I'm not sure what's making people so uppity about such a simple observation. It's like I insulted somebody's mother...
I'm sorry, I think you're missing something. All you prove here is the fact that, clock for clock, the G5 is as fast as a Penryn. So what ? AMD's athlon was clock for clock faster than any P4 that came out, but it still was destroyed when S478 came out. I think the old Athon was even faster then Athlon64, clock for clock. Does that mean the original Athlon still "holds its own" today 🙄.
Have you forgotten why the G5 failed ? It didn't scale properly, remember, because it was way to hot (liquid cooling ring a bell ?) Penryn scales fantastically, so it is definately the better technology. performance per Watt, move over G5 😉
I guess what I'm missing now is what your point is... I said that G5 holds its own performance wise. I think you're agreeing with me there.
I'm talking about performance, not heat. Of course I'd expect a 45nm G5 to be much lower power than its aged cousin was 2 process generations ago, and maybe it could even do without the liquid cooling, but hey-- that never happened. I'd also expect that with all the extra transistors and the improved process in the Penryn we'd see improved performance and reduced heat, but we don't.
Fair enough, but with the lackadaisical way IBM and (especially) Motorola advanced the PowerPC, if Apple had stayed with the chip, we'd likely still be behind the curve compared to where we are now with Intel and would remain so pretty much forever.
So we might have faster PowerPC-based Macs, but they'd still be slower then our current Intel-based Macs and either cost us more or cost Apple more (so they'd be even slower at updating then they are now).
That's the assumption I've been hearing all along, and I think the numbers refute that. There are a lot of reasons to have switched to Intel, but raw performance doesn't seem to have been one of them.
Frequency of updates doesn't seem to be one of them either. Comparing the most stripped down system, the Mini, I'd say cost wasn't a reason.
Just to be clear, I don't expect doubling the number of cores to double performance when all other things are equal. Usually you get something like 80% improvement even when the task at hand is sufficiently threaded. I'm just saying Intel had to put a lot of time, silicon, design and process engineering into making up that difference while also benefiting from improved memory technology, etc. Despite the marketing glitz, switching to Intel didn't put us on a whole new performance curve.
Right, if you make silicon more colder, it will run more faster. Overclocking relies on a combination of this property, manufacturing margins (conservative specs to improve production yield) and reliability targets. If Intel thought enough of the 3.2GHz chips could reliably run at 4GHz with existing heatsinking they wouldn't throw away the profit of adding a higher performance bin.