Amiga Custom Chips

[bloodline]

Does anyone know why Commodore kept the Amiga custom chips in three separate packages? I'm sure they could have save a ton of money by reducing it to 2 or even one chip!!

If you look at the original packages in the A1000, the later amigas (A500, A2000, A3000) kept the same form of package... I'm sure after 5 years Chip technology could have shrunk them to less chips!!!

[vortexau]

By the time of the never-quite-delivered BoXeR Amiga-replacement that ALL-in-ONE Chip Form had actually come to pass.

You have to look at the timeframe. The ORIGINAL Custom Chips existed as whole Breadboards requiring a complete Airline Seat for transport in their Prototype-Form in 1984.

The BoXeR was being developed in 1996-98.

Early Mobile Phones were the size of Cordless Phones, only LARGER, in early '90s timeframe.

Early GVP Accellerators+SCSI+RAM required two physical boards. That's why there is a W-I-D-E gap between the CPU Slot and the first ZorroII Slot in A2000s.

[Colin_Camper]

Good points.

Also you have to factor in the pin count. Reducing the chip count may result in more pins on the ULA/FPLA this may considerably bump up the cost of the device and would also lead to higher PCB FAB costs (due to req for more layers and harder p&p)

[mikeymike]

Also the complexity of the resulting 'all in one' chip makes it much more difficult to maintain and expand.

Chips, like software, should be kept as streamline as possible to perform their desired operations as efficiently as possible.

More complex chips waste more energy, more heat.  More complex software becomes more bloated.

[bloodline]

All good points!!!!

[sgm]

IIRC,

there was the additional problem of re-designing the chips using the CMOS technology from the NMOS used in the original chipset. As you probably know, NMOS is more power hungry than CMOS, and it doesn't scale well.

[olsen]

sgm wrote:
IIRC,

there was the additional problem of re-designing the chips using the CMOS technology from the NMOS used in the original chipset. As you probably know, NMOS is more power hungry than CMOS, and it doesn't scale well.

I'd like to add that the Amiga custom chip set was a so-called "full custom design". The kind of flexibility VHDL grants you today was still not within the reach of the designers back in the early 1980'ies, and Jay Miner reportedly designed the chips "on paper". Thus, in order to shrink or integrate the chip set, it would have had to be in a form that would lend itself to adaptation. That sort of thing is possible today: there are companies which will literally reverse engineer your design and cast it into a more malleable shape. But I daresay that this sort of thing was not possible until the mid 1990'ies. And after that there wasn't much of an incentive to do this any more. Even the Amiga custom chip did age and did fall behind technical development, eventually.