Unless you lived under a rock and have not been exposed to media (not likely since, after all, you are reading this post), you probably have noticed that manufacturers of venerable iPod made change to the other white product line. Yes, Apple still makes computers, and good ones at that, and no, they do not run Windows. However, they are a step closer: now two of four Apple’s computing lines feature processors made by Intel, same as those commonly found in new Windows computers. Now that architecture differentiator has been removed, what lies ahead for the consumer and the industry?
There are two key issues underlying this change: speed and heat. The latter one is also referred to as power, since higher power chips consume, more heat they usually dissipate.
During the annual Macworld keynote, Apple CEO, Steve Jobs showed some benchmarks that pitted G5 (processors based on PowerPC architecture and made by IBM) against new Intel, and claimed that new Intels are 2-5 times faster. There’s been a lot of head scratching about this in user and tech communities since nobody can really believe such a quantum leap. IBM guys are no duds, they know performance computing, and sans specialized architecture optimized code (aka AltiVEC for those of you initiated in PowerPC ways) applications should run comparably, if not faster on PPC solutions featuring similar clock speed. Some benchmarks have been run by 3rd parties to see what’s up. A good comparison is done by Ars guys here, and benchmark results are here. Also, Walter Mossberg in his WSJ column compared usability aspects before and after.
Here are some bottom line(s):
1. Dual core Intel processor = higher performance on multi threaded applications v. single core G5. In real terms that means that stuff that has been developed with multi processor architectures in mind (e.g. ripping a CD or exporting a movie in quicktime) will run faster on dual core intel than on single core G5. DUH!
This, however, is the crux of Intel advantage: ability to provide a solution with 2 cores consuming same power as IBM’s one. Marketingly (is this a word?), Jobs used SPEC benchmarks that favor mutithreaded apps over those that test peak performance to show off new computers. And voila! Performance in real-life apps, however, does not appear nearly that much better: 40% faster on CD ripping, 15% faster on movie recoding, and no advantage in many other apps.
2. Some software is not native and has to run through a PPC-to-X86 translation engine (a.k.a. Rosetta). But, things are not all that bad. Again it’s a 2 core thing: one core translates code and cashes it for later use, while second core executes. On "simple" applications such as MS Office - lag is apparently imperceptible. Photoshop and other heavy duty workhorse programs have issues - obviously.
3. Core graphics performance. This is where G5-based architecture has an edge (as much as 15-20%), which comes as no surprise because Quartz has been tuned for years to run on PPC optimized architecture. Given Intel’s interest in taking on graphics processor manufacturers, I believe it’s a matter of time before they lick this issue.
4. Intel-based system seems to have higher disk performance. This is a bit of a puzzle to me right now, but it’s a welcome development in a long run due to rapid proliferation of files and their size - mostly driven by media content commonly found on our computers.
So, if you are a consumer with regular needs (documents, pictures, home movies, music, and occasional games) this transition is as lateral and as seemless as it probably could be. In a good old Apple way, things just work, and do so out of the box. Wisely, Apple changed one thing and one thing alone, and made sure this change is as imperceptible as possible.
In terms of industry and future developments, Intel and Apple have aligned incentives: low power, high performance consumer devices. Laptops and small-format machines are becoming prevalent and substituting common workhorse desk stations. While it remains to be seen how fast will true benefits of this alliance be realized, exponential growth in performance will certainly be significant as cumulative effects of optimized code and architecture kick in.
What happens next? Well, it’s no news that OS X (Mac operating system) works on already available Thinkpads, Dell’s etc. While compatibility is not perfect (incomplete graphics support and some networking issues), it’s more than 90% there. This opens up a logical question: Will Apple return to clone war days of John Scully and thus launch a frontal assault on MS by licensing OS X for OEM distribution?
While PC industry has been becoming boring for years now - with competitors edging each other out on shipping costs, and although such lateral assault would provide something for journalists to write about, I personally doubt this initiative will give quick results. MS is built its ecosystem very carefully with OS at its base and communication and data productivity applications around it. In addition IT staff (which makes purchase decisions) is more likely than not educated (indoctrinated?) by MS and will go the route they know better - not necessarily one that is better, even if products do offer some advantages.
But is this where the beef is? PC manufacturers are emphasizing laptops more and more as portability becomes crucial and technology allows desktop replacement workstations to be packaged into mobile formats. Apple has been there about four years ago, with Jobs declaring the year of the laptop when Apple launched the new generation of Powerbooks. Now, Apple it’s mostly about consumer electronics and associated devices. All along the industry, in fact, consumer segment is what has been driving growth and novel applications for a while now. Gateway got out of trouble with TVs, Dell followed into PDAs, TVs, projectors, and HP went in with Media PCs. What all PC manufacturers are finding though is that CE space is fundamentally different from PC one. Consumer demands simplicity and performance combined, as opposed to extensive set of features. Just think of VCRs flashing 12 because nobody in the household knows how to program them. Such demands are contradictory with requirements traditional, enterprise-oriented manufacturers are used to, and hence they tend to get products wrong. They attempt too much and achieve too little. Apple’s core, however, is the widget that works out of the box: fully integrated, everything plays together, and thus is positioned to succeed in CE market.
Thus, I don’t think opening OS to OEMs is on top of Apple’s list - no matter how funny it would be to see Dell ship OS X powered computers few short years after Michael himself said that if he were Jobs, he’d shut the company down and return money to shareholders. What I do expect to see, is further strengthening of home-base and building a network of externalities akin to that of MS in enterprise market, and keeping OS wars on a low heat. One way of doing so, perhaps, would be opening some aspects to the open source and letting pent up demand take care of compatibility issues.