[sillyquiet]

> Tramiel never understood or invested in growing a platform or growing, expanding, intercompatible product line. Each Commodore machine under his watch was a brand new machine, not compatible with the others except for maybe a few peripherals.

But, iirc, very very very few computer manufacturers prioritized compatibility in that way in those days though, and for good reason, it would have be stupid expensive. Some manufacturers got around the problem (Commodore did this as a matter of fact) by incorporating all or part of the previous line in the new machines and enabling compatibility modes.

[timbit42]

> But, iirc, very very very few computer manufacturers prioritized compatibility in that way in those days though, and for good reason, it would have be stupid expensive.

You don't recall correctly. The only major brand that didn't prioritize compatibility was Commodore.

Maintaining compatibility limited the ability to add new features but it wasn't 'stupid expensive'. What is expensive is throwing out what you have and creating something incompatible from scratch. When you have hundreds of thousands or millions of units out there, not being backward compatible means you risk losing most of those customers.

Nearly everything that ran on the original Apple II ran on the IIe and IIgs, and nearly everything that ran on the Atari 800 ran on the XL and XE models. Nearly everything that ran on the original 1981 IBM PC can run on a modern PC compatible.

[sillyquiet]

Fair enough, I stand corrected, but I guess I meant maintaining compatibility when you change architectures is stupid expensive.

[timbit42]

With that take, "very, very, very few" means none. Who did that?

[cmrdporcupine]

The Apple IIgs was basically that. The IIgs had a miniature Apple II chipset onboard for backwards compat. C128 was also sorta like that.

But in both those cases, the ISA never changed.

Apple did also provide a card for the Mac that had an Apple II on it.

[sillyquiet]

Commodore, later on, when they put a whole c64 in the c128, but I take your meaning.

[rasz]

c128 was a case of idle engineers with no plan/strategy gluing unsold garbage components (graphic chip left over from earlier failed project and Z80) to c64 and scamming public with misleading name. 2 CPUs, 2 graphic chips, 2 monitor outputs, all in one package.

[timbit42]

Even without a C64 mode, the C128 is a cool system. Dual monitor outputs, 128K RAM, 2 MHz 6502, great BASIC, fast floppy drives. The only thing that would have been better would have been to add these things to the C64. Also nice: dual SIDs.

They could have added all of the above without breaking backward compatibility. This would have encouraged programmers to check for the enhanced features and use them if available.

Of course, long term, they would have needed to go with the 65816 and ultimately switch to another processor architecture.

I think if all of the 8-bit and 16-bit systems had survived until today, they would all be running on Wintel hardware but with their own operating systems, just like Apple Macs.

[cmrdporcupine]

Well, that's simply not true for the Atari 8-bit line (mostly compatible from 1979 right through to the XE series which continued right into the late 80s and even early 90s), the Apple II line (II, II+, IIe, IIc, IIgs, as well as cards that slotted into Macs), MSX and MSXII, and of course the 16-bit era with MS-DOS machines and Macintosh, Amiga and Atari ST lines.

Honestly, Commodore and Sinclair 8-bits are the outliers here? C-128 came quite late in the game.

[timbit42]

Commodore did have a fair bit of compatibility within the PET/CBM line.

The C128 is a good example of how Commodore did compatibility wrong. Instead of creating a 100% C64 compatible mode, separate from the C128 mode, they should have taken the C64 and added a second bank of 64K of RAM, an MMU, an enhanced VIC-III with added registers for the enhanced graphics (eg. colors, resolutions, sprites), a second SID, a 2 MHz CPU, an enhanced KERNAL, and an enhanced BASIC that had the same tokens as BASIC v2. While C128 mode was very similar to the C64 mode, it wasn't enough to run hardly any C64 apps and games without modifications.

I don't blame Bil Herd for what the C128 is though. It is a great computer, especially considering he was fighting the top levels of Commodore to get it out the door. If he'd had more support, it could have been better.

[cmrdporcupine]

I feel like the right way to do it would have been to have shipped a C64 compatibility mode in the Amiga, instead. An A500+ with a 100% C64 compatible mode capable of running a C64 in a window even, IMHO would have been great product that could have sold like gangbusters because the C64 was still a hot selling product right through the late 80s. I can't imagine it would have been that expensive to do given how well understood the C64 architecture was at that point.

Either that or have built the Amiga around a 65816 or similar instead of 68000. Like the IIgs but with even better sound and graphics.

[rbanffy]

> Either that or have built the Amiga around a 65816 or similar instead of 68000.

This would be a huge mistake. The 68000 was much more elegant and had a way forward, something the 65816 never had. Had they gone the 65816 route, there would never be an Amiga with a better CPU except, maybe, a faster one.

And the whole Amiga 1000/2000/500/600 was saddled with TV signal timing dependencies that made it more difficult to have better graphics and ended up having to play catch-up with VGA and Sound Blaster.

[rasz]

65816 at 7MHz would be faster than 68000 despite 8 bit databus. one modified for full 16bit would fly circles around motorola. Bill Mensch would have no problem extending and growing this chip as long as Commodore was buying. 65816 already has reserved mnemonic for expanding instruction set further.

[rbanffy]

The 65816 (much like the 6502 before it) punched way above its weight, but that'd not be enough in the long term.

There was a point neither IBM nor Motorola cared about making PPC CPUs for Apple - that's why Apple moved to Intel. Moving to a bespoke, Apple-only CPU, would only precipitate things by making a series of CPUs just for Apple. Not sure WDC would be able to rival the investment of the big CPU shops of the time.

If, instead of tapping WDC they went to MOS with a roadmap that could accommodate both Apple and Commodore, maybe history would be different. In either case, these are parallel universes I'd love to observe.

[rbanffy]

> an enhanced VIC-III

This chip wasn't ready when they were building the C128. It was later included in the C65.

But, apart from that, there is a lot Commodore could have done incrementally with the C64, making it iteratively more functional without breaking backwards compatibility that the company simply didn't want to.

> he was fighting the top levels of Commodore

The 128 shares a lot of sins with the Apple /// and comes from a very similar story. At least Apple learned something from the /// and made the //e, //c, //c+, and the IIgs (lovely, albeit kludgy, machine, which was an absolutely idiotic project that only took resources away from the Mac). Computer for computer, an LC with Apple //e PDS board was a smarter choice.

[cmrdporcupine]

IIgs was a lovely machine that was kludged and handicapped because they didn't want to take resources away from the Mac. They could easily have clocked it up 4mhz or higher and had a much more capable machine. But my understanding is they didn't want to eat into Macintosh market share.

But schools here in Canada at least were highly invested in the Apple II, and my school acquired a whole fleet of IIgs machines. They were pretty nice actually. If they'd clocked it higher it would have been a very worth competitor for Amiga and Atari ST. But they didn't.

They also priced it too high for what it was.

[rbanffy]

> handicapped because they didn't want to take resources away from the Mac.

They had resources. The IIgs had a ROM Toolbox, a lot of hardware and was, overall, a very complicated machine that took multiple iterations to be developed. Apple was trying to make a 16-bit Apple II since before the Mac.

That they didn't want it to take sales from the Mac, makes sense. The IIgs was an evolutionary dead-end. There was no successor for the 65816.

No matter what they did with the 65816, it'd still have segmented memory, few registers (somewhat aleviated by the single-byte address trick) and no way forward - no MMU, no FPU. WDC still makes them, and never made a 65832 or 65864. The ST and Amiga were excellent opportunities thrown away. Commodore could have the 3000 be the entry-level Unix workstation Sun would sell and, with that, gain a foothold of the technical and financial desktop market. Commodore had the professional video market thanks to the Video Toaster, but made machines where the Toaster wouldn't fit. Atari, OTOH, made a couple decent workstations, but never invested much in anything beyond gaming machines with tiny monitors. It found a niche in the Music segment thanks to its MIDI interfaces, the same way the Mac survived because if was at the right place to take hold in low-end publishing.

[cmrdporcupine]

I rode the Atari ST wave back then. Until I bailed and got a 486 and ran Linux in 93. Yes, Atari half-heartedly marketed the TT030 as a Unix workstation, but it never stuck. I still remember the headline of the little snippet in UnixWorld magazine about it "Up from toyland." Not taken seriously in the Unix market, really, and by the time they had a product, that wave (68k based Unix workstations) had crested anyways.

The real problem for both Commodore and Atari in that era (apart from moribund sales, small market, bad management, etc) was that Motorola was already working towards killing off the 68000 by the early 90s. I'm not sure what Atari or Commodore would have done if they could have held on another couple years. PowerPC ended up being a dead-end, too.

Yeah, I don't like programming the 816, either. But in that era it wasn't the worst. WDC made the 816 just for Apple, really. (And didn't do the greatest job imho, but whatever) If Apple had asked for it, they would have made an 832, added more registers, whatever. It would have been a hack... but so was x86.

Apple's problem was that until the LC (93? 94? 95? I forget. My mom bought one at my recommendation) they didn't have a reasonably priced low-end Mac to sell to people or schools who would have gotten something in the Apple II series in the past. They seemed to have lost a big part of the education market when they dropped/failed-to-advance the II line.

[timbit42]

As a Canadian I am surprised to hear schools, "were highly invested in the Apple II". In Ontario and any other province I've heard about, Commodore PET, VIC-20 and 64 were nearly ubiquitous. What province were you living in?

Even if clocked higher, the IIgs didn't have the video power to compete with the Amiga. No blitter, sprites, or hardware scrolling.

What Apple product wasn't priced too high? Always have been, since day one.

[sillyquiet]

ok, I stand corrected, but I will say that the Apple II line represented smaller iterations of the generally same architecture, right? i.e., Commodore I guess didn't iterate within the same architecture as much, breaking that compatibility more easily.

[rbanffy]

One of the key difficulties of making a faster Apple II was the time-critical routines involved in reading/writing floppies. It was not sufficient to throttle back to 1 MHz when running time-critical loops, but one would need to emulate the timing of a 6502 doing that. Commodore made a better 6502 for the C65, but it'd wouldn't work on an Apple II because it wouldn't have the same cycle timings.

The IIgs and the //c+ was faster and very compatible thanks to a crazy number of hacks in them to acommodate Woz's brilliance.

In retrospect, Apple should have released a Disk II+ that isolated the timings from the CPU and let software break. Commodore should probably have done the same.

[timbit42]

There was a fair bit of compatibility within the PET/CBM line. Outside of that, every system had an incompatible memory map.