Blog Review: June 19

Mentor’s Rebecca Lord digs into signal integrity complications and why today’s high frequency signals make it important to understand the physics of transmission lines.

Cadence’s Meera Collier points to the need to recognize diversity and nuance when compiling AI training datasets and avoid the oversimplification that can lead to bias.

Synopsys’ Deepak Nagaria checks out the new features introduced in LPDDR5 that enable reduced power consumption, secure data transfer, and improved bandwidth.

ANSYS’ Craig Hillman considers the impact of thermal degradation on electronics, how physics simulation can help predict degradation behavior, and the particular challenges faced by a range of components from magnets to capacitors to solder joints.

Arm’s Charlotte Christopherson shares a talk by EPFL’s Andreas Burg on why Gain Cell embedded DRAMs, or GC-eDRAM, could be a replacement for large, power-hungry embedded SRAM even at nanometer nodes.

SEMI’s Serena Brischetto chats with Gaetano D’Aquila, co-founder and CEO of GiPStech, about sensor fusion, augmented GPS applications, and the company’s indoor localization and navigation technology that doesn’t rely on an adequate GPS signal.

Rambus’ Andre Stoorvogel takes a look at the expansion of blockchain technologies beyond cryptocurrencies, from securities trading to the food industry.

Plus, don’t miss the blogs featured in last week’s Low Power-High Performance newsletter:

Editor In Chief Ed Sperling contends that a trade war couldn’t have happened at a worse time.

Synopsys’ Ron Lowman recommends looking beyond traditional design processes when making more effective and optimized AI SoCs.

Rambus’ Sanjay Charagulla observes that as AI processor performance grows, so does the importance of choosing a memory architecture.

Mentor’s Slava Zhuchenya explains why using filters to remove unused and unnecessary layout data can speed identification of high-resistance problem areas.

Adesto Technologies’ Paul Hill demonstrates how serial flash devices that support JESD252 can overcome challenges associated with smaller, lower-cost packages that don’t have a dedicated reset pin.

Arm’s Jason Andrews proposes an end-to-end workflow for deploying embedded machine learning.