Commentary on the transition to Intel based Macs<br>
Well, it has started. The first of the Intel based Macs are shipping, and real world reports of their abilities are starting to show up in the community.

Surface review of the transition
While Apple had put forward remarkable benchmarks for the new systems compared to equivalent PowerPC versions of those product lines (2 to 4 times the speed), it seems that the actually speed gains are in the more realistic range 10 to 20% for native application.

Further, what Apple has traded in the transition is integer performance for floating point performance. This is not much of a surprise people who have been following processors over the years as Intel has often excelled at integer performance while IBM (and Motorola/Freescale) had pushed more for floating point abilities. And when one considers the original aims of IBM's POWER and PowerPC lines of processors, the importance of floating point abilities makes sense.

What the average Mac user is left with from Apple's marketing is a shift from an emphasis on Macs floating float abilities (and terms like "desktop supercomputer" and "Gigaflops") to the new emphasis of integer performance (noted as the more important measure in recent Apple marketing).

There were a number of very good reason for the shift. Some of these were give by Apple and some have been speculated on by people in the community. For my part, Apple's move seem to have been motivated a combination of developments at both IBM and Intel.

On the IBM side, there was the fact that IBM had originally sold the idea of the PowerPC 970 series to Apple as a chip that both computer makers were going to invest in. IBM had place the 970 in the position of being a transitional processor to move it's workstation clients from 32 bit to 64 bit processors. As it turned out, the need for this type of transitional processor wasn't what IBM had originally thought. So that left Apple as the main consumer of the processor line. It was the fact that IBM was going to be invest in the line that guaranteed that IBM would continue to keep the development of the processor on the front burner... once they were no longer invested in the series both development and production suffered.

Add to that IBM's increasing role in the gamestation market and Microsoft's move to PowerPC, and Apple was becoming lower and lower on IBM's priority list.

For Intel's part, Apple has been one of the biggest targets (as a client) for the last 10 years. Intel is, as much as a processor maker, a technology company. Every year Intel puts out technology demo PCs to try to get the standard PC makers to try new technologies that they have developed. And most of the time these technologies are slow make it to market.

A key example is USB. Intel developed USB in the early 1990's, but had a hard time getting PC makers to include it in their hardware because Microsoft wasn't supporting it in their operating systems (the first Windows version to support USB was the second edition of Windows 95, and the first version of the NT line to support USB was Windows 2000).

What move the market on USB was a move by Apple with the introduction of the original iMac. Apple demoed the iMac and then let everyone know that the one way to connect peripherals to the new system was going to be via USB. With that, a ton of USB scanners, printers and external disk drives popped up on the market.

What Apple brings to Intel is the freedom to move very quickly on technology. Where PC makers can really only advance at a rate set by Microsoft's software, Apple (as both hardware and software maker) is able to capitalize on these types of advances.

From Apple's stand point, the attention that Intel is willing to give Apple even though it is not (by any stretch) Intel's biggest processor client is a welcome change compared to IBM's growing indifference to Apple's needs.

The move from a users point of view
I have had a preview of what a move like this would be like having been a user of both NeXT's operating system and Apple's Rhapsody operating system. And the main difference in the case of Mac OS X has to be the range of Carbon development tools.

Cocoa, which is based on Yellow Box and OpenStep, has long been set up for this type of portability. But Carbon, which is based on the original Mac Toolbox APIs, has had a wider range of tools which makes tis type of transition harder for Carbon developers.

As would be expected, many Cocoa apps have been made Universal very quickly while most Carbon apps are taking much longer.

As a consultant to many different types of Mac users, my advance varies from client to client on the move to the new systems. For those who are using mainly Cocoa apps, the move can start right now. And those using mainly Carbon apps I suggest waiting until Universal version of their primary apps are released.

Seeing as I'm rarely at the cutting edge of hardware technology with my own systems, I most likely won't be one of the first to move to the new systems... but because most of my core apps are already Universal (like the Stone Design and Omni Group apps) I could be pretty much just as productive on a new Intel Mac as I am on any of my PowerPC based Macs.

Windows on Macs, Mac OS X on PCs
Ever since Apple released the developer transition kits, people have been talking about running Windows on Intel based Macs. I had warned early on that the systems Apple had given out were based on what the Intel version of Mac OS X was designed to run on, and that Apple most likely wouldn't be following those limitations.

Because the original design of Mac OS X for Intel was to be compatible with standard (Windows compatible) PC hardware, the developer transition kit systems were standard PC hardware. But once Apple had the time to develop new hardware around the new processor, I know this was limitation was going to be gone. Apple had no reason to make Windows compatible hardware just because they were using Intel processors. And making the new Macs different from standard PCs would help insure that Mac OS X would have a hard time running on anything else other than Apple hardware.

Plus, there is the fact that Apple and Intel are working together to push forward new technologies that are not yet supported by Windows.

What this means is that one day Windows may run on Apple hardware, but most likely it'll be older versions of Apple hardware. So some version of Windows Vista may (someday) run on these early Intel Macs, but not on the Macs that currently shipping at that time.

In the same way, I would guess that Mac OS X for these new Intel based Macs may someday run on future PC hardware... but it'll most likely be an older version of Mac OS X running on new PC hardware and not the currently shipping version of Mac OS X at that time.

And from Apple's perspective I think this is a safe trade off. It keeps their current line of hardware and software unique and special so they don't lose much in the way of sales to other PC makers.

Overall outlook of the transition
Apple has done a great job of making sure to produce compelling software to make their hardware just as compelling. While I've often worried about Apple doing this at the expense of third party developers, when it comes to this type of transition, this is a good thing. Apple can move their consumer and professional level apps to the new platform at the pace that they are moving their hardware lines.

The main worry is third party apps that many Mac users require as part of their work flow. Adobe was a little slow on Photoshop for Mac OS X originally, but with the release of Lightroom as a Mac-only public beta I'm under the impression that Adobe does consider Mac users to be an important part of their customer base. Quark was another company that was slow in the Mac OS X transition, but with the major loss of ground to InDesign (largely due to that lag) it seems that they are pushing to make QuarkXPress one of the first major third party apps to become Universal.

The only question that I see as outstanding currently is the pro line of hardware. Right now I don't see Intel producing anything that would match the top of the line PowerMac G5... but I'm sure that they much have shown Apple something on their road map. Apple had originally talked about the move being complete (I assume that means shipping Intel versions of all their hardware lines, rather than discontinuing and replacing all their PowerPC lines) by sometime in 2007. Now it seem they plan on this happening by the end of 2006.

I also full expect that Apple will continue selling PowerPC versions of some of their products for some time after there are Intel versions of all their hardware. Apple continued to sell systems that could boot Mac OS 9 long after their standard hardware had become Mac OS X only, so I expect this to be the same.

All and all, I pretty much expect that this transition to be a lot like the move from 68k to PowerPC processors, only faster. From what I recall of beginning of that transition almost 12 years ago, this one seems smoother... but many of the problem are similar. Apple promised that those first Power Macintoshes were super fast, and people found that with many of their preexisting software titles that the speed was going to be there until native versions were available. Fortunately we have far more native apps ready today than we had with those first Power Macintoshes.

And that difference alone is a promising start for this transition.



Stone Design's Create(tm)
2006-01-31 10:42:56 -0600