The thing about the e600-derived PPC CPUs is that they were still based off of the original core of the G4. Even with an enhanced processor bus and a second core, would it have been able to be competitive with a Core Duo?
This’ll be a response in three parts.
Owing to Freescale’s own work on improving the architecture, one can look to the 1.5GHz Core Solo in the Mac mini, versus the 1.67 GHz in the 7447. Were the e600/PPC8641/PPC8641D to have gone anywhere on the Mac, the previous — Cole Solo versus 7448 — can’t really be compared in terms of performance and time of release, as the e600 would have been the most advanced 32-bit consumer-level offering at the time the Core Solo/Duos reached market.
That said, it is somewhat surprising no one ever tinkered with a PPC8641D (the “D” for “dual”, allegedly consuming less than 25W at 1.5GHz) in a non-embedded use-application
and ran Geekbench on it (ostensibly either in Linux or BSD). That would have been something worth seeing.
Perhaps in 2006, but would it have been able to stay competitive with the Core 2 Duo and the subsequent Core iX series?
Here’s part two.
This is sort of an impossible question to answer, owing to how research development sputtered after the PowerPC’s principal customer bailed out. We also know PA Semi found new inroads with multicore PowerPC architecture, including 64-bit applications, but the principl customer to bail on PowerPC cosplayed a corporate amoeba and devoured PA Semi, leaving us with cryptographically locked/“paired” iPhones/iPads.
Between Apple crashing the AIM Alliance singlehandedly and Apple halting continuing development of PowerPC architecture by any serious start-up it could afford to swallow whole, the entire R&D to have have come to pass is, frankly, unknowable to any of us.
But we can fathom a different Apple which showed a different kind of shrewdness (maybe in another multiverse), in which Apple continued with PowerPC development and PowerPC products, working truly in alliance with Motorola and IBM, whilst at the same time also selling a separate line of Intel based systems running the exact same OS. This would have put a lot of heat on Microsoft, as neither of them were developing an OS for PowerPC systems (that anyone is aware of, at least).
But then you had Steve, who we know to have had a vindictiveness when anything didn’t go entirely his way (witness: Motorola, 3GHz, ZFS, an obsession with hardware products not being user-serviceable, etc.). —_—
I highly doubt it. Not with Freescale pivoting the e600 core towards the embedded market, a market that would be far, far more profitable for them to pursue vs. Apple's share of the desktop market.
And for part three:
Freescale recognized there was an immediate need to be fulfilled in the embedded market (such as in vehicles and fabrication/automation), and that pivot was responsive in the face of losing a major customer from a business alliance (one already hobbled by Steve after banning Mac clones with the launch of the PPC 740 in 1997) — an alliance which no longer existed (because, again, Steve). The e600 was adapted for customers which needed the processing power for their embedded needs, from a company with a lot of chops already under their belt. A best-case business outcome for Freescale would have been to the means to sell to Apple and third-party clones, as well as the embedded market. That, of course, never came to pass.
And that's assuming Freescale could produce a CPU like the MPC8641D at the scale Apple would have wanted, on Apple's timeframe, at the price it would have wanted them (it's always been my impression that price was the reason why Apple balked on the 7448).
A bit ironic, once one considers how much of a retail pricing premium Macs commanded, even then, over their i386/x64 counterparts. As for the 7448, I’m not sure how this would have been a substantially more expensive CPU, owing to its refinement of an existing line which Apple had been using for roughly three years (or whenever the 7455 went on sale).
Apple’s time frame was, from the hindsight of it, a matter of saving face after Steve put unrealistic pressure on the 3GHz performance benchmark, before pivoting completely once having announced the Intel transition (some 18 years ago today). Steve was shrewd and frequently smart, but he was also impressively woodenheaded, often arrogant, and not entirely the near-term visionary I’m guessing he believed in his own headcanon he was.
And maybe I'm wrong since I don't know the details, but would the TDP and performance per watt of Freescale's post-7448 CPUs have enabled them to be put in a machine like the MacBook Air, or offer G5-level performance in a PowerBook? Recalling the original 2005 Keynote, it was clear that Apple wasn't interested in what was competitive now, but what Intel was going to have in its roadmap down the line.
Like you, I lack all the back-end stuff not privy to the public. But considering the way Motorola/Freescale developed the 8641D to function at rated clock speed, as roughly 25W (whether that’s TDP, or a sitting average, I honestly don’t know), coupled with PA Semi’s own development, the lower-power, multicore, and 64-bit tech was coming on line. It just wasn’t coming online by one, single, simple juggernaut wearing blue (the
other blue).
It would have taken a lot on the part of PPC development to delay or even prevent the x86 switch from happening: On IBM's part, they would have had to have the financial incentive and technical chops to focus on significantly reducing the G5's power consumption and heat output, while continuing to boost performance – at least enough to make a PowerBook G5 viable and make good on Jobs' boast of hitting 3 Ghz in the PMG5.
The “G5” was never gonna make fetch happen on portables. IBM never designed the 970 for that use-case. This was Steve’s folly in picking favourites in the disintegrating AIM Alliance: that was a strategic gaffe in the pursuit of a tactical win.
The AIM Alliance really was remarkable, even as Steve had no love for it (again, woodenheadedness, arrogance, farsightedness, etc.). He could have nurtured it when he was brought back aboard, but for whatever reason in his own mind, he chose not to. And maybe it derives from the same reason why he, with NeXT, ditched Motorola in favour of a NeXTSTEP which would run on Intel/i386 hardware: the hardware product mattered less to him than the software (and the technical ability to engineer control over that software, even if he wouldn’t be alive to see that come fully to pass).
Perhaps with more money and resources it would have been possible, but it would happen soon enough to address the growing performance gap between the PowerBook and Wintel laptops? On Freescale's part they would have had to do the same to make a potential "PowerBook Air" not just viable, but competitive too.
We knew Apple had been running OS X on both PPC and Intel architectures from the outset (no doubt a carryover from NeXTSTEP 3.3), and Apple could have completely supported running OS X on two major hardware platforms — one RISC, one MIPS, and putting the pressure on Microsoft. Then again, there just as well might have been a “““gentelmen’s agreement”” between Steve and Bill, when Bill threw nine figures at Apple in 1997, that Apple wouldn’t run OS X on multiple hardware platforms simultaneously.
We, the consumer, lost some choice in that arena, as OS X on either major platform had different customer bases with different specialist needs: customers with professional and semipro-level hardware and software developed by and for PowerPC/RISC gear (think MOTU and a whole slew of music and video production gear developed and made available over the previous dozen years); and customers with bog-standard Intel/MIPS boxes who needed to connect online, print to the department printer, and to tap into using scientific hardware designed only for Intel/AMD systems running Windows (which an Intel Mac could also do natively). We know Rosetta made on-the-fly code translation possible, as it was needed. Rosetta could, in theory, work in both directions. Apple, under Steve, chose to merge those two customer bases as we came to know.
And not just that, but they would have had to have a development path that would deliver CPUs competitive with Intel's roadmap. Given what we've seen from Motorola Semiconductor/Freescale, I wouldn't have had any confidence in that happening (even if we don't count rumored promised-but-never-delivered CPUs like the fabled 7457-RM and the infamous 7500).
Intel’s road map was simpler, because only one vendor/company was involved. The PPC roadmap was, meanwhile, divvied by as many as three different companies (Freescale, IBM, and PA Semi, with the latter, barely two years old, still teething in 2005).
Simplest won out, for worse or better. And once more, even more simple — by keeping it all internal — likely factored into Apple’s move to, ultimately, bring processor development in-house for Silicon and to draw on their years of development with ARM architecture (no doubt drawing on innovations delivered by the consumed PA Semi staff and their pre-acquisition intellectual assets).
Then again, I’m a dork and am talking out of my exit port. 🤓
