Michael J. Miller, ForwardThinking
Mar 24, 2014
Each year after CES and Mobile World Congress, I ponder the shows' announcements and what they mean for the future of mobile application processors. We've certainly seen some interesting developments, including a set of 64-bit chip announcements, some of which are aimed more at mid-range phones, but new 32-bit chips seemed to be the most popular topic of conversation at the high end.
Almost every company that makes chips is talking about better graphics – with huge gains in performance – and all are talking about multiple cores, with 4- and even 8-core chips now becoming routine. What we haven't seen yet are any major applications processors built using 20nm technology (except those from Intel, which controls design and manufacturing for its chips), nor really new high-end 64-bit chips from most of the players. As a result, the changes we're likely to see in the chips for the highest-end phones over the next few months may not be huge, even as the mid-range and low-end phones catch up.
I'll discuss the details of the major chips later this week, but I'd like to start by talking about the basic building blocks that go into the creation of application processors. Unlike in the PC world, in general, makers of such processors tend to use at least some intellectual property (IP), either architectural licenses or full cores, in creating their products. Recall that a typical applications processor today includes a CPU, graphics core, often a baseband modem, and a raft of other features; and many makers license the CPU architecture, graphics, or potentially both. A typical processor maker will combine these features, both those they create themselves and those they license, to design a specific chip for a target market. In this post, I'll talk about CPU architecture, then follow tomorrow with one on graphics design.
Click here to read more ...