy amiga or pc to upgrade

[gimbal]

There was a possibility to have a SID soundcard in your PC at some point, so you could at least pretend for a little while longer

[ Show youtube player ]

[Bruce Abbott]

Originally Posted by ImmortalA1000

Here is a video showing the effect on an 8580 SID in my 1991 64C
[ Show youtube player ]

Very interesting. Most C64 music seems to be heavy on synth effects, but this sounds like a real musical instrument. I must set up my own C64 and try it out!

[ImmortalA1000]

Thinking about what the C64 was in 1982, cutting edge sprites and cutting edge complex sound hardware compared to all the consoles or computers of 1982 along with a truly over the top minimum spec for memory. The CPU was nothing special, all the heavy lifting was being done by the custom chips for the types of games being developed in 1981/82.

On that basis then a 1991 successor to the C64 would need to have better sprite hardware than Sega/NEC/Nintendo 16bit consoles, cutting edge almost commercial quality sound hardware, and an enormous base spec of memory for any home computer so something like 4mb RAM. As character based modes + Color RAM are effectively compressed screen modes vs the way Amstrad gets 16 colours on screen something similar to vastly speed up the manipulation of a 320x200 type screen resolution compared to Amiga's bitmap method of 32kb for a 32 colour screen.

Doesn't sound like the C65 was anywhere near that advanced, better than the A500 for 2D game engines/arcade conversions with it's blitter but still not in the same league as 64kb as standard C64 vs the competition in 1982 despite using the 1981 VIC-II and SID custom chip technology.

[Korodny]

In the late eighties, there would have been market space for a C64 successor to cover the low-end home market. The Amiga was too expensive for that - and since Commodore didn't do anything in that area, Nintendo took over and sold massive amounts of 8 bit hardware.

A C64 follow-up would have had to be (a) as compatible with the old hardware as possible (because of the amount of existing software, and because the C64 was still selling like crazy) and (b) as cheap as possible (to compete with the likes of Nintendo, and because the more expensive segment was already covered by the Amiga)

That pretty much dictates what an updated C64 could have looked like:

• faster disk drive, by fixing burst IEC
• updated ROMs with a much better DOS shell (think something like JiffyDOS) and Autoboot functionality
• DualSID (SIDs were extremely cheap to produce for Commodore)
• additional 80 Column mode that uses the same video output as the 40 column mode. Can be text only, we only need it for productivity apps.
• 128 or 256 K of RAM, including a bankswitching mechanism and/or a blitter
• a CPU that can optionally run at different (higher) clock speeds, if that's possible without breaking VIC compatibility.

With those features, the resulting machine might be about as compatible to the original C64 as a C128 in C64 mode - i.e. extremely compatible. Anything that doesn't write to reserved (i.e. empty in a stock C64) I/O space should run.

[coldacid]

In other words, the C128, with a possible increase to 256kB of RAM.

[Samurai_Crow]

The reason the C128D was discontinued was it was as expensive to build as an A500 but much less capable and sold at a lower price point. Continuing the C128 would have been an even bigger mistake.

[Korodny]

Originally Posted by coldacid

In other words, the C128, with a possible increase to 256kB of RAM.

No, obviously not. The C128 never really took of because of the separate modes, and the requirement for two monitors. My suggestion is to add the C128 improvements (faster disk, better DOS, faster CPU, 80 column mode...) to C64 mode, disabling most of them by default for compatibility reasons.

Originally Posted by Samurai_Crow

The reason the C128D was discontinued was it was as expensive to build as an A500 but much less capable and sold at a lower price point.

What's the source for the claim the C128 was as expensive to built as an A500? That's news to me. It was expensive to build obviously, so the profit margin was small.

My suggestions would not increase production costs (new ROMs, fix IEC), add very little cost (IIRC, Commodore paid 13 cent to produce a SID chip at one point in time, plus cost of a (faster) CMOS 65xx compared to the older nMOS based version). 80 column mode is a 6545 (again, produced inhouse, old design) and some additional logic for color.

The C128 is a completely different beast, featuring two CPUs, two full-blown video chips, and twice the amount of ROMs than a C64.

[ImmortalA1000]

Originally Posted by coldacid

In other words, the C128, with a possible increase to 256kB of RAM.

C128 was not useful for game developers, the VIC and SID are stuck at 1mhz so in 128 mode you are stuck with a 1mhz 6502 compatible with the same identical specs so it's the world's most expensive 64k upgrade to a 1982 computer. It would have made zero difference to gamer developers if they just made a 64k RAM cart for the C64 in 1984.

The 80 column chip is a badly copied EGA knock-off MOS had kicking around, utterly useless for a home machine due to no RGB analogue output, you effectively needed an EGA monitor type RGB digital monitor, which of course only a PC owner would have already.

C128 + 1901/1902 monitor + 1571 disk drive in the UK cost MORE than a launch day 520ST + 1mb disk drive + mono monitor (you can use a colour TV with a SCART RGB cable in the EU even on a 520ST without a modulator like the 1986 520STM). Free bundled 1st Word on the ST is a lot less rubbish than any crusty old CP/M word processor you would need to buy too and the ST had a standard parallel printer port and a hard drive option within 6-9 months. The 128 was a real 'throw it in there and see' type design, didn't help the C64 market in any way that a 64k RAM cart for the C64 couldn't have done. You can't even have the 6502 and Z80 active at the same time, it's one or the other only.

(if you use the border screen blank/2mhz CPU switch trick to get the 25% CPU boost you lose sprites in the borders so your useable area is stuck at 320x200 including scores/status etc unlike a C64 so it's not that useful vs a 1mhz C64).

[TEG]

Originally Posted by Samurai_Crow

The fact that the C128 had a separate memory bus for 80 column video, moving the memory fetching off of the main bus, freed up half of the bus cycles for 2 MHz. Even that was already insufficient.

That's why adding I/O addressing and added coprocessor support would have been superior.

There is a conception flaw in the C128. Bill Herd explained it in a youtube video thanks to the 8-Bit Guy. When he designed the C128 and discussed with the engineer in charge of the VDC he assumed the VDC would have an IRQ line to inform the cpu it was ready. Instead of that, when the chip was almost finished, he discovered that the cpu have to constantly pull the state of the VDC from registers. So the 80 columns display is extremely slow.

[ Show youtube player ]

[TEG]

Originally Posted by Gorf

The 1541 on the other hand was dog-slow without fast-loader because of the famous bug in the serializer of the predecessor, but otherwise very reliable in my case.

I really do not understand why Commodore never addressed that slowness-bug in later revisions.

I guess the idea was a way to let this advantage to the C128 which did not have the flaw (only in 128 mode I guess).

[TEG]

Originally Posted by Gorf

The 1541 on the other hand was dog-slow without fast-loader because of the famous bug in the serializer of the predecessor, but otherwise very reliable in my case.

I really do not understand why Commodore never addressed that slowness-bug in later revisions.

I guess the idea was to let this advantage to the C128 which did not have the flaw and so a justification to buy it.

[ImmortalA1000]

Depends if the change would break compatibility surely?

[luncheon]

the c128 was just about right. just needed a 4-8mhz 8088 to get ibm-dos-pc-cga (through the 80col chip) compatibility instead of the 2-4mhz z80 and cpm compatibility it got

however that would have hurt the sale price, and amiga sales so it was a big no-no for cbm.

[TEG]

Originally Posted by luncheon

the c128 was just about right. just needed a 4-8mhz 8088 to get ibm-dos-pc-cga (through the 80col chip) compatibility instead of the 2-4mhz z80 and cpm compatibility it got

however that would have hurt the sale price, and amiga sales so it was a big no-no for cbm.

The 8088 was much more expensive than the Z80? DOS licence?

[ImmortalA1000]

Originally Posted by TEG

The 8088 was much more expensive than the Z80? DOS licence?

The Acorn BBC Master 512 had an 8088/8086 co-processor board and could run DOS and GEM no problem, it cost probably what a C128 would have cost with these extras and it sold..............bugger all.

The C128 was a huge mistake for at least 2/3 of their market worldwide, nobody here wanted an all-in-one keyboard only style form factor CP/M capable for £280, it was a product looking for a market. C128+1571+1902 bundle was more than the price of a 512k Amstrad PC1512 actual PC clone once you added all the bits needed to use CP/M (1571 disk drive and 1902 80 column EGA style monitor = £800 total price vs £500 for the 512k Amstrad PC compatible with mono monitor, £600 with colour and EGA).

Without a replacement for, or even update to, VIC-II and SID for a true 2mhz 128k C64 successor it was also a failure as an upgrade to the C64. You could actually buy an Amstrad PC1512 AND A C64 bundle in 1986 for less than the price of a C128+1902+1571. VDC is useless as it needs the EGA style monitor frequencies and nobody is going to spend £250 on a 1902 + £280 for a C128 home computer even if the VDC was suitable for home gaming mid 1980s flavour.

Still, it was the most successful 8bit mistake of all time though at 4.5 million units sold in various guises, that's not far off the Amiga 500 to be fair and more than the 520STFM IIRC. I guess people expected miracles from coders that were never going to happen due to the design and lack of VIC-II/SID 2mhz capability just shoved in there. Imagine if your Megadrive suddenly had to run at 3.5mhz if you wanted to use it on a TV!

At least the LCD was top of the class, but some idiot at Commodore took the advice of a Tandy or TI sales rep and cancelled the LCD project as in the words of the rep 'there is no market for portable computers'. Commodore had acquired Eagle Pitcher so the manufacturing cost for the LCD screen for Commodore LCD was a lot less than any competitor.

[r.cade]

Originally Posted by ImmortalA1000

Still, it was the most successful 8bit mistake of all time though at 4.5 million units sold in various guises, that's not far off the Amiga 500 to be fair and more than the 520STFM IIRC.

There is absolutely no actual evidence of this 4.5 million number. Somebody, somewhere, sometime... made it up out of thin air.