Next-Generation Computers Tech Digest - June 2017

Nanosheet transistor for 5nm technology

IBM, Samsung and Global Foundries and other Research Alliance partners have developed a process to create 5 nanometer silicon nanosheet transistors that would enable the creation of fingernail-sized chips containing 30 billion transistors. IBM says that this technology would increase battery life in mobile products by two to three times that of current standards. The increased transistor density could also help accelerate cognitive computing, and other data-intensive applications. The nanosheet transistor is composed of stacks of silicon nanosheets. 

Vibrating chip

University of Queensland, Australia, and Lockheed Martin, an aerospace technology company, have partnered to develop a next generation computer that uses mechanical vibrations inside computer chips to perform computations. This method of computation could make the computer chip more suitable for space applications than chips relying on electric current, the partners say, because it would be more resistant to radiation.  The scientists also see potential for applications in the development of nanotechnologies used in sensing, health and communications, while also improving future computers’ energy efficiency and heat management. 

Nano eraser

Scientists at the Southern University Illinois have developed a ‘nano eraser’ that can be used to correct mistakes when conducting nanolithography – a technique widely used in the production of computer processors. The conical eraser is made of agarose – a relative of cellulous – a spongy material that is capable of absorbing and releasing determined materials. The eraser can be used to clean a surface or to apply a material to a surface. The scientists see the eraser having uses in the fabrication of silicon processors used in computers allowing erasing and corrections of errors at the nanoscale. 

Superconducting nanoscale memory

Scientists at the University of Illinois have developed a nanoscale memory cell that could be used in superconducting processors. The device comprises two superconducting nanowires attached to two unevenly spaced electrodes written using electron-beam lithography. These form an asymmetric, closed superconducting loop called a nanowire ‘SQUID’ (superconducting quantum interference device). The direction of current through the wire dictates the binary state (either 1 or 0). The device can have a memory written by oscillating current at a particular magnitude. To read the memory the current is increased and the current value at which superconductivity fails enables the reading of 1 or 0. The device showed no occasions of memory loss when tested. Future investigations will test the superconductors at higher temperatures with a larger array of nanowire SQUIDS.  

ReRAM memory

4DS Memory, a computer memory company, claims it has developed ReRAM (resistive random-access memory) with speeds approaching that of DRAM that doesn’t require error correction. ReRAM is a type of non-volatile random-access memory that works by changing resistance across a dielectric solid state material. 4DS says it has developed a working Interface Switching ReRAM the same size as a typical FLASH memory chip – about 40nm. 4DS claims the chip has 97 percent memory endurance yield. 

US Department of Energy getting behind exascale computing

The US Department of Energy has awarded six research contracts worth a total USD258 million to companies it believes can accelerate US supercomputing technology. The awardees were AMD, Cray, IBM, NVIDIA, HPE and Intel. 

Microsoft’s AI on a chip

Microsoft has announced its Embedded Learning Library (ELL) that aims to help developers build and deploy machine-learning into embedded platforms such as Raspberry Pis, Arduinos and other microcontrollers. The machine learning model would carry out all processes locally without the need to be connected to the Internet – this should also reduce battery drain. Microsoft researchers demonstrated the machine learning system on a Raspberry Pi 3 trained to detect squirrels and connected to a sprinkler that sprays the animal when it is identified by the computer vision system. 

Investigation of topological excitations

US agency, DARPA (Defense Advanced Research Projects Agency), is carrying out the Topological Excitations in Electronics program that aims to show the uses of topological excitations in various applications including memory, logic, sensors, and quantum information processing. Current devices use superparamagnetism to code binary 1s and 0s. DARPA envisions – amongst other advances -  the possible uses of skyrmions for low-power memory and/or logic architectures due to their small size and potential for manipulation using ultra-low current densities. 

Add this: