In its first public presentation, AMD’s new leadership explained a
new direction of the company to align itself with an evolving CPU and
GPU market. In 2013, AMD will be releasing its first SoCs for desktop
and tablet computers. Execution will be key for the company to succeed
in a highly competitive market that is dominated by ARM vendors and
challenged by Intel.
Rory Read, AMD’s new chief executive officer, did not waste any time
changing AMD’s roadmap and realign the company in quickly shifting
marketing environment for processor manufacturers. Its former GPU rival,
Nvidia is more and more emerging as a leading ARM processor supplier
for tablets, future subnotebooks and smartphones, Qualcomm and Freescale
are additional vendors that are pushing into the computing market from
the low-end and Intel is preparing its first serious attack to defend
its core business and extend its traditional markets into the tablet and
smartphone area in China in Q2 and in more areas of the world possibly
sometime in H2 of this year.
Read and his new executives, many of them with a history at IBM,
crafted the AMD’s new strategy with the bet that hardware will be
consumerized, converged and heavily rely on cloud services.
Consumerization, by the way, is a rather way to describe technologies
that are first established in consumer markets and then extended to
other areas, including enterprise segments. AMD’s client roadmap lists
plenty of new products with enough code-names to confuse analysts and
satisfy the thirst of enthusiasts. However, there are two products in
particular that are interesting – products that AMD will have to keep
pushing until their release.
For 2012, the company announced Hondo, a new Z-series APU that will
integrate one or two Bobcat CPUs and a basic GPU. It is listed in the
tablet and fanless category with a power consumption of 4.5 watts. This
product feels a bit like Intel’s Oak Trail (Atom Z670), which failed to
gain ground in the tablet game as it was a somewhat half-baked product
that did not entirely play by the rules of the tablet requirements game.
There is no information about Hondo, and we remain cautious about its
potential impact, especially if it becomes available as a choice next to
Intel’s potentially more attractive Medfield Atom Z2460 SoC.
Much more interesting is the 28 nm Tamesh, which will be, next to
Kabini, AMD’s first SoC. While Hondo will use the aging Bobcat CPU core,
Tamesh gets two new Jaguar cores as well as a new graphics engine.
Kabini will get up to four cores, but will be reserved for desktop
systems. No information has been provided about Jaguar yet, but AMD
confirmed that it will be used in ultra-low power APUs.
So far, it appears as if AMD will be competing in the tablet space
with x86 processors, which is a brave move as ARM owns the market at
this time and is the de-facto core used by Android products, which has
been accepted by the industry, including Intel, as the platform with the
most promising future next to Apple’s iOS. When available, Tamesh will
be a first-generation product that will have to prove its capabilities
next to Intel’s ambitious SoCs as well as next to Nvidia, Qualcomm,
Samsung and Freescale. It is too early to predict who will win this
fight, but we know that x86 is the underdog today and that AMD will find
itself in a battle against strong rivals that will not take prisoners.
Combining an ARM core with its GPU technology may have been not as
risky as taking the x86 route, , but this is the area of its core
expertise and may pay off in the end.
Wolfgang Gruener in Business Products
No comments:
Post a Comment