Apple buys the PPC CPU manufacturer PA Semi, the maker of the PWRficient 1682M CPU. Link:

http://www.forbes.com/technology/2008/04/23/apple-buys-pasemi-tech-ebiz-cz_eb_0422apple.html

Who else would they buy? The original PA Semi plan was to sell CPUs to Apple. Here is Don Dobberpuhl at the Power.org Venture Conference in December 2005 (read the slide).



Apple's sales on computers were up for the quarter, especially the new thin one. iPod sales continue to be brisk. The iPhone is still doing well. Why wouldn't Apple want to use this design team to create a new iPod/iPhone CPU and better performing computers? AppleTV could use a new twist. It makes sense to us. We bet the venture folks were very happy to get their money back as the traditional semiconductor silicon sales model has changed. HP and Dell will continue to lose market share to Apple.

BTW, don't expect Apple to join Power.org or flock back to the PowerPC fold. The Power Community won't get any more of a boost from this than what it has from the game console business.

Next up?

Cisco makes an offer for Freescale or at least part of it...:-)

R&B :-)

Having thought about it this morning, I think having Dan and team (even with their experience on StrongARM and therefore XScale) work on the iPhone would be a waste of money; $278 million to throw away the Power Architecture design and improve an ARM core in an existing device?

Apple must have something new planned. PA Semi chips are doing well in markets where data transfer and storage are important; high end multi-disk NAS devices for datacenters and so on. It may be that Apple are positioning it for the next version of the Xserve RAID and their SAN software. Apple have traditionally done the bait and switch move where they court a partner to fill a gap in their product line (in this case, the Promise VTrak) and then work to replace it with an in-house product.

Given that you can get PASemi ATCA blades with 9 processors (18 cores!) this is going into the realms where Intel can't compete with just their dual-die quad cores, especially when it comes to performance and power consumption tallied together.

Apple may be bringing Power Architecture back to the Server market, where it has always performed best. Perhaps it was somewhat of a snap decision - the kind where Steve Jobs notices POWER6 got released and beats the pants off of EVERYTHING on the market, and thinks, we need to get back into Power and take on the server market.

There is certainly nothing stopping Apple from having a triple product line with both PPC and x86 on the server side (PPC taking up the big iron and high density market and x86 for the people who need it), and ARM on the side for embedded. Their same OS supports them all brilliantly.

I think it's far more likely that they would go in this direction anyway and go back to their heavy marketing of the clustering solutions which seem to have fallen further and further by the wayside. Maybe we're looking at the birth of the Apple Blade Center. On the outside, we may be looking at Apple's answer to SpursEngine in the video workstation market (Apple already have this market tied up out of sheer force of will of most designers).

But that's the geek in me, the bit that says, what a waste it would be to get PASemi into shoring up the ARM chips in their phones.. and I think moving back to the desktop with some in-house Power Architecture chips designed specifically for Apple is going to piss off more people than ever (not just small developers, but what about all the effort Adobe etc. have put into moving from PPC to Intel, and 32-bit to 64-bit at the same time both on the G5 and Core Duo, only to be told to move right back? That is not to say that the small developers won't also be annoyed, even the Apple Cult won't buffer the effect of having dicked everyone around for 3 years about using x86 chips). In any case all they'd need to do is add USB, FireWire, SATA and drop a graphics chip on the PCI Express bus and they'd have their next gen desktop, faster than all the G5's, in a box the size of the Mac Mini..

http://www.eetimes.com/news/latest/show ... =207401605

Geeeh - and I was thinking of the same things Neko lined out above. Well, last time Apple surprised me in a positive way ist quite a while ago - and I was hoping they would this time... :-(

Forget it, Apple Computer is gone, long time ago.
They bought PA Semi to ditch PowerPC

Oh well. Another one bites the dust. It will be interesting to see how the DoD (referenced in the article above) give Apple a kick about end-of-lifing a chip they're using in so many systems.

It will be interesting if PA Semi didn't know about or didn't tell Apple about these systems that they might have to supply for military product lifetime (which is just about as long-winded as automotive - 15 years or more)

We'll always have POWER6 :)

sucks that apple doesnt offer both, powerpc and x86. perhaps they will. :-D

More thoughts...

There is a game market Apple has not entered. The iPod would make a good remote (an acceleratometer is in the iPod Touch -- the Wii remote has one). The iPod market has matured (see lower price for Shuffle). They need something new and that would be a good 'up-sale.'

We don't buy the 'headhunter' patter. This is bigger.

Also, PA Semi could stay a separate company and just have one shareholder. PA could still sell to others. The semiconductor market is going through a tough time. This would just be gravy for Apple.

R&B :-)

From a good inside source (very close to PA-Semi) I heard that Apple intend to totally ditch ppc and current pa-semi offerings and instead reuse the design team for ipod/iphone. Which means, that apart from current PA-Semi's customers (like the DoD), the rest of us can forget any future PPCs from PA-Semi. Quite sad...

That is what we think Apple wants everyone to believe.

Does the PA Semi license include the CELL technology? It would be logical to assume this. It was all done at about that time.

...back to the game idea wrapped into the whole digital TV package -- you have hardware people at PA Semi that really understand the processor, you have software people at Apple that are certainly in the know about that part of the package -- a digital television needs to frame store and do the conversion from a digital television signal to a scanned system which is really just a computer with some memory -- if that became the rendering engine for the game machine you would certainly be able to do some cool stuff -- especially with an iPod Touch remote.

OK, enough speculation from us!

R&B :-)

I don't buy that either, not for a second. You can never take for granted that people will stay forever (or for a second) even with contractual quarantine periods in the employee contracts. People with these kind of expert skills are wanted everywhere and they tend to move from time to time. The 15th century is long gone after all, you can never *own* people, and you don't have to buy a company for 270+ million of dollars in order to get some clever engineers hired.

Some people suggests that all Apple wants from the deal is to use the skills of these people to "PWRficient-ify" some x86 or ARM design or whatever. I'm not even going to ask the question "why, what's the point when they have already obtained a totally unique PWRficient CPU that beats everything and that they can use as is". All these years of development, all the efforts and enormous investments in dollars, has been funneled into the Power based PWRficient CPU architecture. A lot of value is accumulated in that. And the specialized and unique silicon design is what brings the unique PWRficient features, *this* silicon design is what Apple bought in PA Semi, and that design is not ARM or x86. Changing the "platform" on a silicon level can not be compared to such a "trivial" task as replacing CPU family under the hood of a Mac computer, no it would mean starting all over again, perhaps with accumulated knowledge and experiences but still more or less from scratch. Years and years of *unnecessary* development, instead of continuing down the path they *are already on* and have *struggled* to get up on. And if the plan is to scrap the achievements, start all over again for the sake of it but to do it differently (because of some "Power hatred" at Steve Jobs office or whatever), how many of those engineers will stay on ship and how many will start looking for a more rewarding way to spend their time?

Sooner rather than later, the smart people (from PA) will leave Apple, that's for sure.

You may think what you like, but you're mistaken. We're not talking about word typists, we're talking about chip designers, do you really think that these guys move around and change companies every six months? Sorry that just doesn't happen. Companies might try to develop procedures and documentation, but nothing beats a good developer/chip designer/artist/etc. Sorry that's the truth. If it wasn't for the fact that PA-Semi didn't have a good, no a GREAT design team, it wouldn't have achieved anything. And it wouldn't do it if this team changed every now and then.



This is very likely, I find it very difficult to stay in the same company that would want to ditch my life's work, unless I was in dire financial needs or the replacement project would more than pay for it.

And anyway, I'm not talking out of my head, I heard it from a *very* reliable source. They DO plan to stop sales of the chip to new customers and only continue to support the existing ones. Unless the DoD messes it up for them at least.

Note: I sincerely hope I'm wrong! Obviously my guess is as best as my source's.

Indeed the PA Semi team of engineers created a great design (where did I diss that?). And indeed the feudalism is dead, so there is nothing that physically or legally (except potential time limited quarantine obligations (often with full payment during quarantine period)) that stops any of the designers/engineers from leaving for whatever reason (not agreeing to new management, want to get on with life elsewhere, simply bored, whatever).

I'm not claiming that Apple weren't interested in the skills of PA Semi employees, *of course they were*. But I seriously doubt that this was their primary reason. I think they saw something they liked *beyond the people* (that can leave anytime anyway), I think they saw some potential in a dual core processor with ahelluvalot of integrated high speed controllers, that only consumes 15 Watts running at 2GHz (and even less when running slower one would presume), and a roadmap to go with that. I don't know about PA Semi's Product Life Cycle policy (them being a startup and all, perhaps with uncertain funding), but chances are they have at least one more chip nearing completion, and have already getting started outlining a third generation.

All in all, what I'm saying is that I don't think it's very likely that the employees of PA Semi *alone* was the prime reason for Apple making this takeover. I sincerely believe that Apple realized PA Semi had something going, something existing or upcoming, something that will fit very well into whatever upcoming Apple product (to come within a year or so and *not* after half a decade of "PWRficient-ifying some x86 or ARM design").



Of course you are right. Apple has designed their own proprietary chipsets etc before, and have you ever seen them selling these chips as parts on the open market? No, you have only seen Apple chips if you open up the case of an Apple end-user product.

But that doesn't necessarily mean that DoD (or whatever existing customer) won't get their supplies of CPU's. AFAIK, PA Semi issued some note to its customers that "we cannot guarantee the supply of CPU's in the future". Of course they have to say that. However, Apple may decide to let PA Semi continue operating in the shape of its current organization, and they could very well decide to honor any existing contract way into the future. But I would be very surprised to see PA Semi CPU's freely available on the market for new, third party, none Apple products, in the future. I think that from "our" perspective, PA Semi has left the building.