Blog Review: Nov. 21

Cadence’s Paul McLellan looks at why specialized architectures will be the future of processor development, why general purpose processors are a poor match for AI, and other highlights from the recent Linley Processor Conference.

Mentor’s Harry Foster focuses on what’s happening in FPGA design and the factors that are adding to increasing design and verification complexity.

Synopsys’ Lewis Ardern helps online shoppers brush up on security concerns with a guide to the ways scammers try to get personal information as well as practical advice for identifying phishing and other attacks.

Arm’s Charlotte Christopherson points to a new research program focused on developing a wearable device consisting of an odor-detecting sensor, sensor interface, and machine learning processing engine manufactured on a flexible plastic substrate.

SEMI’s Paul Semenza warns that while increasing production of electric and hybrid autos means opportunities for MEMS and sensor makers, automotive suppliers focusing solely on MEMS could be left behind.

A Rambus writer notes that as memory has become an increasing bottleneck to performance and monolithic scaling has slowed, technologies like system-in-package and high-bandwidth memory can pick up the slack.

And don’t forget to check out the blogs highlighted in last week’s Manufacturing & Process Technology newsletter:

Editor In Chief Ed Sperling looks at why AI, and systems companies designing their own chips, could alter semiconductor manufacturing.

Executive Editor Mark LaPedus digs into why DRAM and NAND fell off the cliff in ’18, with no relief in sight for ’19.

Applied Materials’ Michael Stewart contends that chip startups see opportunities to change the AI ecosystem amid a dramatic increase in investment.

GlobalFoundries blogger Gary Dagastine finds that scaling still has a place, but existing technologies hold the power to push computing in new directions.

SEMI’s Ayo Kajopaiye digs into efforts to optimize processes based on analysis of production and sensor data.

Arm’s Kelvin Low considers how to deal with complex power issues at advanced nodes.