In less than a decade, that smartphone you're holding could have 32 times the memory, 20 times the bandwidth and a microprocessor core no bigger than a red blood cell, the CTO of chip design company ARM said on Thursday.
ARM has already helped develop a prototype, implantable device for monitoring eye-pressure in glaucoma patients that measures just 1 cubic millimeter, CTO Mike Muller said at ARM's TechCon conference in Silicon Valley Thursday. The device includes a microprocessor sandwiched between sensors at the top and a battery at the bottom.
Strip away those extra components, rearrange the transistors into a cube and apply the type of advanced manufacturing process expected in 2020, and you'd end up with a device that occupies about the same volume as a blood cell, Muller said.
ARM designs the processor cores used in most of today's smartphones and tablets, and smaller cores are generally more energy efficient, he said. That helps to extend battery life.
That's a good thing, because battery technology is advancing much more slowly, and Muller expects only twice the improvement in battery performance by the end of the decade.
That could be a gating factor for all the other improvements, so the electrical systems inside portable devices will have to be redesigned so that people don't have to recharge them multiple times a day.
For example, smartphones today contain basically a single compute system, with one type of CPU and some memory attached. But the tasks performed by smartphones, such as making a call or playing a 3D game, require very different levels of performance.
So in the future, MulIer said, "some systems will have entire subsystems within them, including their own CPU and their own memory," devoted to a particular task such as music playback. That way, other subsystems in a device can be shut down, conserving battery life.
It's a model ARM is already pursuing with its Big.Little architecture announced last week. That design will see two types of processor core in the same device, one powerful and one less so, and uses the most power-appropriate device for the task at hand. The idea of entire subsystems takes that a step further.
The bandwidth gains in 2020 will come mostly from advances in topology, according to Muller -- basically increasing the number of cellular base stations. Spectrum, and the technologies used to send bits across that spectrum, won't advance much, he predicted.
That's okay for people in cities, where it can make financial sense to install more base stations. "If you're out in the middle of nowhere, I'm sorry, there's not going to be much big change for you," Muller said.
Sign up for MIS Asia eNewsletters.