Carbon nanotube random access memory, phase-change memory and ferroelectric RAM are among the next-generation memory technologies that could replace conventional memory types in the future. However, many new memory types are expensive and "rely on exotic materials and switching mechanisms, making them difficult to fabricate and/or operate in the field," Mark Lapedus writes.
SK Telecom has chosen Xilinx field-programmable gate arrays to supply artificial intelligence acceleration for its automatic speech-recognition application. The use of Xilinx FPGAs has improved performance by up to five times while delivering 16 times better per-watt performance, according to SK Telecom.
Intel is extending its work on artificial intelligence with the acquisition of Vertex.AI, a startup focusing on the development of "deep learning for every platform." Intel said it will be putting to work Vertex's entire team and intellectual property "to further enable flexible deep learning at the edge."
AVSdsp has teamed with electrical engineering professors from National Taiwan Normal University to develop the Mipy chip, an artificial intelligence processor that is said to represent the smallest size and lowest cost on the AI market. The deep-learning image recognition AI chip's low cost makes it a good fit for applications such as home products, automotive products, toys and industrial equipment.
It's becoming more difficult to find defects in unpatterned wafers due to shrinking critical dimensions, writes Mark Lapedus. "Increasing requirements to detect and classify defects like pits and slip lines add challenges and complexity to the inspection process," said Felix Moellmann of Rudolph Technologies Systems.
The sheer number of technology conferences have made it difficult for organizations to differentiate their events, write Ed Sperling and Katherine Derbyshire. Within the semiconductor industry, new markets that cut through a number of segments make it harder to group events under one theme, they add.
The Center for Process Innovation and OXIS Energy are working on the Lithium Innovations for Future Electric vehicles project together, with CPI using OXIS coating technologies and running protective coating trials on lithium-metal cells.
Square has reduced the processing time of its chip readers and registers to just two seconds. Square implemented Visa's Quick Chip and Mastercard's M/Chip Fast technologies to reduce the speed from 3.6 seconds, which was achieved in 2017.
FD-SOI allows more versatility for developers who can decide to reduce the power in ways that aren't possible with finFETs, says STMicroelectronics Director of Technical Marketing Giorgio Cesana. Meanwhile, FD-SOI appears to boast more long-term reliability and functionality at multiple temperature ranges than standard complementary metal-oxide-semiconductors, said James Lamb of Brewer Science.
- Page 1