East was asked how ARM continues to remain focused on the design
needs of its partners now that it is expanding rapidly into new
sectors that require different expertise than its traditional domain
of mobile phones. ARM IP is in servers, myriad consumer products,
automotive and is now headed towards the Internet of Things.
"The real answer is it is hard. It's about execution, But it's what
we do," East said. Some things help such as the work with the Linaro
organization to create open-source software to sit on ARM cores for
mobile phones, which is now being replicated in the server space, he
said. "We're re-using the ecosystem for servers that we have used
for mobile phones," said East.
ARM also has an internal platform approach in which the company
effectively mocks up what it imagines to be a particular type of
system of the future and can do so for different markets, such as a
superphone or specialist server. ARM does this in consultation with
its semiconductor customers to help it make decisions as to how to
make performance-power trade offs for its forthcoming cores and
decisions on memory layout and connections.
"We have created an ARM platform made of test chips and so on. It's
not suitable for sale but it helps us understand memory
architecture, low power. It's only by building real things that you
can understand that stuff," said East.
It is understood that a particular mobile platform build is being
planned for 2013 around first silicon of the Cortex-A57 and
Shifting back to ARM's just-announced endorsement of the Weightless
standard, East was asked whether the proposal was broad enough to
cover all or most eventualities for machine-to-machine
"We get behind standards where they can reduce
market friction but we are not afraid to be agnostic. East was also
asked whether Weightless, originally drafted by Cambridge startup
Neul Ltd., and now backed by ARM, CSR plc, Neul and Cable &
Wireless is too much focused in the U.K.
"I did want Microsoft and Qualcomm to be part of the announcement,"
said East. He added that talks on forming special interest groups
can take time and in this case it was decided it was better to get
something going rather than wait.
East said that if it turns that another group is formed, on
different technology or geographic lines, then at least the IoT
universe would have been reduced from an enormous number of
proprietary solutions to a just two sets of standards, which would
still be a tremendous win for the industry.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.