While looking up information on IBM POWER chips, I ran across a nice page which listed a ton of microprocessor information. The real gem of the page wasn't relating to the information I was seeking but rather a blurb about John Sculley talking to DEC's President about using Alpha chips to replace Motorola 68K processors. I knew Apple looked at several different companies for chips at the time but it is interest to know why any relationship with DEC never happened. Apple history would have been radically different had DEC been more enthusiastic about the wonderful Alpha processor they had engineered during these discussions.
http://www.cs.clemson.edu/~mark/architects.html
(scroll about half way down or just search for Sculley)
Yes they'd be up a creek without a processor. The alhpa never really was a serious alternative. Yea it was fast, but the migration would have been much more difficult.
For its time, the Alpha processors were extremely high performance pieces. The 21364 chip (EV6 line) arguably contains the most efficient CPU design to ever grace a computer. The Alpha ISA is also the best 64 bit implementation I've heard of. DEC/Compaq/HP management never took the processor seriously. The Compaq/HP merger is effectively what killed the Alpha as HP was too invested in Itanium, not because the chip was slow or arcane.
The migration from 68K to Alpha would have been the same basic road as 68K to PPC Mac users have witnessed. A 68K emulator for Alpha would have been a key part in the migration just like the 68K emulator for PPC. In fact, the Alpha line would have had an advantage over the PPC in the long term considering Apple's OS strategy. Right now the 64 bit migration for applications is beginning with OS X 10.4. Had Apple gone with the Alpha, once OS X arrived it would have been fully 64 bit from the very beginning with the only 32 bit addressing applications running inside the classic environment.
I have to disagree there. Motorola bent over backwards in both design and support to make it easy for developers to port their code over.
Apple didn't have enough volume to save Alpha. I'm not saying it was a bad CPU, but it was WAAAAAAY too expensive to be a serious contender in the personal computer market. The affordable Alpha processors performed far below equivelant PowerPC processors.
Do you think the SPARC platform was ever an option to them? I know it's open source now, but I don't know if it was open source back in 1993.
Until the Ultra series I don't think the Sparc line was all that much better than the 68ks, for desktop use generally. I'm not positive when Sparc went Ultra or PCI, but the Ultra 1 still had Sbus.
All the UltraII systems had PCI available. In fact the UltraSparcIIi cpu had a PCI controller integrated intot he CPU itself. It was one of the first to do that, though nowdays it's becomming more common.
As for competition, the ultrasparc CPU's tended to kinda suck at interger and kick butt at floating point. Not really the CPU you want for a personal computer. Also they were usually a step or two behind the times in both performance and manufacturing process. I think in another couple years, Sparc will be as dead as Alpha, save for the limited legacy embedded platform use.
It looks like all of the little processors are dying off. MIPS, SPARC, and Alpha are all sentenced to death.
I agree with you about sparc and alpha, but the mips processor still has quite a future in embedded technology. All linksys wireless products are shipping with a mips processor.
And SGI still sells massively parallel systems with MIPS CPUs. And Sun still sells massively parallel UltraSPARC systems. Alphas are not totally dead, but are being slowly killed off.