Slashdot Mirror
Apple Buys a Chip Company for $278M
An anonymous reader writes "Apple's just bought a chip company, P.A. Semi that could make chips for iPhones and maybe iPods. Apple wouldn't reveal the exact plans, but Dan Dobberpuhl, lead designer of Alpha's chips, is known for making super efficient processors, like a 64-bit dual core last year that was supposedly about 300% more efficient than the nearest competition, using only 5 to 13 watts at 2GHz. Apple's quarterly results are later today, so we might hear more about the deal. This is something of a blow to ARM, especially with the mobile chip market heating up recently, with forays by Intel and Nvidia adding to competition from established players like VIA."
The Apple Product Cycle
Discuss.
Apple is increasing moving into embedded and mobile markets more and more with iPhone, iPod, etc. I think we're going to start to see more small footprint devices from Apple in the future, maybe even something that creates a whole new product category. Information-based devices and appliances are the future, and Apple is one of the companies poised to do great things in this market.
This is a precursor to some big things and I think Apple is taking itself in an entirely new direction.
Just me $0.02.
My blog
Maybe, but the PA Semi guys have already shown that they can produce good designs for two ISAs, adding a third wouldn't be beyond their abilities. I wouldn't be surprised if we start seeing their PowerPC chips in things like the AppleTV and ARM cores designed by the same team in handheld devices.
I am TheRaven on Soylent News
G5 LAPTOP WOOOHOOO!!!!!!!
5 to 13 watts at 2GHz? And 100 cycles per command? What kind of range is that "5 to 13"? And obligatory car analogy: 20 to 52MPG seems rather large deviation...
I'm very curious where Apple is going with this. P.A. Semi so far has only put information about one design up on their website and it's been there at least since the rumor that Apple was going to buy them (shortly before they went to Intel). That chip, being somewhere between Atom & Core2 I suppose, doesn't seem to me to good fit to any of Apple's existing products.
The idea that hidden up their sleeves P.A. Semi has an ultra efficient SOC design for a next generation iPhone/iPod/Tablet is sort of interesting but I'd be really surprised if a dark horse came out nowhere and outdid the various upcoming Intel offerings or even the existing ARM SOC designs. Intel is very, very proud of their Low Voltage and Ultra Low Voltage parts but surely that added cost doesn't make it worth Apple's while to go out a buy a company.
The idea that P.A. Semi has a next generation chip suitable workstation or home computer applications for me is even more unlikely. I think it would have to some chip to really motivate Apple to go away from Intel for their Mac lines.
Nothing in the world is more dangerous than sincere ignorance and conscientious stupidity.
There comes a point where designing/making your own chips is more profitable (or less costly) than having someone else design/make them.
Jobs is a control freak, so maybe cost has nothing to do with it, but as a business move, it has the chance to work for you.
As for being a good value for the purchase, Apple seems to think so.
Bearded Dragon
They can buy Enron for all I care.
Invenio via vel creo
Ultra-low power chips are enormously important for several key Apple areas. They're buying technology; but also expertise.
There are all sorts of things that Apple could be looking at this for Apple TV, iPhone, Tablet's, Apple EEPC/Macbook Air, Newton, iPod or even something different.
But at the same time they like to work with Intel on chip designs. They had one specially made for the Macbook Air. Besides the implied threat of an ability to go their own way they might find that collaborating with Intel on design may give them a massive say in the ultra-low power chips end up.
Without directly using PA Semi chips they could use PA semi to improve their own power consumption. Ultimately, $278m isn't actually that much money given the importance of low power performance to Apple across most of their product line.
the more they over-think the plumbing the easier it is to stop up the pipe
I've never been a big fan of Apple chips. I prefer sea salt and vinegar. Preferably kettle cooked.
FOXTROT UNIFORM CHARLIE KILO
Except that, ARM unlike all the other companies named, doesn't actually _make_ chips. It designs cores, which it licenses to other companies - many other companies in fact.
Apple changing architecture I suppose is possible (not like they haven't before), but it seems like an odd step when there are many ARM-based manufacturers to choose from. ARM themselves wins no matter what.
I've always wondered why Steve Jobs didn't announce a dual-architecture strategy from the get-go. But perhaps that was the plan all along, and Apple simply needed to announce "Intel only" to get all their developers moved as quickly as possible to universal binaries. Now that Microsoft and Adobe, the last holdouts, have complied, Apple can go back to a dual (or even tri, with iPhone's ARM) architecture approach, choosing the right processor core for the right device and maximizing its flexibility and distinctiveness.
For example, the PowerPC core would be perfect for AppleTV and possibly a new Mac nano, where the cost of an Intel chip simply doesn't make sense. Apple is probably losing money on every AppleTV box right now. Every universal binary already runs on PowerPC, so all the applications and development ecosystem are already in place. The fact VMware and Parallels don't run on PowerPC is a feature, not a bug: Apple can wean some more users away from Microsoft Windows as certain devices hit the market and get some better market segmentation. Users who want Intel can buy Intel, and users who want alternative form factors, alternative power consumption profiles, lower cost, and/or new device categories can get PowerPC under the hood and still run the full Mac OS X portfolio of software. And having their own chip company helps keep Intel honest. Apple probably didn't like Intel's forced march from Santa Rosa to Penryn. That was inconsistent with Apple's longer product cycles. And all the game consoles are PowerPC-based, so that could be appealing if Apple ever wants to entice some game developers over to some of their devices. (Games do tend to work down on the iron.) IBM continues to underwrite PowerPC for its own server lines and has cranked up POWER6 to 5.0 GHz in its servers, way beyond Intel's best, so it's still an architecture with a lot of interesting advantages.
This isn't that surprising really. Apple is very much a hardware company these days. They must ensure that they have control over not only the software and external design of their products, but also the "guts" of their products.
Jobs is obviously a fanatic about design and as time goes on, we'll see more and more "vertical integration" like this. Also remember that Apple is growing and is relatively flush with cash.
This is perhaps a very solid investment (bean counting as it were), outside the realm of design, software, hardware, and other technical matters.
How to Download YouTube Videos
With the iPhone, iPod, and so on, to save power and keep prices down you really want system-on-a-chip designs. But if you buy commodity, then you get the same system-on-a-chip everyone else can get. It's hard to do something different. For desktop machines, you can distinguish yourself by the combination of features (even though Apple machines aren't that different to anyone elses these days, except possibly for firewire), but you can't do that in the embedded/mobile space if cost and power dictate it's a single chip design. So, my guess is they want their own in-house capability to build system-on-a-chip designs that are different from everyone else. Different in what way though, I have no idea.
It's debatable. There have been plenty of people who've said that if they were in Apple's position, they'd license the Mac OS and let others battle it out to make the hardware. That would be a seriously foolish move for Apple partly because they make so much money off of hardware, but also because one of their main design philosophies is designing the whole "widget". Apple seems to really like being in direct control of as many pieces of their products as they can. I bet if they thought they could realistically design and manufacture their own CPUs, they'd do that to.
Maybe they've got something in mind but that they don't think they can convince different chip makers to move in that direction. They've just got a ton of cash laying around, maybe they felt like taking a little risk is worth it to get certain types of chips that they really want. This isn't Apple just blindly jumping into an industry that they have no idea about. There's got to be a specific reason for this.
One time I threw a brick at a duck.
Universal Binary.
Hey, you never know.
Before you say "Apple will never do that", let me remind you of some things we all heard before:
- Apple will never release a low-cost computer
- Apple will never make a music player
- Apple will never enter the cellphone market
- Apple will never dump support for Mac OS classic
- Apple will never switch to Intel
Vertical integration is not necessarily stupid.
This company that they bought simply licenses the Power architecture from IBM, and then makes manufacturing/design changes to make the chips more energy efficient. They make a dual-core 2GHz power chip with 2MB of cache, and integrated DDR2 controller with DMA controller... burning just 5-13W. AFAIK, no one else is making anything similar. Atom seems similar on the x86 side, but is larger and does not have the same features.
If Apple gained the ability to produce a product that others cannot match, then the move was not stupid. If Apple bought a commodity chip maker, then the move was stupid. They can always spin it back off if the product becomes a commodity.
W..w..W - Willy Waterloo washes Warren Wiggins who is washing Waldo Woo.
This was stupid.
Unless they have something really, really specific in mind that the market can't provide.
Remember, macs had SCSI despite the expense because the market had nothing that did anything similar available at the time. They had SCSI until peripheral busses like Firewire and USB arrived, at which point they dumped it like a dead rat. See also their long string of proprietary monitor connectors - the 25-pin mac standard, Applevision, ACD - the latter two of which provide essentially the same functionality for two different generations of technology.
In my experience, Apple's the kind of company who's willing to let other companies make the bits (including software), if the bits do what they need. If they can't get anything useful from third parties, they will make it themselves. The best example there (after the Mac itself) would be software - MP3 players on MacOS were unstable, crash-prone winamp clones until Apple bought an audio software company and then iTunes came along.... and entry level through prosumer (and even pro, depending on who you talk to) video editing on the mac SUCKED ASS until Apple bought a chunk of video editing software and twisted it into the awesome that is Final Cut Pro.
My blog
Maybe they want to release a PPC Mac (maybe as a set-top or something) just to keep price and development pressure on Intel? If there is a new PPC Mac in the field, software vendors might feel compelled to continue shipping universal binaries instead of going Intel-only.
The problem with my little hypothesis there is that Intel already has price pressure from AMD on the laptop/desktop and the various ARM players on the embedded front.
So here's a better hypothesis: These guys have figured out a manufacturing process to take hungry chips and make them into thrifty chips. Apple would have loved their dual-Power chip that uses 5 or so watts back before the switch to Intel. The G4 that they had in their laptops never got anywhere near that, and ran at a slower clock-speed, and was single-core! Even if they never make another "Mac" with PPC, they might use this technology to adapt other cores - or release a OSX-based non-Mac product (like they have with the iPhone, iPod, and AppleTV).
W..w..W - Willy Waterloo washes Warren Wiggins who is washing Waldo Woo.
When you are in Apple's position, you leave the semiconductor development to others and let them battle it out to make the best component at the best price.
This was stupid.
But what if Apple just denied all other small device makers the use of a chip that's three times more efficient than the competition?
Then Apple has the best chip, that no-one else can have at any price...
It's easy to say something looks stupid now, but without the roadmap for the company and for Apple you are just guessing. And Apple has a record of making smart choices, especially in the last few years. Therefore, we can reasonably say it's way more likely your pronouncement is of the "iPod Lame" sort that SLashdot is so famous for.
"There is more worth loving than we have strength to love." - Brian Jay Stanley
When you are in Apple's position, you leave the semiconductor development to others and let them battle it out to make the best component at the best price.
The semiconductor industry is really only interested in creating the best "general-purpose" semiconductor. If a hardware company like Apple has specific or divergent needs, their choices are to pay through the nose to "partner" with a chipmaker to accommodate their requirements (if they can find one willing), or to buy a chipmaker outright and do whatever they want.
Apple doesn't want to be stuck in the situation they were just a few years ago, needing IBM to improve the PowerPC so their business could move forward, but finding IBM uninterested in investing the effort because it wasn't profitable enough to them.
The odd thing is that it is a POWER instruction set CPU. Whilst I know that Apple are flexible, it seems an odd move.
Far more likely is that Apple want them to design a 2GHz dual-core ARM compatible CPU. Depending on the design of their current CPUs, it could be possible that this work could just affect a relatively small part of the overall CPU (although still a lot of work).
Then again, why not move to using POWER in Apple's mobile devices instead of ARM... hmm.
It's less than 2% of Apple's savings, and I believe the company already has clients and sales so it could just be a good investment in the long term.
Well, if you own a large percentage of a market, and that market requires a specific type of processor, then you're essentially - by buying large quantities of it - paying for this processor's development anyway, thus subsidising your competition.
Also, if you have a chip that is better than everyone else's, and you own that chip, that's a huge competitive advantage.
I'm not saying this is the case here. Just saying that there may be sound reasons for a move such as Apple's.
We need to move away from x86 to a modern design, rather than one that has gradually been modified beyond all recognition and hacked to gain 32 bit then 64 bit compatibility, etc. As an Amiga/Mac user for most of my early life, I've always thought of x86 as an inferior and inefficient chip design. Apple has demonstrated twice now how well they can adapt their OS for any architecture.
Apple is no longer free to change their CPU architecture. They are "locked in" to x86 due to their dependence on running Windows. Mac market share jumped significantly when they switched to Intel and jumped significantly again when they offered Windows compatibility. While Windows emulation has been available since at least G3 PowerPC Macs, it suffered greatly because it had to emulate the x86 CPU instruction set. The switch to Intel made that unnecessary and made emulation viable. The decades old question, should I go Mac or PC, largely ended. You could have both on a single machine. When dual boot became an option then the last barrier fell, those who needed absolute performance, gamers for example, could now have both on a single machine. Switching to a non-x86 Mac architecture would probably destroy the 50% increase in market share, 4% to 6%, that Intel brought them.
You are echoing the same argument that the PowerPC consortium made in the very early 90s. The flaw in their logic and yours is that Intel can overcome x86 inefficiency and difficulty of working with it by spending more money. PowerPC was more efficient and a modern design that could more easily be enhanced, but Intel could throw 10x the resources at x86. PowerPC did not really fail because it failed to improve, it failed because no one ever imagined that Intel could get the x86 to the levels of performance that they did. The PowerPC folks expected Intel to try to move x86 users to a new CPU, Itanium as it turned out, and that would break the x86 lock and allow buyers to consider other non-x86 alternatives. I believe you are making the same mistake. Consider that the x86 architecture is really a facade, that underneath this facade Intel is free to change from one modern RISC design to another, or to whatever is next, allowing them to increase performance without breaking compatibility. On the fly translation of x86 operations into RISC micro-ops combined with reordering and other technologies is going to be far harder to overcome than you suggest.
One little detail you overlooked is important to understanding what Apple might possibly do with this stuff.
Apple doesn't have much in the way of ARM code at all, to the extent that nearly all of their ARM code is generated by a compiler. Apple has C and Objective C code, and has LLVM sitting between the hardware and the Apple application source code. Apple can run on any hardware platform they like. They can support more than one hardware platform at almost negligible marginal cost. While the rest of the industry flails about, with their obsolete notions of "platform wars", Apple can simultaneously participate on the industry standards platform (or platforms as the case happens to be) and also invent a better platform, for one or many other product categories. Those can also overlap.
Apple is essentially platform agnostic, with respect to hardware.
If you mod me down, I shall become more powerful than you could possibly imagine.
There was a lot of talk a while back of Apple buying up Sun or SGI, or having a merger, which would enable them to muscle into the hard-core server market, soup up their unix-based OS, and get a ton of chip-level hardware wizards on board. Now knowing that this is basically impossible, I wonder if execs at those two now also-rans are getting ready to byte the bullet or bail out.
That's a gross distortion of reality. PowerPC always did up through the G4, and the G5 comes close as well. I'd consider the G5 a pretty modern desktop processor. Yeah, it uses a handful of instructions that are cracked or microcoded, but those are the exception, not the rule, and it is easy to build fully-functional code that doesn't use any microcoded instructions. With x86-derived CPUs, microcoded instructions are the rule, and it is almost impossible to write working code that isn't heavily microcoded in the CPU....
I think you mean instruction cache. Instead of the instruction cache, though, you now have truckloads of glue logic to split those instructions back into their RISC form internally. You only think you're saving die space by going CISC.... The only thing you really save by going CISC is disk space/RAM to hold the code, and since that's usually tiny compared to the size of data, that's just not a particularly important savings....
There are plenty of valid objective reasons to switch away from x86. The people arguing that RISC was inherently better than CISC weren't wrong; the people building the CISC architectures simply had orders of magnitude more R&D money to throw at the problem. While it is amazing that Intel has been able to wring as much performance as they have out of the x86 ISA, it is still important to note that modern x86 CPUs are basically RISC CPUs with massive microcode engines and cache wrapped around them. It should therefore be plainly obvious that if we used that underlying RISC instruction set directly instead of the CISC wrapper ISA, we could get rid of a huge chunk of the die size, representing a huge power win, a huge thermal win, a huge manufacturing cost win, and a huge manufacturing yield win, all without changing the actual performance of the chip in the slightest, memory bandwidth for instruction prefetch notwithstanding....
Of course, my comments here are all about the x86 architecture in the long run. Eventually, Windows (or at least 32-bit Windows) will be a footnote in computer history. When that happens---when only application emulation (rather than OS emulation) is needed to maintain backwards compatibility---the x86 architecture's legacy support will no longer be as critical as it is now, and we will likely see it start to fall away as legacy cruft inevitably does. In the short term, though, that legacy compatibility is still at least moderately useful, so the industry puts up with the x86 ISA....
Just my $0.02.
Check out my sci-fi/humor trilogy at PatriotsBooks.
Apple is essentially platform agnostic, with respect to hardware.
Due to economies of scale, Apple must choose hardware with somewhat similar production volumes as the most popular hardware platform in that category. For desk-/laptop, that means x86. If you remember, Apple dropped PPC because of lower production volumes which translated to less R&D and cutting edge fab investments, which in turn led to PPC falling behind x86 in price/performance. On the handheld/ultramobile side, any new hardware platform will have to compete with the large existing ARM market (and soon the well-funded and x86 compatible Intel Atom). Unless Apple can get economies of scale within shouting distance of the most prevalent hardware platform in a given category, they will be in for an expensive lesson in how "obsolete" the "platform wars" are.
Not to mention that while Apple might be hardware agnostic, 3rd party software isn't.
Anyway.. they bought a chip design company that specializes in high performance low power network storage server chips, not mobile/laptop chips. In the short term nothing is likely to come from this except perhaps a hefty upgrade in the next XServe RAID.
If J.K.R wrote Windows: Puteulanus fenestra mortalis!