IBM Research plans to present two papers on improving the efficiency of artificial intelligence technology in visual analysis applications. The company will detail its BlockDrop software, which will be offered as open-source code, and its neuromorphic hardware for stereo vision.
The choice of a wireless connectivity standard for internet of things devices can be complicated and limiting, Brian Bailey writes. Richard Edgar of Imagination Technologies' Ensigma group says, "Wireless protocol is dictated by the infrastructure, but there are a number of layers to this question."
Chip design houses in China and Taiwan are contending to secure 8-inch wafer fabrication capacity at local foundries, as a short-term capacity shortfall threatens to become a long-term issue and wafer pricing will rise 20% to 30% during the third quarter, industry sources report. The phenomenon is driven by increasing demand for automotive ICs, internet of things chips, microcontrollers and other devices, while foundries are finding it difficult to obtain 8-inch wafer fabrication equipment, those sources note.
Macronix is planning a $1.5 billion expansion in production capacity during 2019 due to high demand for its memory devices, says Miin Wu, the chipmaker's chairman and CEO. "We believe this situation will continue for a few years, as we are to roll out new products that will consume more capacity," he says.
Rice University researchers report synthesizing laser-induced graphene by using a laser to heat polyimide films, then sculpting 3D blocks of graphene foam. The material could be used in batteries and supercapacitors, while also serving as a mold for conductive, flexible sensors, they say.
The PE29101 gallium nitride field-effect transistor driver for solid-state lidar systems was introduced by pSemi. Lidar technology is used in advanced driver-assistance systems and is expected to be a key element for fully autonomous vehicles.
Microchips designed with 10-nanometer and 7nm features are experiencing performance failures related to near-threshold approaches, due to process variation and timing issues, this analysis notes. "Near-threshold design brings new challenges because a lot of the statistical parameters become non-Gaussian," says Oliver King of Moortec Semiconductor.
Peter Schneider of Fraunhofer's Engineering of Adaptive Systems Division sees increasing complexity as the semiconductor industry's largest current challenge. "You have complexity in the system, in the structure, in the function, in the value chain, and in customer supplier relations," he says in this interview.