Will Apple move all of its products to its
own chip architecture?
Think the iPad mini is the most important
tablet Apple has released this year? There’s certainly plenty of evidence to
support that belief. Its size, weight and form have hit a sweet spot that
balances comfort with usefulness. The price appears to have deterred very few,
and early reviews, including have been enthusiastic.
There’s another new iPad that’s every bit
as important for Apple’s future, however. One that, while given stage time at
Apple’s launch event, almost slipped out unnoticed amid the excitement of the
mini release. The iPad 4 (I refuse to call it ‘iPad with Retina display’)
arrived barely six months after its predecessor. Why so soon? Some commentators
assumed it was an admission from Apple that the iPad 3 was a lackluster release
that had failed to offer enough to tempt new customers or those seeking to
upgrade an original iPad or iPad 2. Sales figures for July-September seem to
corroborate that. Fourteen million units sound impressive, but that’s fewer
than analysts had expected. Yet speak to anyone who has used one for any length
of time and you won’t hear a word said against it.
iPad
4
The real answer, I believe, lies in the
benchmark tests that have been run on the iPad 4 since its release. They show
that while the iPad 2 and iPad 3 made incremental performance improvements on
their predecessors, the iPad 4 is more than twice as fast. Both the CPU and GPU
in the iPad 4’s A6X system on a chip (SoC) blow previous iPads out of the
water. Apple had the A6X, a modification of the A6 chip in the iPhone 5, ready
to go. It simply couldn’t sit on it and allow Samsung and others to render it
ordinary.
Talking of Samsung, the A6X, a dual-core
SoC, scores marginally worse in benchmarks than the quad-core Exynos 4 used in
the Galaxy Note 10.2 ‘phablet’. Take out the tests optimized for multiple
cores, however, and the A6X is comfortably faster, meaning that for most
real-world tasks, the iPad 4 is more powerful. If Apple had stuck to its annual
iPad refresh, the A6X would have lain unused for six months and this
competitive advantage would have been eroded.
The performance of the A6X is testament to
how far Apple has come in the very short time it’s been making its own SoCs. In
a few short years, it’s caught up with and overtaken Samsung, a company that
has been developing its own chips for a great deal longer. Apple now has one of
the best teams in the business when it comes to designing the low-power,
high-performance SoCs needed in today’s smartphones and tablets, thanks to its
acquisitions of Instrinsity and PA Semi, and some astute recruitment. The
question now is how will it use that expertise in the future?
A6X
chip
There have been rumors in the past about
Apple producing a MacBook Air running on an ARM-based SoC, but there was
little to suggest it was something the company was seriously considering. Now,
however, with the recruitment of Jim Mergard from Samsung and the establishment
of Bob Mansfield’s Technology division, the landscape has changed considerably.
US business magazine Bloomberg ran a story earlier this month in which it
claimed Apple was ‘said to be exploring a move away from Intel.’ It cited
‘people familiar with the company’s research’, but the story didn’t contain a
direct quote from anyone with any knowledge of Apple’s plans. Nor did it
provide any real evidence that Apple was planning a switch from Intel.
Nevertheless, the possibility of a Mac running on an ARM-based Apple SoC looks
less remote now than it did at the beginning of the year.
There are multiple hurdles
to be overcome, of course, and they may not all prove surmountable. But the
first of them was cleared at the end of last month, when ARM announced the A-53
and A-57 processor cores, complete with support for 64-bit computing, which is
necessary to run the 64-bit OS X.
The second obstacle is raw power. While
Intel’s x86 architecture focuses on performance at the expense of power
consumption, ARM’s cores do the opposite. For mobile devices and small laptops,
that’s great. It’s less good for the Mac Pro. Would Apple transition to
ARM-based chips for part of its Mac range while leaving its pro machines on
Intel? That seems unlikely. In order to persuade software developers to make
the necessary changes to their code, it might have to sell it as an all-
or-nothing deal. Where would that leave the Mac Pro?
Where
would that leave the Mac Pro?
Then there’s OS X. While iOS is based on
the same kernel as OS X, it’s very different in use. Would a transition to ARM
cores mean iOS running on a Mac? Or would Apple attempt to produce a hybrid of
its two operating systems, as Microsoft has done with Windows 8? I suspect one
of the first jobs of Mansfield’s Technology division will be to answer those questions
and many more like them. While they’re doing it, expect to hear much more about
the possibility of Apple shifting towards its own chip architecture. If nothing
else, it will be a useful tool for bargaining with Intel and pushing it to
produce better low-power chips of its own. iPad running on Intel, anyone?