Techniques For Improving Energy Efficiency of Training/Inference for NLP Applications, Including Power Capping & Energy-Aware Scheduling


This new technical paper titled "Great Power, Great Responsibility: Recommendations for Reducing Energy for Training Language Models" is from researchers at MIT and Northeastern University. Abstract: "The energy requirements of current natural language processing models continue to grow at a rapid, unsustainable pace. Recent works highlighting this problem conclude there is an urgent need ... » read more

Technical Paper Round-Up: March 15


Research is expanding across a variety of semiconductor-related topics, from security to flexible substrates and chiplets. Unlike in the past, when work was confined to some of the largest universities, that research work is now being spread across a much broader spectrum of schools on a global basic, including joint research involving schools whose names rarely appeared together. Among the ... » read more

Mapping Transformation Enabled High-Performance and Low-Energy Memristor-Based DNNs


Abstract: "When deep neural network (DNN) is extensively utilized for edge AI (Artificial Intelligence), for example, the Internet of things (IoT) and autonomous vehicles, it makes CMOS (Complementary Metal Oxide Semiconductor)-based conventional computers suffer from overly large computing loads. Memristor-based devices are emerging as an option to conduct computing in memory for DNNs to make... » read more

PTAuth: Temporal Memory Safety via Robust Points-to Authentication


Authors: Reza Mirzazade Farkhani, Mansour Ahmadi, and Long Lu, Northeastern University Abstract: "Temporal memory corruptions are commonly exploited software vulnerabilities that can lead to powerful attacks. Despite significant progress made by decades of research on mitigation techniques, existing countermeasures fall short due to either limited coverage or overly high overhead. Further... » read more

FORMS: Fine-grained Polarized ReRAM-based In-situ Computation for Mixed-signal DNN Accelerator


Abstract: "Recent work demonstrated the promise of using resistive random access memory (ReRAM) as an emerging technology to perform inherently parallel analog domain in-situ matrix-vector multiplication—the intensive and key computation in deep neural networks (DNNs). One key problem is the weights that are signed values. However, in a ReRAM crossbar, weights are stored as conductance of... » read more

Manufacturing Bits: March 9


Finding cures for coronavirus The Department of Energy’s Oak Ridge National Laboratory (ORNL) is using the world’s most powerful supercomputer to identify drug compounds and cures for the coronavirus. [caption id="attachment_24162601" align="alignleft" width="300"] Summit supercomputer. Source: Oak Ridge National Laboratory[/caption] The supercomputer, called Summit, has identified 7... » read more

Non-Traditional Chips Gaining Steam


Flexible hybrid electronics are beginning to roll out in the form of medical devices, wearable electronics and even near-field communications tags in retail, setting the stage for a whole new wave of circuit design, manufacturing and packaging that reaches well beyond traditional chips. FHE devices begin with substrates made of ceramics, glass, plastic, polyimide, polymers, polysilicon, stai... » read more

The Week In Review: IoT


Finance NXP Semiconductors reported its Secure Connected Devices group posted revenue of $569 million in the fourth quarter, a gain of 10% from a year earlier. NXP CEO Richard Clemmer said in a statement, “All major product lines contributed to a seasonally solid quarter.” The chip company reported Q4 revenue of $2.44 billion and 2016 revenue of nearly $9.5 billion. Consortia Bosch, C... » read more

Manufacturing Bits: Dec. 29


Printing hair Using a low-cost, 3D printing technique, Carnegie Mellon University has found a way to produce hair-like strands and fibers. The printer produces plastic hair strand by strand. It takes about 20-25 minutes to generate hair on 10 square millimeters. A video can be seen here. [caption id="attachment_24544" align="alignleft" width="300"] 3D printed hair (Photo: Carnegie Mellon... » read more

System Bits: Aug. 25


Quantum computer building block In a finding that could ultimately be used to produce key components of quantum computers in the future, a team of researchers led by MIT have analyzed an exotic kind of magnetic behavior, driven by the mere proximity of two materials, using a technique called spin-polarized neutron reflectometry. This discovery could also be used to probe a variety of exotic... » read more

← Older posts Newer posts →