[Bruce Abbott]

Originally Posted by hammer

The original intent for AGA's object manipulator was $20 DSP3210 with Fast RAM.

To put this price into perspective, all the MOS chips in the A1200 (Alice/Lisa/Paula/2xCIA/keyboard MCU) combined cost $19.85. Budgie cost $10, 68EC020 was $7.61, and the 4 DRAM chips cost $50.32.

At $20 the DSP chip would raise the retail price by ~$50 just for the chip itself, not counting support chips. The high speed static RAM chips wouldn't be cheap. This would certainly exceed the 'magic' £399 retail price that was a big part of the A1200's popularity.

The 'original intent' of putting a DSP chip in the Amiga was for video and audio processing, and perhaps for a 'software' modem. AGA's 'object manipulator' was the Blitter. AFAIK the DSP chip was not intended to take over this role.

A1200's Budgie has Fast RAM controller support baked in which is missing on A500's Zorro I edge connector.

This statement - while technically true - is a bit deceptive. The A1200's expansion was similar to Zorro I except for being 32 bit. It was the same idea as VL bus on PCs, a direct 32 bit CPU interface without any fancy protocols. This scheme was envisioned for the A1000+ too, but Commodore's European subsidiaries wanted at least two Zorro-II/III slots.

Lew Eggebrecht intended to restore the original AGA+DSP3210 bundle and Lew made sure the $20 argument is focused on.

I'm not sure about that. I do know that Lew Eggebrecht didn't think the Amiga's architecture suited 'high-end' machines. He thought an entirely new architecture using more 'standard' parts (including a RISC CPU) should be produced for that market. At the lower end he thought they should concentrate on improving the A500, which was Commodore's main money spinner.

The engineers didn't like this at all. They wanted high-end machines with super-powerful graphics that would beat PCs for years to come, and stuff like DSP chips for real-time movie playback etc. The fact that they couldn't achieve this in a reasonable time frame didn't seem to faze them.

Bill "IBM PCJr" Sydnes made A1200 to be another JR.

This isn't quite right. Mehdi Ali put pressure on Syndes to get the A1000+ out - AA or not - because Gould had publicly promised a new machine and was expecting it. Sydnes didn't have a lot of choice. Personally I would have been cracking the whip a lot earlier to get the AA chipset finished, even if some features were missing. It all came down to engineers focusing on what they wanted, not what was needed.

The 'A1000jr' didn't get released because Lew Eggebrecht sensibly polled Commodore's subsidiaries to gauge demand - which turned out to be zero. Everybody wanted something with enhanced graphics, not another ECS machine. That's when Commodore finally realized that they needed to get AGA systems out ASAP. When the pressure was put on the engineers did a good of the A1200. Just pity they hadn't done it earlier!

Reading Brian Bagnall's books 'Commodore the Amiga Years' and 'Commodore the Final Years' one common theme stands out - product releases were constantly slipping due to engineers not focusing enough on getting the job done. This was more apparent after Jack left, though not always. The A1000 was produced quite quickly, as were the A500 and A2000 once the rivalry was sorted out. However after 1987 they started way too many projects with dubious prospects and little urgency to get them finished.