Blog Review: Sept. 23

Arm’s Matthew Mattina introduces a method to reduce the cost of neural network inference by combining both low-precision representation and the complexity-reducing Winograd transform while maintaining accuracy.

Cadence’s Paul McLellan checks out some of the biggest machine learning systems from Nvidia, Google, and Cerebras that were presented at the recent Hot Chips.

Mentor’s Robin Bornoff explains ECXML, the ‘Electronics Cooling eXtensible Markup Language’ defacto file format standard for thermal modeling that was recently published by JEDEC

Synopsys’ Godwin Maben considers the new low power design challenges posed by AI chips, including increasing leakage and the need for hardware and software teams to work together.

Rambus’ Frank Ferro and IDC’s Shane Rau discuss how AI enables useful data processing, various examples of AI silicon, and the evolving role of DRAM in advancing artificial intelligence.

Ansys’ Robert Harwood checks out what consumers around the world think about the trend towards electrification of vehicles and planes and how the U.S. differs from the rest of the world when it comes to thinking about carbon emissions.

In a video, VLSI Research’s Dan Hutcheson chats with Emmanuel Sabonnadière of CEA-Leti about how R&D needs have been accelerated by COVID-19 and some of the particular R&D efforts they are undertaking.

In a blog for SEMI, eda2asic’s Herb Reiter shares highlights from the recent Smart MedTech webinar series, from how sensors enable care at a distance to efforts to grow replacement tissue.

And don’t miss the blogs from the latest Manufacturing, Packaging & Materials newsletter:

Editor In Chief Ed Sperling warns that making all the components in a heterogeneous package work properly is just the beginning.

Executive Editor Mark LaPedus sees that while forecasts look up, challenges loom.

Custer Consulting’s Walt Custer offers an improved outlook as automotive sales rebound and consumer electronics build up for the busy pre-holiday season.

Coventor’s Pradeep Nanja demonstrates how to calibrate a process model to ensure it reflects actual process behavior and can generate silicon-accurate images.