Snow Leopard: Apple's "Big Bang" for Next Year's Developers' Conference?

(Revised June 16, 2008) Yesterday, Apple announced its next version of OS X, but it did it so quietly that the announcement seemed more like compensation for a mistake (a premature press release from Apple Canada) than a planned event. The new version, named Snow Leopard, is intended to improve performance and stability rather than introduce lots of new features. However, a new feature appears to be "priming the pump" for some major new hardware announcements at Apple's 2009 Worldwide Developers' Conference next June.

It's called "Grand Central," and it appears to be a rewrite of those portions of OS X that aren't multiprocessor- (or multicore-) aware. The issue is that we've gone about as far as we can go with increasing clock speeds in order to get better performance, so the approach favored by Intel, AMD and just about everyone else is to add more cores to each processor. However, unless the software is designed to take advantage of multiple processors/cores, you don't get any performance advantage. Even in those cases where the software is fully multithreaded or multiprocessor-aware, the performance doesn't scale linearly as more processors are added; typically, the second processor or core only improves performance by 80%, and adding more processors results in even smaller incremental gains.

Apple claims that Grand Central will make all of OS X fully multiprocessor-aware (in essence, fully multithreaded,) and it will also make it significantly easier for application developers to write fully multithreaded software. Intel's Nehalem family of processors is scheduled to be released in Q4 2008, starting with four-core/eight-thread models for servers and high-end desktops. The performance kick from Nehalem over Intel's current Penryn generation of processors is expected to be big--approximately 20-30% overall, but closer to 70-80% for multithreaded applications.

I expect to see a dramatically refreshed, or possibly even completely redesigned Mac Pro at or prior to the Worldwide Developers' Conference next June, released concurrently with Snow Leopard. Apple should be able to stay with its existing two-processor/eight-core designs, but get big performance boosts at lower power consumption with Nehalem. At this point, unless you need to get a Mac Pro in the six months, you're probably wise to wait until next June for a Nehalem-based system. Snow Leopard will provide significant performance advantages even on current-generation systems, but it will need Nehalem in order to take full advantage of what it will be able to do

The Real News from Apple

Today's announcement of the iPhone 3G at Apple's Worldwide Developers' Conference has been rehashed and dissected by reporters and news anchors all day. However, the iPhone announcement itself was fairly anticlimactic; the new features had been well-covered in leaked reports prior to the announcement. (In fact, the lack of new features beyond 3G and GPS was somewhat surprising.) Even the dramatic price drop had been foreshadowed by widely-publicized reports.

I think that the real news wasn't the iPhone announcement, but what came before: A slew of application demos, plus the announcement of Apple's MobileMe service. Let's take the applications first: While they were being demonstrated, some of the live bloggers griped that they were tedious and just went on and on, but that was the point: In just a few months, Apple has built a bigger, more productive developer ecosystem than Symbian has been able to do in years, and no one else (including RIM, which just recently launched its own developers' program) is even in the same ballpark. Even Microsoft's smartphone platform can't deliver applications with the quality of experience of those designed for the iPhone.

Over and over again, the story was: "We've covered all the bases." Enterprise applications? Check. Exchange integration? Check. Desktop application support? Check. Location-based applications? Check. Games? Check. All of this was on top of the basic capabilities of the iPhone, now fortified with sufficient 3G speed to make heavily data- and media-centric applications work.

The other part of the story is MobileMe. At its heart, MobileMe is a centralized storage and synchronization application that allows iPhones, Macs and PCs to be synced to a single, central database that's managed by Apple. Now, all of the capabilities of MobileMe are available in some form from a variety of vendors, but they don't necessarily work very well. As a long-time ActiveSync user, I can tell you that getting my PC notebook to stay in sync with my old Windows Mobile PDA and current Windows Mobile Smartphone (let alone my MacBook) can be an exercise in frustration.

MobileMe is aimed at two targets: The large body of Windows Mobile users who are frustrated to death with ActiveSync, and everyone who has held off on buying an iPhone because it doesn't have the "it just works" syncing capabilities of RIM's Blackberry. It's far too early to tell just how well Apple has implemented MobileMe, and it may very have its own frustrations and limitations. However, it has the potential to be a very appealing alternative to Microsoft's and RIM's offerings.

The one big frustration that I have with the announcements is that the iPhone 3G still doesn't have video camera capabilities. A 3G iPhone with the video capabilities of, say, a Nokia N95, would be a multimedia killer product, and I'm still not giving up hope that Apple with do something in this space in the future.