Macs on Intel: The 80% Factor
June 8th, 2006Just think about it: A year ago, Steve Jobs promised that Macs would switch to Intel before the end of 2007. To think that this daunting process is nearly all done 18 months early. Right now, in case you tuned in late, everything but the Power Mac and Xserve have been upgraded, and since Intel’s new desktop chips are due beginning next month, it’s reasonable to expect that upgrade will be completed by fall with few serious complications.
But it’s a good time to look back at what Apple has wrought, particularly in light of the mixed reaction to its Intel switch. Today, it seems that it was the only logical choice, but I’m sure some of you wondered whether Jobs was pilling some sort of scam with such a vast change in direction. No doubt developers felt the shock and awe, even if a few had an inkling of what was coming. Here they’d spent years moving their work to Mac OS X, fine-tuning performance, dealing with the problems occasioned by the regular stream of updates, and now they had to do it all over again.
Of course, Jobs told thousands of developers who attended last year’s WWDC how easy it would be to switch apps to Universal, to run on both PowerPC and Intel chips. Just click a checkbox, make a few tweaks, and that would be all that was required for most, though a few might have to work harder. Of course, that’s not much better than the spin Apple put on the alleged ease of converting software from Classic to Mac OS X, but in the real world, the process wasn’t near as daunting as some feared.
Unfortunately, this isn’t true for everyone, and not every Universal application is as efficient as it should be. Worse, developers who didn’t use Apple’s Xcode environment to begin with must first import their code and clean it up. When you have sprawling productivity applications such as Adobe Photoshop and Microsoft Word undergoing this transformation, it’s no surprise that the projects may take months or years to complete. It also requires a large investment to pay the salaries of the employees who are doing the work, so it makes sense that the funds have to be recouped somehow. So you just know that some of these cherished Universal versions will carry a standard upgrade price tag.
The only hope is that the new versions will contain a sufficient number of new features to justify the money. I do not want to see any repetition of what Microsoft and Quark pulled with their first Mac OS X upgrades, which was to deliver native versions at the full upgrade price with a paucity of new features. If they try that scheme again, I would rather stick with the existing versions and let someone else pay for the beta testing.
As to Apple, I think it’s done a great job. Yes, I understand there were some issues with the first shipments of the MacBook Pro. I’ve heard of early battery failures, excessive heat and possibly other failures. The latest problem involves the MacBook, where Apple’s factories put a plastic seal on the top cooling vents to keep dust from creeping in. If you don’t remove the seal, it gets too hot.
Now, whenever I unpack a new Apple product, I spend a little time going through the nooks and crannies and the external fittings, such as power adapters, to make sure all the plastic is removed. So there’s a little common sense here as well, and maybe some folks, anxious to put their new note-books to use, have forgotten to check things carefully before pushing the power button to see what happens.
Another issue, which is something we’re going to have to live with, and that is that the word “laptop” is no longer in Apple’s lexicon. They are telling us that these things can run hot, because of the powerful processors inside, and you just have to get accustomed to it. On the other hand, lots of Apple lap — make that note-books — have tended to be a little too warm for knee placement, so this shouldn’t come as much of a surprise.
In fact, I’m rather pleased with the way this processor switch has gone. Rosetta emulation compatibility is quote good, and if you don’t expect miracles, performance should be acceptable for all but power users in most cases. Moreover, I’m particularly impressed with how Mac OS X sings and dances on an Intel chip, as if it were meant to reside on that platform.
Best of all, don’t forget that this is only the beginning. As Intel rolls out its higher-performing chips over the next few months, Apple will, I expect, be quick to put them into production. Already, preliminary reports are showing that AMD’s highly-touted advantage in high-end desktop processors will be short-lived. So, despite some skepticism, it does appear that Apple’s pact with Intel was not a pact with the devil. It
was exactly the right move, at exactly the right time.
Print This Post
When Jobs announced the first G5 at 2.0 GHz and promised a 3 GHz version for the next year, I was really excited and finally relieved that Macs aren’t at the mercy of Motorola’s whims.
When the next year came and the G5 didn’t reach 3 GHz, I started feeling uneasy. When the second year rolled around and we didn’t see more than 2.7 GHz G5, I knew that Apple should be moving to x86 processors. IBM turned out to be almost as bad as Motorola.
Motorola’s and IBM’s unwillingness to compete with the x86 camp in the desktop/notebook spaces cost Apple a lot and held the Mac back for very long (can anybody really forget the 1st year of the G4?).
I remember couple of years ago at WWDC Jobs strongly encouraging developers to move their apps to Xcode and thought, what an odd thing to do and poor Motorola (who owned code warrior). I guess he couldn’t come out and bluntly say it. But then again, Motorola had said at one point that they’ll be concentrating code warrior’s development on the embedded space, so that should have been a hint. Oh well, hind sight is always 20/20.
One thing that shows how lazy Microsoft has been with respect to updating Office on the mac: you will notice that they still can’t do live window resizing on Word or Excel (maybe all their apps, haven’t checked). This is an indication that they haven’t even updated their software to use Carbon Events, and that they’re still using the same old WaitNextEvent loop that is a relic of the OS 7 8 9 days. I can only imagine all the old cruft that is in the Office for Mac codebase…no wonder they take forever to get new versions out.
It is these kinds of recurring expensive decisions that lead many to develop software on other platforms. If Apple had skipped the G5 to begin with and released the MacIntels earlier, maybe it wouldnt have been so bad. As long as Apple sticks to Intel AND Intel doesnt do it’s own dramatic makover, Apple will be OK for the long term.
Many talk about this “miracle” of being able to switch processors in such a relatively short time. This is just marketing spin. We forget that Apple has been secretly working on this for years prior to Job’s Intel announcement. Also we forget that Microsoft has done the same behind the scenes. Microsoft’s XBox 360 uses a modified version of Windows that runs on PPC processors. The initial xbox 360 game demos were actually running on G5 PowerMacs. To the great embarassment of Apple, it runs games better than any PowerMac (using OSX). What does this tell you? Not-so-secretly, Microsoft appears to be doing the same thing… building on OS that can run on other processor architectures (in this case non-x86 architecture). It also tells you that Apple did not fully utilize the maximum power of the G5 (and Microsoft did and is still doing so today). How embarrasing… game machine that can beatup a PowerMac.
Garret,
The XBox running games faster than OS X tells you that M$ designed it to run games. The test boxes for the XBox were G5s, that’s true, because M$ had a version of NT running on PPC machines back in the late 90s. I’m sure they eliminated the parts of the OS that weren’t needed for gaming and fine tuned those that weren’t. Don’t compare apples with…you know.
Apple wasn’t secretly working on x86 OS X for years, OpenStep ran on x86 before it became OS X…that’s why the PPC version was getting faster with newer builds, as they continued to optimize it…and they always had the x86 version being updated internally as an alternative.
I don’t think Apple is going to stay still on this OS/processor combination for more than 5 years. Apple has been doing a major hardware or software change every 5 years for the last 20 years. They stayed still too long in the OS 8 and 9 years and they lost a lot of market share. Steve Jobs won’t do that again. One of the reason the Mac is so good is that they update it to make use of the latest technologies. Imagine if they were still selling DOS on Apple machines. People might think that OS X is secure, safe, and easy enough for their current uses but in 5 years it will be antiquated. Who knows if the majority of users will be users personal computers then?
Apple’s move to Intel has been a good one, but they shouldn’t completely abandon the PowerPC just yet. If PA Semi can deliver their 64 bit dual core processor with Altivec as scheduled, the PowerPC could easily remain viable for the duration yet. If the specifications are true, the chip should very soundly trounce any Intel chip in performance per watt. PA Semi has a fairly aggressive roadmap when it comes to multiple cores and if they can hold to the roadmap, will easily compete with Intel’s offerings. Even surpassing them in low power/high performance computing.
Apple’s decision to go with Intel was still the logical choice as PA Semi is just a small start up and hasn’t even delivered working silicon yet. After all, if IBM and Motorola with their far superior resources couldn’t deliver, why should PA Semi be any different? However, Apple should consider this carefully. After all, Microsoft with their superior resources haven’t been able to deliver their OS upgrade to XP even as Apple has delivered a number of upgrades in the same time span. Perhaps PA Semi will be able to deliver where the behemoths weren’t.
In any case, Apple really needed to move away from IBM. IBM could really have delivered far greater performance for the G5, but were too afraid it would cut into the sales of their Power4 chips. IBM could never really be serious in delivering a true high performance chip to Apple as they were too afraid that doing such would cannibilize their own hardware sales. Jobs knew this and knew that Apple needed to go elsewhere. Freescale’s ability to deliver anything even remotely competitive to Intel was essentially nil. To risk everything on a start up that hadn’t even delivered working silicon would have been very irresponsible. Still, I suspect that if PA Semiconductor delivers a compelling part, Apple will very likely use it. Apple wants the best processor available, not the spoils in the way that IBM was doing.
Regarding the poor OS X performance on the G5 compared with Windows on the same chip, it really shouldn’t come as a surprise. OS X’s performance is severely hindered by Apple’s use of an archaic monolithic kernel. The Linux people already know this and much has been written about it. If Microsoft used even a reasonably efficient kernel on PPC, it would easily have bested OS X. However, people use OS X not for maximal performance, but for its vastly superior user experience. An experience that allows for greater productivity even with the compromises in performance inherent in using OS X on the present PowerPC chip. After all, my Powerbook will link up to wireless networks with relative ease and the process is nearly painless. A process that is painful with respect to my Vaio whenever I try to link up to a new network. I can care less if the Vaio runs faster with its Pentium M processor compared with my G4. I can get far more accomplished with the G4. Besides, Microsoft and its hardware vendors haven’t figured out how to do that instantaneous wake from sleep thing nearly as well either. As far as Linux waking from sleep on a laptop, well, let’s just say that the Linux vendors have to make much more substantial progress in this area.
So, hopefully, Apple will be able to maintain both processor lines in the way of x86 and PowerPC. I also expect them to eventually move away from the monolithic kernel in a future OS upgrade. They have focused their energies precisely where they needed to be. My Linux using friends are correct that OS X on PowerPC is slow, especially when it comes to running a server. However, OS X is light years ahead of Linux in terms of useability. That’s what matters to me as it does with the vast majority of consumers. I just don’t want to be spending hours tweaking Linux to use my new digital camera or printing on a new printer. Microsoft is really out of the game at this point. Their OS has OS X beat in a small number of areas in terms of useability, trails badly in many more areas and is a security nightmare. Linux walks all over Windows as it does OS X in performance. As Vista attempts to catch OS X, Apple will be releasing 10.5. As OS X has matured, Apple can now go back and focus more on performance of the OS whether on PowerPC or x86. It should still be exhilirating ride, that’s for sure.
Before you make assumptions about the alleged “inferior” kernel in Mac OS X, folks, I suggest you read the fascinating article at roughlydrafted.com on the subject. Further discussion is welcome on this — but after you read the article.
Peace,
Gene
OS X uses a microkernel architecture and Linux is a monolithic kernel. I thought I should clear that bit up before anyone gets terribly confused by glasspusher’s post. To be exact, Wikipedia lists the OS X’s kernel as a Hybrid Kernel, although that definition seems dubious — http://en.wikipedia.org/wiki/Hybrid_kernel .
Otherwise, I enjoyed the article and Gene I should let you know, your new design for The Mac Night Owl is fantastic.
Nathan