Next Generation Computing Tech Digest - April 2017

Canada’s new supercomputer

Simon Fraser University (SFU), in collaboration with Compute Canada and WestGrid, has launched its supercomputer, called Cedar. The computer, located in the SFU Data Centre on the Burnaby Campus, is the most powerful academic supercomputer in Canada, according to the SFU website. It has large memory nodes (up to 3 terabytes), on-node storage, and GPU nodes. It’s designed to optimally run multiple simultaneous jobs of up to 1024 CPU cores each. The launch is the first step toward consolidation of the 27 data centres across Canada into five to ten data centres by the end of 2018. 

Intel’s memory

Intel has launched its Optane Memory solid state drive (SSD) which promises ‘read speeds’ of up to 1350Mbytes/s and write speeds of 280Mbytes/s  on the 32GB drive. The initially available 16GB and 32GB models are meant to be used as support drives for a traditional hard drive. SSD technology achieves faster reading and writing speeds as it doesn’t rely on the spinning component in traditional hard disk drives. 

Boson sampling machine continues journey towards quantum computing

Chinese scientists at the Chinese Academy of Sciences, the University of Science and Technology of China, and Zhejiang University have developed a multi-photon quantum computing prototype (or a Boson sampling machine) to run the Boson sampling task. The sampling machine can only carry out a single predefined task; as such it isn’t universal. It does this by splitting a beam of photons arriving into the machine at a number of parallel input ports into a second set that leave via a set of parallel output ports. The boson sampling task involves working out the probability that a certain input configuration will lead to a certain output. The sampling rate of the Chinese team’s machine was 24,000 times faster than other similar devices at calculating the probability with five photons, according to the scientists. 

Chemical company to get supercomputer

BASF, a German chemistry company, has announced that it will collaborate with HPE (Hewlett Packard Enterprise) to create a 1 petaflop supercomputer for industrial chemical research at the chemical company’s HQ in Ludwigshafen. The computer will be used to promote the application and development of complex modelling and simulation approaches for BASF’s research. It is expected to reduce time to market and costs by, for example, accelerating the design of new materials with pre-defined properties. 

Using supercomputers to understand natural processes

Scientists have been using the Oak Ridge National Laboratory’s supercomputer to gain new insights into an unorthodox photosynthesis method widely used by succulents in the hope of understanding how to genetically engineer feed stock, food crops etc. to consume less water and possibly grow in desertified environments. Crassulacean acid metabolism (CAM) is a method of photosynthesis that helps plants conserve water by blocking stoma (small vents/orifices on the leaves of plants and responsible for gas exchange) from opening during the day (when the plant would lose water due to evaporation). Instead the stoma open at night to collect carbon dioxide. 

Processor

Researchers from the Graphene Flagship have developed a functioning microprocessor that is made of 115 integrated, layered transistors. The team used transition metal dichalcogenides (TMDs) (semi-conductors that can be separated into sheets a few atoms thick), and specifically the TMD molybdenum disulphide (MoS2) to create individual transistors and in turn the microprocessors that can perform 1-bit logic operations with possibility to scale the design to carry out multibit operations. The MoS2 transistors are flexible and compact, while also requiring little power compared to larger transistors - facts that the team believes lends them to use in wearables, sensors, e-paper, and integrated with logic circuits to make smart sensors. The team plans to create an 8 bit design on a single chip. 

Open source chips

SiFive is taking the open-source architecture of RISC-V and turning it into two silicon chip designs that can be licensed. The company’s 32-bit E31 Coreplex processor design is targeted for use in Internet of things devices, wearables and low power devices. The 64-bit CorePlex E51 could be used in servers, network processors, storage devices and edge computing. SiFive will charge a licensing fee but will not charge royalties unlike ARM. While this isn’t the only open source chip SiFive is positioning it as the first to be devised for use in everyday electronic items. SiFive also has plans to begin manufacturing chips by the end of 2017.

Samsung roll-mat computing

According to Patently Apple a Samsung patent has been published that shows a computer-like device that can be rolled into a cylinder. The device has a flexible display on top of which sits an infrared ray output device that would be used to project a virtual keyboard down onto a surface. 

 

Add this: