[X4U] PPC a superior chip to Intel?

[David Ledger]

>From: nk <netkat at comcast.net>
>a few weeks ago, a buddy of mine, who owns an intel Mac Book Pro said
>that, while his new machine is faster by the specs, he feels that the
>PPC was just a better chip.
>He didn't give specifics, just a sense overall, after having used both.
>Lately, I have been working on-site at a client who has me on a Dual
>Xeon Mac Pro. I can't say that it's noticeably faster (most of what
>they have me doing is CS3-based work).
<snip>
>Any thots on whether the PPC was a better chip than the (supposedly)
>faster Intel?

The PPC family _is_ better than the Intel x86 range of chips. But 
remember that we only ever got the base level chips in Macs. The PPC 
chips in the big iron hardware leaves any personal computer standing 
(but the CPU unit alone will cost more than a top end Mac). Likewise 
the Intel IA64 family is on a par with, or outstrips, the top end PPC 
stuff. Apple went with Intel x86 because IBM wouldn't advance the 
base level PPC chips as fast as Apple needed.

Intel x86 has the great disadvantage that it's been developed in 
parallel with Windows. There's never been a chance to start from 
scratch. It's still based on the old 8080 concepts. PPC was designed 
once the designers knew what the requirements were. The 8080 was 
designed for a different world and its successors are only viable now 
because Windows gave it that large market that made it worthwhile 
continuing development. Intel's excellent designers have managed to 
produce a chip with sufficient compute power at a low price, based on 
a far from ideal (for the modern world) architecture.

At the time the Intel 8080 and the Motorola 6800 came out, virtual 
memory meant using _more_ physical memory than the processor could 
address. When that changed to meaning _less_, some time after the 
advent of 16 bit processors, by chance the 68000 architecture handled 
it more easily than the 8086. The first Unix machines I used in 1983 
were 68000 based; Unix wasn't viable on Intel until ten years later, 
with Linux. * The 80286 had just about enough power for the extra 
software procedures required to give what the 68000 gave inherently, 
without crippling the throughput *. SCO had its Unix-like system 
before that, but it was severely limited.

* - * Told to me by colleagues in the mid-80s who worked on this 
stuff (not from my own experience).

David


-- 
David Ledger - Freelance Unix Sysadmin in the UK.
HP-UX specialist of hpUG technical user group (www.hpug.org.uk)
david.ledger at ivdcs.co.uk
www.ivdcs.co.uk