Despite Intel Migration, a Mac is Still a Mac

Thursday Jun 16th 2005 by John Welch
Share:

Datamation columnist John Welch says it's important to distinguish between the emotional impact and the real world impact.

On Monday, June 6, (a bit of irony in the timing) Steve Jobs, CEO of Apple Computer, announced that they will be migrating the Mac to Intel. This is a statement that, within the Mac community, had the impact of a rather large asteroid... one moving at almost light speed.

However, it is important to distinguish between the emotional impact and the real world impact.

That the emotional impact is huge, there can be no denying. Apple has always made much of the fact that they ''thought different''. They didn't dance to Intel's tune. They went their own way. In 1994, they went with the PowerPC (PPC), and for 11 years they stayed with that architecture -- the 601/603/603e/603ev/604/604e/G3/G4/G5. But it hasn't been a perfect relationship between Apple and Motorola/IBM.

Speed always has been an issue. Yes, there are many learned, and correct, opinions on the Instructions Per Cycle (IPC) superiority of the PPC architecture, particularly with Altivec. For a long time, PPC CPUs ran cooler. PPC had a more elegant and forgiving Instruction Set Architecture, (ISA). But not everyone gets that Altivec can have a 40 percent efficiency advantage over SSE2/SIMD. Everyone does get that 3.2 GHZ is faster than 2.7GHz, and that hurt Apple.

In the last two years, it's become obvious that while the G5 is a great chip, IBM just doesn't have the interest in doing the work to make it the chip Apple wants. No notebook version, minor speed increases. To be fair, of late, Intel hasn't been blinding the world with speed increases either. In fact, they've kind of gotten out of the ''clock speed is all'' business.

But there's another fact to consider, and its an important one. Neither IBM nor Motorola consider general purpose CPUs to be their primary business, or even a primary business.

Motorola is all about embedded microchips. Cell phones, radios, satellites, your car... that sort of thing. IBM, and more specifically IBM Microelectronics, is a bit split. They care a great deal about the high-end server CPUs, for their z/iSeries servers. However, a Mac built on a POWER chipset would increase your home utility needs to where you would be classified as a utility. So that does not do Apple a lot of good. The low-end for IBM is the Xbox/PS3 market. Well, the Xbox 360 is, from what I've heard, liquid-cooled. No laptop love there. Both the PS3 and the Xbox 360 are running different CPUs than Apple anyway, and that's radically different in the case of the PS3. As well, in the game console market, CPU upgrades only happen with new versions of the console. So that's a new version every three to four years or so, which is a far cry from Apple wanting new toys every six months, if not more often.

So for Motorola and IBM, Apple is a partner that makes them a little money, but causes them a lot of work and pain.

Intel, however, is different. (Ah, the irony in this transition is everywhere).

While Intel has a good sideline in embedded chips (the 486 that runs Airport base stations is proof of that), and they have a good server chip in the Itanium2, their primary business is in 'desktop' CPUs. They make their bread and butter in the exact spot that Apple lives. More importantly, they recently announced that they are focusing future research and CPUs on the Pentium Mobile design.

Where has Apple been hurt the most by IBM's lack of desire? Mobile computing. So with Intel, not only do they have a partner focusing on the precise area they care most about, and not only do they have a partner who is used to the needs of consumer computing, (indeed, they defined much of it), but for the first time in Apple's history, their CPUs are going to be made by a company for whom making CPUs is a primary function.

So we get an idea of why this is good for Apple. Intel does this for a living, they're quite good at it, and they care deeply about this arena.

Continue on to the next page to read about how this move will benefit Intel and what IT administrators need to think about.

Benefits to Intel

But what does Intel get out of this? Well, there's the obvious PR win. They have gotten the last great holdout. The last rebel now is on the side of the empire. That's a PR bonanza, and I would not be surprised to see Intel doing a lot of marketing based on that fact.

But there's a lot more here for Intel than just PR and the smugness of having that last couple of percentage points. Think about that flat-ish sales curve for Intel computers that Steve showed at his recent keynote. There's two factors to that. The first is Intel's fault. It's really quite hard to convince most folks that they need to dump a 3.0GHz PC for a 3.2GHz PC. Outside of some vertical markets -- like multimedia, gaming, 3D design -- most people just don't care. I know that my company's replacement schedule is driven by amortization and other financial considerations, not technology.

The other part of this is the people using Intel chips -- primarily Microsoft. While Microsoft is very good at pushing Windows, and applications, they haven't pushed the majority of the PC user base to get a new computer since Windows XP, and that's quite a few years ago. Windows Media, applications that rely on Windows Media, and games may need faster hardware on a regular basis, but those don't push a lot of Intel tin. Moviemaking in the Wintel world is nothing like it is on the Mac side. The applications that make it as simple as iLife still aren't available on Windows. Tablet sales are still so small as to be nonexistent compared to desktop sales. And PocketPC/Smartphone sales are spread across a few CPU companies, of which Intel is a major player, but by no means the only player.

So while Microsoft is very pleased about tablets and handheld devices, the net benefit to Intel is kind of small.

Things like Office, and other mainstream Microsoft applications haven't had a major update in years, so they aren't pushing a lot of Intel tin. XP Service Pack 2 isn't going to push a lot of tin. If your system runs XP well, it will run XP SP2 well. Longhorn will probably push a lot of tin, but it's not coming out for well over a year to a year and a half. Even with that, a lot of Longhorn is being backported to XP, so you'll have access to a lot of the Longhorn coolness on your existing machine. Microsoft says, probably correctly, that Longhorn will run these backported technologies better, but as of yet, 'better' doesn't have a meaningful number.

Even if Longhorn is as cool as Microsoft says it will be, in the enterprise there will be a six-to-nine-month delay before widespread rollouts begin. In the home, you'll have an initial spurt, then a drop, with a steady flow of people upgrading. However, if XP is doing the job for home users, that flow could be smaller than Microsoft wants it to be.

So even if Longhorn is released on Dec. 1, 2006, the corporate world won't really start buying until the start of the second half of 2006, or possibly even the last quarter of 2006. (If then... Microsoft's success at getting customers to migrate from Windows NT 4 to the Windows 2000/2003 product line has not been a mind-boggling success.) For Intel, that's not good.

However, there's another issue, as well.

Microsoft only pushes new technology when it pushes Windows sales. As long as Microsoft's sales are where they want them to be, Microsoft, as a company, doesn't really care about new chip tech. Microsoft has never done a whole lot with SSE(2)/SIMD -- certainly nothing close to what Apple is doing with Altivec. Microsoft may have supported USB first, but Apple created the USB market. It took them 90 days after the first iMac. Apple helping to create FireWire was the big impetus for USB2. Face it, if you want to push truly new tech, you go to Apple. If you want to sell astounding amounts of existing tech, you go to Microsoft.

Linux certainly sells a lot of boxes, but they focus more on using existing hardware better and longer, if nothing else, for better TCO. And Linux is primarily a server OS as of yet, and servers are updated or new servers are bought based on functional or technical need. If you don't have to buy a new server for a couple years, that's great. Servers may have more CPUs per box than desktops or laptops, but their sales numbers are far smaller than desktops or laptops.

There is also a philosophical compatibility between Intel and Apple that doesn't exist with Microsoft.

Continue on to find out what IT administrators need to consider...

The Zen of the Union

Intel, while a business giant, is at heart, an engineering company. When they come up with SSE, Hyperthreading, etc., they get excited. I used to hang out with some folks who did design work for them. When they have something cool, they get excited like little kids on their birthday. Running to a company like Microsoft with a cool bit of tech that may not be immediately useful tends to get you a ''Wow, that's um... really nice. How will it help us sell Windows?'' It's not a negative reaction, but it's not the reaction you want. The reaction you want is ''WOW!''.

Apple is, at heart, an engineering company. I know a lot of those guys. They love new, cool stuff. At their World Wide Developer's Conference, they have plugfests with FireWire and USB, just to see how weird you can build a functional, fully populated chain of devices. In the first WWDC I went to, they pulled off a maxed-out USB bus. It was critically important, because it beat Intel's record at the time.

MacHack. Apple engineers loved MacHack because it was a way to do really cool stuff, even if it was not always useful.

Apple creates hardware. They design motherboards. So does Intel. They both get excited about the same things at an engineering level. They sometimes had different, even opposing business aims, but the engineering teams of both companies live to do what no one else has ever done before. And that's reflected in the business aims of both companies.

This philosophical compatibility brings us to an important point: Intel now has a partner that can be a new source of ideas for chip design.

Apple knows hardware as well as Intel does. They help with microelectronic design. They're good at design. Intel's good at design. They're working together. I have to think that there's a very thick Intellectual Property sharing agreement between the two companies in this deal. Apple has never been, is not now, and never will be a Dell-type company, doing nothing but slapping logos and cases on motherboards from Intel. It is literally unthinkable to anyone familiar with the company that this would happen.

There's a final thing that Intel gets. What is the one arena where, in the last few years, they have been consistently beaten... and beaten badly? Scientific/High Performance Computing (HPC). Who has been beating them? Apple. Look to see some major improvements in HPC from Intel thanks to this partnership.

Impact on Corporate IT

Now that we've looked at some of the background of this deal, what does it mean to sysadmins?

Well, in the next six months, not much. If you need to buy a Mac, buy a PowerPC Mac. There's nothing wrong with them, and contrary to some of the more alarmist reports and rumors, PPC Macs are not going to be useless and/or nonfunctional on Jan. 1, 2006. They'll work even after the January 2006 Macworld Conference and Expo Keynote. They're great boxes, they will continue to be so for as long as you need them to be. As you get into 2006/2007, well, then you have to make some purchasing decisions... maybe.

I will now list all the major facts that we have on this decision:

  • Apple will, starting in 2006, transition its computer hardware product line to Intel CPUs;
  • They will not actively prevent you from trying to run Windows on these machines, but they will not actively encourage it;
  • They will actively prevent you from running a future build of Mac OS X on Dell boxes;
  • Apple is using technology from Transitive for their ''Rosetta'' feature, so that PPC applications can run, for the most part, unchanged on Intel-based Macs, and
  • Apple released an updated version of its development tools, Xcode, so developers can create Intel binaries.

    That's it. Five facts.

    There are a few more details, but nothing that contradicts this. We don't know exactly which CPU they'll go for, what the motherboard design will be, etc. We don't know what machines will be the first to make the transition, although it looks to be the lower end of the line, as it will be the easiest transition and cause the least amount of pain. However, PowerBooks and iBooks could be first, as well, since they need performance gains the most.

    We know some basics about Rosetta, in that so long as an application doesn't require a G4/G5, doesn't need drivers, and is generally well-behaved, it'll run fine.

    Rosetta is an important point here. It's not an emulator, but more of a translator. So there's much less overhead. There's a speed hit, although from what I've seen, it's not ridiculous, depending, obviously, on the application.

    So even if you buy an Intel-based Mac, it's not like your current software is automatically dead. Obviously, Apple learned some things from the 68K-PPC transition, which was much harder, for reasons I don't have space for here. One thing from that transition I will point out is that with this transition, the entire OS is running natively on both architectures, something that was most definitely not the case during the 68K-PPC transition.

    So, in 2006/2007, you'll have to look at the Intel hardware available and the available software. If the software you need isn't available, then is Rosetta a viable solution? If yes, get Intel. If not, can you wait for Intel? (Note to Apple...Roadmaps would be very good here. I know you don't like them, but in this case, make an exception.) If you can't, get the PPC.

    Again, major numbers of Intel-only applications will lag the new hardware by a year or so. That won't really be an issue until the 2008/2009 timeframe. The head of Microsoft's Mac Business Unit, Roz Ho, said the next version of Office will run on both. Adobe didn't say this, but unless CS Suite 3 isn't ready until 2009, it's a safe bet the next release of its applications will run on both, as well. I can't see everyone else being Intel-only by January 2006.

    Those are the main issues, and as you can see, they're small. For IS staff doing Perl/PHP/Shell/etc., it won't make a difference. All that stuff lives far above the hardware. In other words, this really is not anything to panic about. It's going to make for interesting purchase decisions in 2006/2007, but Apple makes that interesting anyway. Anyone who has dealt with Apple is well-prepared for that.

    Move calmly, plan, make sure your applications are available for the hardware you buy, and buy accordingly. It's just like every other hardware purchase decision that a sysadmin has ever made.

    Yes, Apple going to Intel is momentous, but it's not nearly as disruptive as the more emotional reactions I've seen indicate. It's still going to be a Mac, it's still going to be Mac OS X, it's still going to be running on Apple hardware. After all, that's why you buy Macs in the first place, and that's not changing.

    So stay informed, but don't get overly upset. In the end, I think it's a great decision.

  • Share:
    Home
    Mobile Site | Full Site
    Copyright 2017 © QuinStreet Inc. All Rights Reserved