Dojo (HowTo)







  Easter Eggs




  Martial Arts

Is the PPC on a limb?
Response to an Article

By:David K. Every
©Copyright 1999

There was an unusual and small article in "The Register UK" that was:

Basically the article was about how Apple is going to abandon the PowerPC and go to Merced (IA64).

While article may or may not have a point in there somewhere, it is so oversimplified as to cause dangerous misconceptions.

Servers First

The basic physics are that the Merced (IA64) looks like it will be a big, hot, expensive but powerful processor. It is designed for SERVERS! And it has a monster connector that is designed for a lot of heat, and won't go into portables or the low end. This is not a mainstream processor. The follow-on chip, McKinley, is also likely to be a server chip. After that, there is little doubt that they will eventually move the processor down the product line -- *IF* they can gain momentum and get market share and get software adoption. But those are a lot of "if's".

Those are also pretty big "if's" as well. Merced will be competing with big, bad boys on the block, like Alpha, Power3, UltraSPARC3 and HP-PA. Those competitors can currently eat up anything Intel has to offer. Merced's success is not as guaranteed as some pretend. I do think Intel's size will make it successful, eventually -- but it will take take time. But frankly I think every time a company says they are going to evaluate the Merced (and see what it has to offer), the Press reports it as another migration. I wonder how many spiffs Intel has to give them for their souls?

This whole premise that Apple is going to "exit the PowerPC model in 3 years" is someone's delusion. If Motorola and IBM ABANDONED the PowerPC tomorrow, and Merced shipped tomorrow, it would take longer than that 3 years for Apple to make a successful migration.

It took around 3 years for Apple's last migration (to the PowerPC) -- one of the smoothest transitions in computer history -- but that the first PowerPCs came out. There was after about 2 -4 more years of hardware and software development before the first machines shipped that people aren't factoring in. And of course the PowerPC was going to be a product line covering the spectrum of CPUs, not just a server chip, and many other variables that people aren't considering.

Covering your butt

For about a decade now, Microsoft and Intel have talked about redesigning the basic PC platform and what it will contain, to get rid of all the legacy crap. As soon as they do that, it does make sense that Apple consider offering something (hardware and / or software) that uses that simplified (engineered) PC-platform.

Software on Merced

Apple has been putting a lot of effort into making the Mac OS portable (cross platform capable). Most of the Mac OS can already move -- and is already moved and running on PCs today (meaning it is cross-platform).

  1. QuickTime really has a large part of the old Mac OS underneath it -- that is why it runs better than the Microsoft equivalent solutions on Windows (because it uses the MacOS low-level services when possible).
  2. The Carbon API is about making all the new Mac Libraries cross platform (and modernized).
  3. Most of OS X Server is cross platform in both the low level (Kernel) and in the high level API's, as is Yellow Box (Open Step and NeXT's stuff).

So in other words, Apple can move the platforms if they need to. Right now they do not need to, and Apple would be hurt if they tried -- so they won't, but they could.

There is little doubt that some day in the future, Apple will offer their software on other platforms -- it makes sense to do so (to gain market share). Will Apple offer a subset (like the server stuff), or the whole thing (like Mac OS X, Blue Box, and so on)? We don't know. The former is likely sooner than the latter. The former makes more sense as a business strategy, but both are possible eventually. It makes sense that either way, you wouldn't replace the PowerPC -- just offer Merced options. Why give up your market that is feeding you? Apple can offer their server software for other companies to use, and still sell PowerPC hardware, and gain marketshare and everyone is happy. If Merced gains big marketshare then this plan would offer Apple the possibility of some migration.

Hardware on Merced

If I ran Apple, I would certainly be investigating how to go to Merced if MIA (AIM -- Apple, IBM, Motorola) dropped the ball on the PowerPC Processors. Not that I expect it to happen -- but in business it is wise to not be restricted by a single-source, and it is a good idea to have contingency plans! Always remember though, exploring a path, or even starting down it (investigatively), is not the same thing as abandoning the PowerPC.

There is a lot of evidence that the PowerPC has far more potential in the low-end that Merced (IA64) or x86 (IA32) will, ever! Already the PowerPC has a serious advantage in portables. As I mentioned in another article (Parallelism), I think the PowerPC has the potential to crush Merced and still take over the high end. But despite the fact that IA64 may have little to offer, it is STILL a good idea to be prepared.

Now let's play pretend -- and assume that Merced were coming out, it is going to be much faster, smaller, cooler, better than anyone thinks. As Merced is coming out, and lots of people are offering Merced based machines, it would be a good time to consider offering a machine as Apple. If that were to happen, then maybe Apple would consider some hardware on that new platform. Let's also pretend that Intel finally gets a new platform for Merced designed and working (and can force adoption down the PC markets throat -- kicking and screaming). If Intel does all that, and if Apple uses it, then I still don't think this would replace the Mac line -- just augment it for many years. As I said before, if Apple starts down the Merced path it will take time -- if Apple did want to really migrate their product line, it is a 5+ year migration, assuming things go well for Intel. It is going to take that long until PC manufacturers start using IA64 across the board, and so it will be that long before the economies of scale really kick in -- and before process technology will make it affordable for low-end and portables.

So when is all this even possible? Late 2000 is the earliest delivery of Merced. 2001 is when the McKinley flavor of the chip ships. It would take a couple years after that until any momentum is gained and the product line spreads -- so we are talking timelines like 2003 or 2004 before they could start the migration -- that is the soonest Apple could even CONSIDER making that move (unless Intel has a lot of surprises). Then there is still a 3 - 5 years migration time. That is way past the 3 year timeline offered by The Register UK. This is assuming that Merced is just awesome and offers a bunch of compelling surprises that make it an absolute "must have", which would be a complete shock to the entire industry which is monitoring Merced.

Apple can't move now -- If Apple was going to port to Intel immediately, they would need to have to support IA32 (x86 / Pentiums). All the economies of scale, and all the portable solutions, are in the IA32 (x86) arena. That is why moving now makes no sense -- they would have to move twice, once to IA32, then again to IA64.

So long term it may make some sense that Apple is thinking about possibly supporting Merced / IA64 -- but that is more a 5 - 10 year plan, and based on a lot of "if's" -- not a 3 year plan that is absolute.

What likely happened

If Apple wants to offer some higher end server solutions (hardware and/ or software) in their 3 - 5 year plans, and they would be fools not to explore IA64 (Merced), then they would have to have some contingency plans to migrate if Motorola and IBM don't get off their butts and offer some higher end PowerPC's.

Probably someone heard a fragment of some contingency plan, or the product plan for a SINGLE product in the Server Arena (hardware and / or software) -- and then they went crazy with the story. Which shows you why you have to be skeptical about what you read in the press. Odds are we will hear more about this rumor over the next few weeks. But 10 or 100 people all talking about the same rumor doesn't make it true.

You must also remember negotiation ploys and power. Apple can threaten to go to Merced to motivate Motorola and IBM to be more aggressive on the high end of PowerPC's. That is a far cry from Apple really following through with it. Yet it isn't a far leap for one person to mishear a fragment of what was said, assume the worst, leaking it to a Newspaper that runs with it (and tries to fill in gaps on their own)!


Migration of the low-end, like set top boxes, iMacs and other low end and Portable machines, are not likely to make the transition to Merced within 3 years. I don't think Apple and Intel could do it in 5 years. Merced just isn't the right type of chip (cost, size, heat, etc.).

Migration, or more choice in the high end, is not a bad thing. The fact is that a little diversification can be good -- offering PowerPC based portables and low end machines along with Merced Servers might be some longer term strategy that could go on indefinitely. I could imagine Apple splitting the line between servers, clients and workstations. But I still think it is more probable that the PowerPC has enough in it to compete (or beat) anything in the Merced camp -- yet, you might offer the inferior Merced as well, just to appease some IS requirements or corporate or government mandates.

I am not afraid that Apple might make a Merced (IA64) Mac OS X based server in the future. I wouldn't be surprised if they did -- though I'm not counting on it either.

  1. If Apple did use Merced, it would be in the high end arena and not the mainstream.
  2. The only way Apple would consider mainstreaming the IA64 machines would be if Intel just has dozens of surprises up their sleeves (and Merced turns out to be very small, very fast, very cool, very low-power, and very very cheap, and so on). But then if Merced is all that, I would be glad that Apple is getting ready to use that processor! I'm not holding my breath on any of that.
  3. Even if Merced were everything, and Apple was going to migrate, it just wouldn't make sense for Apple to do it for years. It would have to be a long slow migration -- well beyond 3 years.
  4. Apple may be able to get Intel to fund some of the cross platform development (because Intel doesn't like being tied to Microsoft any more than anyone else with a brain does) -- so I see even the investigation as a possible win for Apple (and Mac Users).

So when you see an article that doesn't make sense, stop, breath, and think about what does make sense -- or what makes more sense. I'll play Johnny Cocheran -- "If it don't make sense, then it ain't worth 2 pence!"

You "common sense" approach can be wrong -- sometimes the world is all messed up and illogical -- but far more often than not your intuition and logic will win out other than some reporters half-informed story. If you doubt me, just review history and look at how many of the presses brain-dead predictions about Apple have failed to come through?


Of course days after they made some extreme and ridiculous claims, the truth of the story comes out.

Basically, they got the information from one Intel source, and it was somewhat extremely presented, and the reporter seems to have gone a little crazy with -- all about what I guessed.

Kudos to TheRegisterUK for offering a correction and explanation.
Shame on TheRegisterUK for not doing their due-diligence better BEFORE running the first article, and for not putting the source and supporting evidence in the original article.

Created: 04/13/99
Updated: 11/09/02

Top of page

Top of Section