HD For Indies

UPDATED-see bottom

Just watched the Apple WWDC Keynote by Steve Jobs.

After all the hyperbole leading up to this, it starts to make some sense. Apple wants and needs viable, low heat processors for portables - the G4/G5 line wasn't going to get there. They also need to continue to scale up the G5 processor's performance, and keep thermal issues relatively in check, which seems to be a problem (who ELSE has liquid cooling in their mainstream desktops? I mean, really?) There were probably some issues with multiple dual core systems as well - how would IBM feel to be placing iSeries and pSeries systems up against PowerMacs and XServes at a fraction of the price based on very similar silicon?

Apple is going to do a gradual, easy transition for developers and users. My earlier concerns about panicked developers and confused buyers aren't as bad as I thought. Rather than announce "We're switching, here's the new machines for sale (consumers), start developing today (programmers)!" they are going to take a couple of years to do it, and new hardware won't even ship until the second year (or almost that point).

Developers are getting plenty of heads up warning to port their code, even if they need to port from Metrowerks' CodeWarrior over to Xcode. Consumers are being told in plenty of time that new hardware is coming, but your current hardware and near-term purchases won't be affected. For high end users, even their medium term purchases (in a year or more) won't be affected. With universal binaries and Rosetta, developers won't be panicked to get their code working smoothly on Intel hardware, unless they develop with Metrowerks CodeWarrior, and/or have a lot of G4/G5/Altivec specific code. For developers with a lot of Altivec enabled, G4/G5 code done in CodeWarrior, they had better get to work porting to Xcode 2.1 sooner rather than later, as the first "MacIntel" (Intel based Macs) are likely to ship in a little under a year from now. Fortunately, since the high end machines are rumored to be last to be switched to Intel architecture, heavy performance apps (those most likely to be G4/G5/Altivec specific code laden, and possibly developed with CodeWarrior) will have about two years to port their products - plenty of time, IF they are willing to dedicate the manpower to it. Or they may bail on the Mac platform due to this requirement. Wait and see, wait and see. It'll be interesting to hear developer reports in 3-4 months as they get into their porting efforts, and we see how hard it is to port from Altivec laden CodeWarrior to Xcode 2.1 for MacIntel systems.

Universal binaries will allow a single, double clickable application to run on either Intel or PowerPC based systems, so developer support efforts are kept as simple as possible under the circumstances, above caveats aside.

The transparent Rosetta technology will allow existing, PowerPC based applications to run on Intel based Macs without too much difficulty, but with some caveats - I read that G4 and G5 specific code, as well as Altivec specific code, will not run under Rosetta - thus endangering current versions of Final Cut Pro, DVD Studio Pro, and perhaps Motion and Soundtrack Pro from running as is successfully under this new "MacIntel" hardware.

The soon-to-be-available Developer Kits will be a boon for those wanting to port their code - they'll have until the end of 2006 to use the Intel processor based Power Mac cased hardware to develop their code on at $1000 apiece.

From a consumer side, nothing will happen for nearly a year - in the meantime, I expect dual core G5, PCI Express based Macs to ship in 2006 as we await the upgrade for the high end desktop machines to be Intel based sometime in 2007. I would also expect faster G4 based PowerBooks to ship later this year or early next year.

Considering that it'll take time for apps to get ported to Intel (although Apple seems to be making excellent strides to aid in porting efforts, witness Mathematica's two hour port to get a running version), the first Macs to use Intel chips will be the Macs typically used for non-performance critical applications - and I'm betting that'll be iMacs, Minis, eMacs (if that line continues), iBooks, and PowerBooks (and that's the order I'd do it in). PowerMacs and XServes would be last, sometime in the latter portion of 2007 is my guess. For users merely needing MS Office, web, email, photo & simple iPod/audio needs, these new "MacIntel" machines will be fine for their needs. The higher end machines will ship as/when the new software will be ready for them, hopefully. Again, a well thought out strategy.

Apple will have their hands full developing new hardware to work with new processors - what will happen to the high speed bus (HyperTransport) that the G5 brought us? I think it'll go away, even though AMD was going to support it with their procesors (in fact they licensed it to Apple, as a reader informed me).

It's interesting that Apple went with Intel over AMD. AMD is the hungry upstart, and uses HyperTransport as well (that high speed bus the G5 uses), which would have aided in their hardware transition by using similar technology. But AMD is also the lesser player; and perhaps it was because Apple has grown weary of being a minor client to second tier vendors that they just decided Bag It, Go With The Big Guy and Be Done With It. By partnering with Intel, they know that they'll have leading edge tech at bulk pricing discounts. Perhaps IBM's excitement to make volume deals with Microsoft, Sony, and Nintendo played into this and burned Steve's ego overmuch; we'll never know.

So again, for the HD For Indies crowd, it's likely that a PowerBook would be their first relevant machine to be ported. And that's not likely to be any sooner than April of 2006, circa NAB. As Apple usually launches new versions of Final Cut Pro at NAB, I wouldn't expect them to launch a "MacIntel" compatible version at that time, but they might. More likely they'd do it at NAB 2007, when Intel powered Power Mac towers would be likely to be not too far off. Wait and see, wait and see. One the one hand, if porting is as easy as they make it out to be, NAB 2006 seems likely. On the other hand, FCP (and other portions of Final Cut Studio) lean so heavily on Altivec code it might be a substantial porting project to achieve performance parity on the new hardware.

One of the biggest concerns about the future of HD editing on Mac has to do with Altivec - Apple's vector processing unit that does exceptionally well at pixel and audio types of processing. Altivec has been responsible for a goodly bit of Apple's realtime capable heavy lifting in recent versions of Final Cut Pro, and without it, there will be some ground that has to be made up. Intel doesn't have a (truly) matching technology (and the portable stuff won't have the SSE3 functions anyway), so one of the inherent advantages of the OS X/PowerPC architecture will be removed. It'll be interesting to see what changes are made, or what progress doesn't happen, or what differences might occur between PowerPC and Intel, in a Universal Binary version Final Cut Pro 7 (or whatever it is to be called in 2007). (See here for some further info on where Final Cut Pro 5 stands for porting to Intel.)

But in the meantime, for HD editing, these new announcements won't possibly change anything for at least a year. If you think you'll need a machine for editing in the next 10 months, a PowerPC is 98% likely to be your choice if you want to edit with Final Cut Pro. It won't be until spring of 2007 that I predict it'll be worth pausing in your purchasing decision to decide whether PowerPC or Intel is the way to go, even if those Intel based high end desktop Macs aren't available yet.

Keep in mind, you can plan for the future all you want, but if you need to edit next week (or next year, or possibly even up to mid 2007) Final Cut Pro will run, and probably run best, on a PowerPC based Mac. And even after the "MacIntel" based machines ship, it's guaranteed that Final Cut Pro, DVD Studio Pro, Motion, and the rest of Apple's professional tools, as well as Adobe's (After Effects, Photoshop, etc.) will ship their latest versions for YEARS TO COME in Universal Binaries, capable of executing on both Intel and PowerPC (G3/G4/G5 based) hardware. And since the "on the front lines" lifespan of a high end desktop is usually only a couple of years, I take this to mean that if you're purchasing this summer, Monday's announcement DOES NOT AFFECT your near and medium term purchasing plans.

So while it's nice to plan for the future, if you need to work in the here and now, and even in the medium term future, a G5 and Final Cut Pro 5 will work as they do now, and will be viable and software upgradable for years to come. I wouldn't even anticipate the possibility of not being able to upgrade a G5 based Mac to the lastest version of Final Cut Pro until at least version 8 or even later. So don't sweat it. Buy whatever you were going to buy this year.

There is also the outside possibility that, using Intel's forthcoming paired motherboards and processors, that one might be able to run multiple simultaneous operating systems on a single machine. Thus WinXP/Longhorn in one memory partition, and Tiger "OS X86"/Leopard in another, allowing switching users to keep using their existing software on this new Apple "MacIntel" hardware. But that's a maybe, and quite likely not an up-front supported feature. If ever, likely to be a Leopard/Longhorn combo down the road. But a fun idea, and not entirely impossible. Wait and see, wait and see.

OK, that's as much after-three-beers analysis as I'm up for tonight. What a day!

-mike

UPDATE, written the next morning - as a commenter pointed out, supposedly not just the OS, but all PROJECTS have been dual platform for the last several years. This would imply the possibility that there is a Final Cut Pro existant in the dark corners of Apple that already runs on MacIntels (Intel based Macs - just cuz I like the word I made up).

So it's a matter of what Steve defines as a project - is it chunks of OS code, like QuickTime 7 or Spotlight? Or does it also imply other software products as well? In the end, I think not. Steve was showing iPhoto when he was discussing Rosetta - therefore it's still PPC, unless they wanted to hide the native Intel based iPhoto for some reason (which I doubt). If they had some of their apps already converted to Intel, they would have shown us. Definitely. Even for something like Mail, that has no funky Altivec code. So I'd interpret this to mean that by "projects" he meant operating system parts, not "products" which would imply other software as well. Correction: a sharp reader pointed out that iPhoto was in the general demo, and it was Photoshop and Word demonstrating Rosetta. So it could well be that Apple's applications have been kept running on Intel. So disregard the above

As for the difficulty of porting, Apple made it seem as if it would be fairly easy to port to OS X86 (Intel based OS X, again using my phrase just for fun), so perhaps we'll see an Intel optimized Final Cut Pro at next NAB after all. It doesn't seem entirely unreasonable, and Apple would be incentivized to "lead the pack" by having it's apps out native first. The timeline is right - so I hereby predict FCP 6 will ship as a Universal Binary. I'm just not saying when it'll ship. ; )

The good news is that we know for a fact that the Big Apps we use - Final Cut Suite, Adobe's toolset (CS & After Effects) will positively, definitively be available for G4/G5 Macs for YEARS TO COME, and the Intel optimized versions will be available promptly, or at least in a reasonable timeframe.

The biggest concern I have is that there might be some defections from smaller companies that make plugins and tools that aren't using Xcode (the tool to make universal binaries) and will decide it's too much effort to port their code and support two hardware platforms. One of the good things about this move, however, is that eventually, developers will only have to support one hardware platform between Macs & WinTel boxes - they'll only have to optimize their code for one processor in 5 or more years.

As for processor costs, I've read that PPC chips are cheaper than equivalent Pentium chips. Apple will just have to bite the bullet on that one. But hopefully SOME of that increased cost will be defrayed by the use of yet more standardized parts that are available from plenty of vendors at reduced costs. Macs are already using standardized ports, memory, drives, etc., but some motherboard level stuff might cost less. Not enough to entirely defray the higher chip costs, but some. More important than the chip costs, however, will be the satisfaction and comfort on knowing that they'll be on par with whatever the other PC manufacturers can offer in terms of speed, utilizing the latest Pentium chips.

The current G5 holds up pretty well against Xeons - close enough to make Macs a valid choice. But the chips that IBM and Intel are expected to have in a couple of years were looking like there would be a significant difference - Intel's roadmap lifts away from IBM's in terms of performance. And that, I think, is one of the biggest things driving this change, especially for portables and low end machines.

-mike