Whats Right About the PowerPC?

 
 
By David Morgenstern  |  Posted 2005-06-05
 
 
 

Whats Right About the PowerPC?


If anything were guaranteed to set teeth gnashing in the executive suite at Apple Computers headquarters, it was Intels Thursday announcement of Yonah, the dual-core version of its Pentium Mobile processor line due next year for Windows notebooks. Or maybe not.

Over the weekend, the rumors of Apple switching from its PowerPC-based platform to some flavor of Intel chip kept growing. Perhaps these reports (or more likely the television crews camped out in Cupertino) caused apoplexy in the upper ranks as well. Or not.

The only certain bet is that on Monday morning, CEO Steve Jobs will take the stage at Apples Worldwide Developers Conference in San Francisco and pitch developers (again) on the goodness of the companys recently released "Tiger" version of Mac OS X. Everything else is speculation.

eWEEK Labs took a close look at Mac OS X 10.4, aka "Tiger," and said both merit strong consideration from non-Mac shops. Click here to read more.

Now, Intels Yonah platform sounds interesting. The two cores will boost performance for multithreaded applications (and will give some oomph to the next-generation version of Windows, which will no doubt need it).

But when running on the battery, the processor can shut down one core to improve power consumption. This will only sap the performance of processes that can take advantage of multiple cores—usually a minority of applications or functions, albeit important ones. The single core can also control all of the chips integrated 2MB Level 2 cache, instead of having a pair of smaller, dedicated caches.

Unlike the clarity from Intel on its mobile processors, the roadmap for the PowerPC G5 chips, Apples branding for the PowerPC 970, is missing from IBM and Freescale Semiconductor (formerly Motorolas SPSS chip group). This lack of fresh marketing pitches on the topic is often a portent of forthcoming announcements.

Click here to read more about forthcoming low-power and multicore PowerPC processors.

I find it difficult to understand the insistence by some in the industry that Apple would be attracted to Intels processor lines since, for the most part, the PowerPC is working well for it. Of course, the slow progress in some areas—the 3GHz speed bump and mobile procession—have produced some growing pains for its product grid (no relation to grid computing).

When Steve Jobs returned to Apple in 1997, he griped about the dozens of product lines in the company. Competing development teams produced lines with different logic board architecture or proprietary expansion cards. Or, the different marketing groups would work against each other, at times limiting the growth of some systems in market segments or dumbing down a systems capabilities to fit the positioning of a certain segment. In other words, a mess.

With the introduction of the PowerPC G3 models, Jobs slashed the number of designs and offered up a simple four-slot product grid: a desktop and notebook for professional customers (the Power Macintosh and PowerBook, respectively) and a similar set for consumers (iMac and iBook). Servers dont count. This strategy has worked well for Apple, and only a few expectations have strayed outside the grid, such as the Power Mac G4 Cube and the new Mac Mini models.

Next page: A few hard facts about Intel vs. PowerPC.

Whats Right About the PowerPC? - Page 2


Interestingly, Apple has pushed each of the centers along, especially on the consumer side, instead of limiting performance at the low end to help differentiate the professional and consumer capabilities. The company should be given a lot of credit for this.

For example, instead of holding back the iMac by running the older G4 processors, Apple moved it quickly to the G5 line. Still, theres a significant performance difference in the professional desktop side, since the Power Mac G5s dual-processor architecture, expandability and eight memory slots really produce workstation-level performance for 64-bit content and sci-tech applications.

The problem is in the notebooks. The difference between the G4-based iBook consumer and PowerBook professional lines has shrunk (again thanks to Apples driving the bottom forward and ignoring the potential pain for the higher-end line). Or customers dont perceive the value in the higher-priced spread.

But will this performance gap be better filled by Intels processors or some forthcoming solution from Apples current PowerPC partners? Or has Apple grown tired of the manufacturing and design excuses? Thats something only Steve Jobs can answer.

A dual-core version of the PowerPC G5, the PowerPC 970MP, looks to be on its way. But the mobile G5 still looks like a difficult fit into the kind of machine that Apple wants to offer: sleek, tony and power-consumption-savvy, yet full of performance.

Here are a few hard facts about the PowerPC vs. Intel case to consider:

  • Apple pays less for its PowerPC processors than the cost of comparable Intel chips, and analysts say the cost is less than half as much for some models. Moving to Intel would hit Apples bottom line.

  • The PowerPC G5 is a smaller, more efficient chip than the competition. The G5 has performance comparable to any current x86 chip, which is really all thats necessary to persuade Mac fans to stay with the platform. Mac OS X trumps Windows any time, and all thats needed is a machine that can hold its own against Wintel machines, not necessarily beat them.

    According to Peter Glaskowsky, analyst for The Envisioneering Group, in Seaford, N.Y., the G5s "superior system architecture allows its performance to grow faster, as we see with each new G5 release. The Pentium 4 isnt showing that kind of headroom. IBM also has far more experience with dual-core design than any x86 vendor."

    Intel is more than a processor company. The rumors of a partnership could extend to other initiatives important to Apple, such as WiMax networking or PCI Express.

    Of all the predictions on Apples mobile chip dilemma Ive seen, I like the one from Kevin Krewell, editor-in-chief of the Microprocessor Report, in San Jose, Calif. Its the path of least resistance.

    To arrive at the right mix of frequency and performance, he suggests that IBM and Apple should build a new G4-G5 hybrid instead of waiting for a low-power G5. This chip could be used in future versions of the Mac Mini, as well.

    What is my prediction? All of the above. And Mac OS X Tiger. Gotta love it.

    Check out eWEEK.coms for the latest news, reviews and analysis on Apple in the enterprise.

  • Rocket Fuel