Forecasting Apple's Intrinsity Acquisition

IEEE Spectrum had perhaps the last Intrinsity interview before the Apple acquisition

3 min read
Forecasting Apple's Intrinsity Acquisition

Now that Apple has officially confirmed that it has purchased the Austin, Tex.-based smartphone CPU redesign firm Intrinsity for an estimated $121 million, tech blogs have been buzzing about what it all means. Despite all the rampant speculation, though, one of the only things known for sure is: The now-Apple-owned Intrinsity designed the iPad's CPU— basing it around Intrinsity's "Hummingbird," a modified ARM Cortex A8.

But will Intrinsity's secret CPU hot-rodding technique (which IEEE Spectrumdescribed in detail in January) be a game changer for the next-generation iPads and iPhones?

Cue industry analysts who will say, essentially, "Maybe. Although... maybe not."

Here's one other thing we do know, though. Before Steve Jobs' borg descended on Intrinsity and put Apple's trademark cone of silence over it, IEEE Spectrum had a lengthy sit-down with the management and engineering team (Sept. 2009) for what was one of the company's final pressers as an entity that could speak on the record.

At the time, of course, the iPad was just one of numerous fanboy pipe dreams of an Apple netbook/tablet that could dominate the market the way the iPhone overtook many other smartphones. 

Intrinsity's CEO Bob Russo told Spectrum that the company was applying its same chip streamlining techniques to ARM's Cortex A9 -- the logical successor to Hummingbird and now speculative candidate for future iPads and iPhones. ("We are engaged in a multicore A9 development. ... I just can't tell you who it's for," Russo said at the time.)

Below are some further excerpts from that confab on a stormy Thursday morning in Austin. Watch out for interesting mentions of Apple’s earlier chips design purchase, PA Semi.

IEEE SPECTRUM: How do you take a pre-existing ARM chip and make it faster?

Bob Russo, Intrinsity CEO: There are two ways to enhance a chip. One, you can take the existing product and enhance that core -- and keep it totally software-compatible for the customer at the end....The other way is to go in and change the entire architecture. That's a bigger undertaking. Companies have done that. Qualcomm's done it, for example, [with their redesigned A8, "Snapdragon."]

SPECTRUM: So "Hummingbird" follows the first of those two models, right?

BR: Yes. That's the model that we like, that we're set up to produce. And we're trying to make it so that we produce 8 or 9 of these a year.

SPECTRUM: Eight or nine chip redesigns?

BR: That's the ultimate goal here. To enhance the cores. You can't do that if you fool around with the [CPU's] architecture. But if you have the right tools, technology, and know-how -- which we do -- you can take on multiple customers who require a higher frequency part than what's available in the standard marketplace.

[Russo describes Samsung's desire to speed the 650 MHz A8 up to 1 GHz.]

There are only two ways to do it. One, you have your own internal team to develop it, like Qualcomm has. And you go for all the work to develop a high-speed part. Or you come to us.

Apple went and bought PA Semi for that very reason. Apple bought PA Semi because they wanted to control more of their destiny...to have their own team to be able to come out with a part that is faster than is currently available in the marketplace.

Brent Chambers, Intrinsity director of engineering: We're taking Ferrari technology that typically takes Ferrari development time and Ferrari dollars to afford, and we're bringing that to the masses in a much lower cost and much more specific application. It gets you 90 percent of the speed for a fraction of the cost.

BR: We're an order of magnitude less expensive than everyone else who developed this [enhanced A8 core] on their own.

SPECTRUM: And that's because you streamline the CPU, but only in the choke-points, right?

BC: That's what gets us in the power envelope for the mobile space. But the cost is the automation.

SPECTRUM: So the way you automate your streamlining of a CPU is how you can cut your costs.

BR: It's all the patents we have, all the circuit technology we've developed. There's another component here, too. We're a premier shop of talent. We're the last-standing independent, high-end CPU design house in the world. There's no one else left but us. The last one was PA Semi, and they got bought by Apple.

SPECTRUM: So Hummingbird wasn't designed with a netbook in mind, was it? It's more for a smartphone, right?

BR: I would think that's the case. Where it ends up, I believe -- my opinion, I've never been told this -- it's going to end up in more places than just a smartphone. And one of the potential places could be a netbook. Potentially.

Photo: Robert Galbraith/Reuters

The Conversation (0)

Why Functional Programming Should Be the Future of Software Development

It’s hard to learn, but your code will produce fewer nasty surprises

11 min read
Vertical
A plate of spaghetti made from code
Shira Inbar
DarkBlue1

You’d expectthe longest and most costly phase in the lifecycle of a software product to be the initial development of the system, when all those great features are first imagined and then created. In fact, the hardest part comes later, during the maintenance phase. That’s when programmers pay the price for the shortcuts they took during development.

So why did they take shortcuts? Maybe they didn’t realize that they were cutting any corners. Only when their code was deployed and exercised by a lot of users did its hidden flaws come to light. And maybe the developers were rushed. Time-to-market pressures would almost guarantee that their software will contain more bugs than it would otherwise.

Keep Reading ↓Show less
{"imageShortcodeIds":["31996907"]}