Why Apple stayed with PPC so long!

[toRus]

Actually, Apple stayed with PPC because PPC was better than Intel/AMD/Cyrix chips. And at the time that really mattered. MacOSX might have been developed x86 compatible before the 2005 Intel transition/regression but mainly because "it could be done" rather than the superiority of the technology. The performance benefits came long after the G5 was introduced and the PPC project was abandoned. Apple have chosen their companion and has stayed loyal up to now. This offered many opportunities but also had many risks. Apple simply wanted out of the CPU business and motherboard design. Time (not just 5-10 years) will tell whether this was a wise decision.

[toRus]

This was debunked by several benchmarks. Care to restart K8 Athlon 64 vs PowerPC 970 and K7 Athlon vs PowerPC G4 debate?

Benchmarks are just biased statistcs. And you do know what they say about statistics, don't you ? It's interesting how many people still interprete the G5 vs Pentium benchmark fiasco the wrong way.
Anyway, the PowerPC looked good but there was too much animosity between the partners. Apple failed to push it hard in the desktop, Motorola was about to restructure and IBM underestimated the Mac OSX momentum.

[toRus]

And they did that with all the "benchmarks" that Apple chose to prove that per mhz the PPC could beat a pentium.  Whether the benchmarks truthfully represented performance differences in the real world for the average user or were contrived tests that worked better on PPC architecture is another argument.  Jobs just had to run a few photoshop filters that worked faster on PPC to prove his point to the masses that PPCwas indeed faster.  Nowdays the hardware point of difference is largely aesthetic, which works better than  ever to get people to get a mac.

This is a practice x86 chipmakers do for over 2 decades. When the G5 came out, innocent x86 consumers complained because Apple's benchmarks didn't use Intel's heavily optimised compiler instead of gcc. They really thought their hardware was better than what it was in reality. Instead of questioning why their hardware was noticeably slower when not using Intel's C++/Fortran they blamed Apple (just if Intel's compilers were available in PowerPC or gcc wasn't already more optimised for x86 than ppc).
Intel has done so much false advertising (not only against other platforms but also aginst AMD) in the history of computing it's ridiculous. MMX/SSE were worse than 3DNow, let alone Altivec but prevailed and thus evolved because Intel pressed and used monopolistic practices. If using Altivec vs SSE benchmarks via Adobe filters Intel is guilty as well since most benchmarks and OS/application features depend on them today (even to a greater extent).