The Apple tablet has been a topic of discussion ever since Steve Jobs yanked the Newton out of the product line in the late 90's. Speculation was rejuvenated at last month's conference call, where Jobs himself was on hand to say of the Netbook category:
"entrance into that category" is the iPhone. But then he admitted that Apple is going to "wait and see" how the category evolves and "we've got some pretty interesting ideas if it does evolve."
To many, including me, that seemed to scream "tablet." To get a better idea about this category, however, I talked to Bob Morris, director of platform enablement for ARM's mobile processor group, who had a wealth of information but (of course) could not speak about anything specific to Apple.
It comes down to a couple of things:
1. ARM and Intel are doing battle because they are encroaching on each other's turf in much the same way that Verizon and Time Warner are in the Cable/Phone/Internet space. ARM is getting fast enough to build a netbook while Intel is getting miserly enough in power to be put into a phone. 2009 will be a huge face-off between the two in this field.
2. Intel's business model doesn't allow for switching between hardware vendors. They design the chips AND build them. You want an x86-based system, you have to go to Intel (or sometimes AMD or VIA) for the chips. With ARM, there are any number of vendors working on the platform. TI, Marvell, Qualcomm, Samsung (like in the iPhone), NVIDIA, and even Intel and Apple (rumored) are all ARM licensees and are working on chips. The competition is fierce, not only in processor speed, but in price. Low power ARM chips have come down in price to below $10/each for some models. They come in many flavors and varieties.
3. As things stand now, ARM chips can do almost everything (besides run Windows XP) that a Intel chip can. Recently some Intel execs publicly said that Atom would kill the ARM Netbooks in browsing. One doesn't have to look further than the iPhone to see that isn't quite the case.
If you have any doubt that ARM's Cortex can hang with Intel's Atom processors, take a look at the video below.
4. The first netbooks (that Bob can talk about) are coming out in mid 2009 running Linux. They will likely be running on the Android platform. Two different vendors are said to be working on Android for a Netbook.
5. While pretty close in speed, the ARM Cortex A8 absolutely kills the Intel Atom in power efficiency. We are talking Apples and Oranges here. Not double, but an order of magnitude better for ARM. Some ARM chips routinely use 10-20 times less power than Intel for similar operations. Battery usage with ARM chips in prospective netbooks could be measured in days, not hours much like smartphones.
If you recall, the reason Steve Jobs moved from PowerPC to Intel was performance/Watt. If an Intel processor uses 18 times the power per unit of processing performance, "We know what we have to do".
6. Motheboard design is a huge consideration. With SoC (System on a Chip) an ARM-based motherboard can take up 1/10th the space of an Atom chipset. NVIDIA, for instance is building ARM-based systems that eschew the motherboard alltogether. Just power in, ports out.
All of this is just theoretical, however. Intel already has the big lead in the MID/Netbook space. With the exception of HP's older Mini (VIA), just about every Netbook out there is using an Intel Atom processor. On the ARM side, you have the Nokia N810 line of tablets, the iPod touch and not too much else, unless you travel down into smartphone land.
One very interesting product that shows what a Cortex-based product can do is the Open Pandora. While decidely clunky and unrefined in appearance, the guts of this little machine tell a different story. It lists the following Specs:
To recap, here are the reasons, I believe that Apple will choose the ARM platform for their upcoming Netbook/Tablets.
OK, if reason doesn't work, check out ARM's internal Powerpoints.
Notice that they have a certain light Apple Notebook pictured in the top right corner?
update: a comment from a forum post:
As an ARM assembly coder from back in the day (and by back in the day I mean for Acorn, who invented the ARM chip before they spun off ARM to handle it) I have a very positive feeling about this.
Apple has much experience with OS X on ARM, and likely has internal XCode with ARM as a checkbox option. Their use of universal binaries supports more than just two architectures. It would be trivial from Apple's POV to use the forthcoming drop of PPC support to add ARM support.
One of the most attractive features of the ARM chipset is price. Back when I was designing boards around the ARM7500, one could be had for $5, or less in quantity. That is for the processor and entire chipset, including video. Granted, back then it was a 40 or 50MHz device, but it drew less than a watt. Now, we have 600 and 800MHz parts that equal the performance of the Atom, draw similar currents and have bluetooth, wifi, ethernet PHY and video on die.
So, picture if you will that Apple decides to support the ARM architecture with OS X... Think a simple line of low cost 10" netbooks, think of possibly sub-$300 OS X based consoles (or STBs as we used to call them before we knew what to do with them!)
When thinking of some of the accomplishments of ARM, remember: ARM is the definitive architecture for STBs, embedded boards, controllers, and cellphones. ARM cores outsell EVERY other architecture. There are more ARM cores in the world than every other architecture combined.
They are cheap at every level: cheap to license, cheap to fab, cheap to design boards with (so many reference designs), cheap to manufacture (you can buy full ARM computers for under $100) and have very cheap power budgets - usually around 1W.
If Apple has an ARM license, a chip design house with ARM experience and two existing products with ARM cores and good OS X support, it would be simply good use of resources to use this IP more broadly.
How? Well, that's what this thread is all about