Posted by aaron at 08:01AM, Monday, December 06th, 2004
Oh no, MacTel, or, my take on the Apple PPC to Intel thing
I've read a lot of buzz about what this means for Apple and for its users, and (I know this is a shock) much of it strikes me as pretty naive.
I've used Macs since 1989. I've written software for Macs since 1989. I've been through the migrations from System 6 to System 7, from 680x0 to the first PowerPC Macs, and from OS 9 to OS X. I've learned a few things in that time.
Hardware-wise, I can see some good reasons to go to a Pentium architecture: it's standard, and its performance has been scaling better than PowerPC lately. I can also see some reasons not to: I've been reading for years about how the G4 and G5 have the potential to scale much better than Pentium, with the addition of stuff like multi-core architectures and easier multiprocessor support, and their historically superior performance-to-power-drain ratio. Anyone remember how underpowered PC laptops were all of, oh, two years ago, while my then-current PowerBook was nearly as fast as anything Apple offered?
Also, the whole game platform thing makes this a strange-sounding decision. Microsoft and Sony are both going with multi-processor G5s for their new systems -- presumably because of the sort of vector-based calculations that AltiVec does so well, and that Mac users have historically touted as making their platform more powerful for Photoshop and music apps and things like that. G5s are also cheaper than most flavors of Pentium, which is great for a consumer box. Clearly, Apple's pissed that IBM won't give them the time of day, and that it hasn't been able to deliver a G5 usable in a PowerBook. But now they're going with a commodity vendor, who's not going to give Apple any custom chips; Apple will have to deal with what Dell and all the other PC makers get. So why not deal with what Sony and Microsoft get for G5s? I mean, presumably IBM's going to have to make a low-power high-performance G5 for the next portable PlayStation, right? It's not as though the PowerMac G5 is getting bashed in speed tests, either.
The other major hardware issue is Apple's sales figures. G5s are doing pretty well. The current iMac G5s are hot. I'd thought about buying one of them soon -- but why should I, now, if the entire PPC platform is going to be obsolete in a year? Apple's making a lot of money on iPods right now, but their bread and butter still comes from CPU sales, and I can't see how those won't be killed by Jobs' announcement. Making the transition so high-profile without a product to ship is a stupid, stupid move, and reminiscent of the Amelio years. "Just wait for Copland!" Those were great for Apple profits and market-share.
One aspect of this deal that's not getting a lot of press is the political one. Intel's a pretty red company; their execs donate mostly to Republican campaigns. BuyBlue.org rates Intel as significantly redder than IBM. Intel CEO Craig Barrett came out in an issue of Vanity Fair last year as supporting Bush for President. That may not be a big deal for a lot of people, but for some of us who tend to wield our buying power as true capitalists, it's important. The Republicans don't need any more money from me, as far as I'm concerned, and it will be tough for me to fork over a big chunk of change for the new, more-expensive Intel processor in a Mac, so that Barrett can donate to politicians, like Bush, dedicated to rescinding First Amendment rights, instigating violence against any progressive judges, tearing down the separation of powers, and demonizing me, my friends, and my marriage.
From the user standpoint, a lot of commentators are out there saying that it's "stupid" for the users to give a shit about the underlying hardware. That's patronizing, insulting, and wrong-headed; those guys should go back to Slashdot where they belong. Look, I remember the days when us Mac users gleefully pointed out the arithmetic errors in the early Pentiums. That was a hardware issue that certainly affected some users. I like having a fast laptop that doesn't set my lap on fire. That's another hardware-architecture issue that directly affects users. What I'm most concerned about in this switch is my software. Apple's been saying it has some great emulation technology which will allow a large number of existing apps to run with decent speed on the new Pentium Macs. That's great. But I'm not worried about applications like TextEdit, or Xcode, or AppleWorks -- any of the stuff written by Apple and following their APIs and guidelines to the letter, which probably makes it super portable.
I'm worried about stuff like Peak, and ProTools, and Finale, and Photoshop. I'm worried about all the specialized software I use, much of it written by small, specialist companies, who go around the usual APIs, and make lots of low-level system calls, and do a lot of optimization for AltiVec. Sure, anything AltiVec-optimized but that also runs on a G3 might run in emulation -- but it will be unbearably slow. I'm also worried about my drivers and system extensions for legacy hardware, since from what I've read, the emulation layer won't do diddly squat for all of that: it won't support any current kernel extensions, or any code that inserts a preference in System Preferences. That's every audio driver I own. That's my scanner. The AltiVec stuff is going to be a big performance hit for Photoshop, even before the emulation hit. All of this sounds to me like it will make for a transition more like that from OS 9 to OS X than from 680x0 to PPC.
My transition from OS 9 to OS X was not a happy one. For programming, it was great -- once all the hurdles with X11 support and networking had been smoothed out. I assume the Intel transition will be similar. For the applications I use for other stuff, mostly music apps, it was absolutely horrible, and had me seriously considering switching to Windows for music notation. Finale, the notation app I use, wasn't supported under OS X until a little over a year ago. It didn't work in Classic -- at least, not if you wanted it to talk to MIDI, which is the whole point. There was no other OS X notation program that I tried that didn't suck for the kind of stuff I do. I still don't have a software MIDI management system as flexible as what I had under 9.
I was also using an Audiomedia III card from Digidesign for audio recording. I was using Peak and the bundle version of ProTools with it. It rocked under 9. It was years before there was a driver that worked for X, and by the time there was, Avid had end-of-lifed support for the card. Basically, I got about a year of use out of the card since I bought it -- and that was on my 6500 running OS 8. After that, I switched to OS X, since I'm a programmer, and I don't have the money or space for multiple fast machines. No ProTools in Classic or X. Then Digidesign released a driver -- which was incompatible with Jaguar, which had just come out, if I'm remembering this correctly. At this point, although the card "works" with Panther (but not Tiger), I've ripped it out from my machine, since it generates a lot of errors in the Console, and only works with one application at a time -- and an arbitrary one at that.
Of course, I've learned from that experience, and won't be buying anything from Digidesign again. That's fine, since there are hardware alternatives from places like MotU, and software ones from BIAS, etc. Finale is more problematic, since I haven't found a single viable alternative. (And, no, Sibelius is not an alternative to Finale for the stuff I do.)
I also wonder about the graphics market, which is a pretty big segment of Mac fans. Last I heard, there still wasn't a decent version of Quark for OS X. Adobe just released a bunch of new versions of things like Photoshop and Illustrator; I can't imagine the performance optimization that their customers demand for those apps is going to be easy to implement for Mac on Intel, given the heavy reliance on AltiVec right now.
Overall, I'm sure the transition will be very smooth for people like my parents, who pretty much only use the bundled applications. For those customers who've long supported Apple, but rely on software solutions Apple doesn't provide -- musicians, graphic designers, hobbyists, photographers who use VueScan and Photoshop and can't afford a $500 software upgrade every year -- well, I'd like to see this transition go smoothly, but, historically, there's no precedent for that. The emulation layer sounds a lot like Classic to me, and Classic was only really useful for the kinds of apps that were easy to port to OS X anyway -- the kinds of applications my parents use, which all come bundled with the CPU.
Steve Jobs has done a good job of marketing to folks like my parents. But at this point, the long, hard-core Mac faithful have suffered enough under the OS 9 to OS X transition. If the transition to Intel is anything like it, I may be jumping ship and buying a Windows box for audio. People who've heard me bitch about Windows know how painful that move would be, but, if my computer won't run the software I want to use right now, what choice do I have? Apple's in very serious danger of alienating "the rest of us" for whom it once claimed to exist.
All content copyright © 2001-2009 the owners of http://www.circa75.com/