The Kynix Blog - Memory

Last March, the AI program AlphaGo beat Korean Go champion LEE Se-Dol at the Asian board game. "The game was quite tight, but AlphaGo used 1200 CPUs and 56,000 watts per hour, while Lee used only 20 W. If a hardware that mimics the human brain structure is developed, we can operate artificial intelligence with less power," points out Professor YU Woo Jong.In collaboration with Sungkyunkwan University, researchers from the Center for Integrated Nanostructure Physics within the Institute for Basic Science (IBS), have devised a new memory device inspired by the neuron connections of the human brain.The research, published in Nature Communications, highlights the devise's highly reliable performance, long retention time and endurance. Moreover, its stretchability and flexibility makes it a promising tool for next-gen soft electronics attached to clothes or body.The brain is able to learn and memorise thanks to a huge number of connections between neurons. The information you memorise is transmitted through synapses from one neuron to the next as an electro-chemical signal.Inspired by these connections, IBS scientists constructed a memory called two-terminal tunnelling random access memory (TRAM), where two electrodes, referred to as drain and source, resemble the two communicating neurons of the synapse.While mainstream mobile electronics, like digital cameras and mobile phones use the so-called three-terminal flash memory, the advantage of two-terminal memories like TRAM is that two-terminal memories do not need a thick and rigid oxide layer."Flash memory is still more reliable and has better performance, but TRAM is more flexible and can be scalable," explains Professor Yu.TRAM is made up of a stack of one-atom-thick or a few atom-thick 2D crystal layers: One layer of the semiconductor molybdenum disulfide (MoS2) with two electrodes (drain and source), an insulating layer of hexagonal boron nitride (h-BN) and a graphene layer.In simple terms, memory is created (logical-0), read and erased (logical-1) by the flowing of charges through these layers. TRAM stores data by keeping electrons on its graphene layer. By applying different voltages between the electrodes, electrons flow from the drain to the graphene layer tunnelling through the insulating h-BN layer.The graphene layer becomes negatively charged and memory is written and stored and vice versa, when positive charges are introduced in the graphene layer, memory is erased.IBS scientists carefully selected the thickness of the insulating h-BN layer as they found that a thickness of 7.5 nm allows the electrons to tunnel from the drain electrode to the graphene layer without leakages and without losing flexibility.Flexibility and stretchability are indeed two key features of TRAM. When TRAM was fabricated on flexible plastic (PET) and stretachable silicone materials (PDMS), it could be strained up to 0.5% and 20%, respectively.In the future, TRAM can be useful to save data from flexible or wearable smartphones, eye cameras, smart surgical gloves, and body-attachable biomedical devices.Last but not least, TRAM has better performance than other types of two-terminal memories known as phase-change random-access memory (PRAM) and resistive random-access memory (RRAM).Reference:MT16JTF51264AZ-1G6M1MD2202-D192MD2203-D576

B&R has selected the Hyperstone S8 Flash Memory Controller for their recently announced X90 mobile automation product line. The X90 family of products are designed for securing mobile control and I/O tasks in demanding environments experienced by industrial and mobile machines.To meet all requirements of their robust applications B&R implemented the Hyperstone S8 Flash Memory Controller, which operates over an extended temperature range and ensures the highest reliability and endurance even under extreme environmental conditions. The S8 Controller enables the use of advanced technology NAND flashes in demanding, rugged applications e.g. in commercial vehicles and outdoor applications. B&R X90 products deliver robustness and durability hence the choice of Hyperstone products.X90 mobile With a comprehensive set of standardised components, the X90 product line offers automation solutions perfect for flexible automation concepts used in mobile and outdoor applications found in agriculture and forestry, construction or municipal vehicles. Extended temperature range, strong vibrations or shocks and influences such as salt, oil and UV light must be handled by the system. Basic features of the X90 family include interfaces for CAN, USB, Ethernet and the real-time POWERLINK bus system. A flexible set up allows the addition of I/O channels and interfaces. The X90 series was introduced at Bauma 2016 in Munich.S8 The Hyperstone S8 Flash Memory Controller and its application and Flash specific firmware offer a platform for industrial high endurance Flash Memory cards or modules which are compatible with SD interface host systems. Together with Hyperstone’s proprietary hyReliability firmware, S8 provides enhanced endurance and data retention management, as well as rigorous fail-safe features, all of which are mandatory for industrial applications. Implemented hardware features and special firmware developments enable the highest data retention and refresh mechanisms, even when storage applications are faced with extreme environmental conditions.Reference:MT16JTF51264AZ-1G6M1SDUS5EB-001GMD2202-D192MT9VDDT6472HY-335 F2 

Astute Electronics has strengthened its new memory division portfolio by signing a UK and Ireland deal for Innodisk’s industrial embedded flash and DRAM storage products and technologies. Innodisk focuses on industries such as aerospace, defence, industrial and cloud storage to name a few, and where a commitment to quality service is vitally important. The company manufactures all its products in a purpose-built memory production facility located in Taiwan.Mick Martin, Memory Division Manager at Astute says; “Innodisk’s product range matches our customer base exactly. Its product design features fixed-BOM control to ensure product quality, longevity and stability, and the use of industrial-grade components means that products can withstand temperatures ranging from -40 to 85°C. Sample parts are available very quickly and typical production lead times are only two weeks. The company is also able to customise devices to suit specific applications.”With over 62 patents, Innodisk is a technical innovator, pioneering Pin 7 VCC technology in the SATA interface which eliminates the need for power cables.This led to the development of the SATADOM form factor which is certified by Intel and widely adopted by industrial and embedded system designers worldwide.Innodisk offers a wide selection of flash memory form factors. Flash products are available in single-level-cell (SLC) and multi-layer cell (MLC) styles and a comprehensive range of DRAM modules are also included.Specifically, InnoRobust devices have a hi-rel feature set specially designed for aerospace and defence applications, while ISO/TS 16949 approval guarantees that that the company can deliver reliable storage and memory solutions for in-vehicle computing system applications.Reference:MT16JTF51264AZ-1G6M1SDUS5EB-001GMD2202-D192MD2202-D192-X 

New research, led by the University of Southampton, has demonstrated that a nanoscale device, called a memristor, could be used to power artificial systems that can mimic the human brain.Artificial neural networks (ANNs) exhibit learning abilities and can perform tasks which are difficult for conventional computing systems, such as pattern recognition, on-line learning and classification. Practical ANN implementations are currently hampered by the lack of efficient hardware synapses; a key component that every ANN requires in large numbers.In the study, published in Nature Communications, the Southampton research team experimentally demonstrated an ANN that used memristor synapses supporting sophisticated learning rules in order to carry out reversible learning of noisy input data.Memristors are electrical components that limit or regulate the flow of electrical current in a circuit and can remember the amount of charge that was flowing through it and retain the data, even when the power is turned off.Lead author Dr Alex Serb, from Electronics and Computer Science at the University of Southampton, said: "If we want to build artificial systems that can mimic the brain in function and power we need to use hundreds of billions, perhaps even trillions of artificial synapses, many of which must be able to implement learning rules of varying degrees of complexity. Whilst currently available electronic components can certainly be pieced together to create such synapses, the required power and area efficiency benchmarks will be extremely difficult to meet -if even possible at all- without designing new and bespoke 'synapse components'."Memristors offer a possible route towards that end by supporting many fundamental features of learning synapses (memory storage, on-line learning, computationally powerful learning rule implementation, two-terminal structure) in extremely compact volumes and at exceptionally low energy costs. If artificial brains are ever going to become reality, therefore, memristive synapses have to succeed."Acting like synapses in the brain, the metal-oxide memristor array was capable of learning and re-learning input patterns in an unsupervised manner within a probabilistic winner-take-all (WTA) network. This is extremely useful for enabling low-power embedded processors (needed for the Internet of Things) that can process in real-time big data without any prior knowledge of the data.Co-author Dr Themis Prodromakis, Reader in Nanoelectronics and EPSRC Fellow in Electronics and Computer Science at the University of Southampton, said: "The uptake of any new technology is typically hampered by the lack of practical demonstrators that showcase the technology's benefits in practical applications. Our work establishes such a technological paradigm shift, proving that nanoscale memristors can indeed be used to formulate in-silico neural circuits for processing big-data in real-time; a key challenge of modern society."We have shown that such hardware platforms can independently adapt to its environment without any human intervention and are very resilient in processing even noisy data in real-time reliably. This new type of hardware could find a diverse range of applications in pervasive sensing technologies to fuel real-time monitoring in harsh or inaccessible environments; a highly desirable capability for enabling the Internet of Things vision."Reference:KY259-SDUS5EB-002GKY259-SDUS5AB-002GKY259-SDUS5AB-001G 

Toshiba Corporation today announced that it has developed the world's first 15-nanometer (nm) process technology, which will apply to 2-bit-per-cell 128-gigabit (16 gigabytes) NAND flash memories. Mass production with the new technology will start at the end of April at Fab 5 Yokkaichi Operations, Toshiba's NAND flash fabrication facility (fab), replacing second generation 19 nm process technology, Toshiba's previous flagship process. The second stage of Fab 5 is currently under construction, and the new technology will also be deployed there.Toshiba has achieved the world's smallest class chip size with the 15nm process plus improved peripheral circuitry technology. The new chips achieve the same write speed as chips formed with second generation 19 nm process technology, but boost the data transfer rate to 533 megabits a second, 1.3 times faster, by employing a high speed interface.Toshiba is now applying the 15nm process technology 3-bit-per-cell chips, and aims to start mass production in the first quarter of this fiscal year, to June 2014. The company will develop controllers for embedded NAND flash memory in parallel and introduce 3-bit-per-cell products for smartphones and tablets, and will subsequently extend application to notebook PCs by developing a controller compliant with solid state drives (SSD). 

Random-access memory, or RAM, is where computers like to store the data they're working on. A processor can retrieve data from RAM tens of thousands of times more rapidly than it can from the computer's disk drive.But in the age of big data, data sets are often much too large to fit in a single computer's RAM. The data describing a single human genome would take up the RAM of somewhere between 40 and 100 typical computers.Flash memory—the type of memory used by most portable devices—could provide an alternative to conventional RAM for big-data applications. It's about a tenth as expensive, and it consumes about a tenth as much power.The problem is that it's also a tenth as fast. But at the International Symposium on Computer Architecture in June, MIT researchers presented a new system that, for several common big-data applications, should make servers using flash memory as efficient as those using conventional RAM, while preserving their power and cost savings.The researchers also presented experimental evidence showing that, if the servers executing a distributed computation have to go to disk for data even 5 percent of the time, their performance falls to a level that's comparable with flash, anyway.In other words, even without the researchers' new techniques for accelerating data retrieval from flash memory, 40 servers with 10 terabytes' worth of RAM couldn't handle a 10.5-terabyte computation any better than 20 servers with 20 terabytes' worth of flash memory, which would consume only a fraction as much power."This is not a replacement for DRAM [dynamic RAM] or anything like that," says Arvind, the Johnson Professor of Computer Science and Engineering at MIT, whose group performed the new work. "But there may be many applications that can take advantage of this new style of architecture. Which companies recognize: Everybody's experimenting with different aspects of flash. We're just trying to establish another point in the design space."Joining Arvind on the new paper are Sang Woo Jun and Ming Liu, MIT graduate students in computer science and engineering and joint first authors; their fellow grad student Shuotao Xu; Sungjin Lee, a postdoc in Arvind's group; Myron King and Jamey Hicks, who did their PhDs with Arvind and were researchers at Quanta Computer when the new system was developed; and one of their colleagues from Quanta, John Ankcorn—who is also an MIT alumnus.Outsourced computationThe researchers were able to make a network of flash-based servers competitive with a network of RAM-based servers by moving a little computational power off of the servers and onto the chips that control the USB flash drives. By preprocessing some of the data on the flash drives before passing it back to the servers, those chips can make distributed computation much more efficient. And since the preprocessing algorithms are wired into the chips, they dispense with the computational overhead associated with running an operating system, maintaining a file system, and the like.With hardware contributed by some of their sponsors—Quanta, Samsung, and Xilinx—the researchers built a prototype network of 20 servers. Each server was connected to a field-programmable gate array, or FPGA, a kind of chip that can be reprogrammed to mimic different types of electrical circuits. Each FPGA, in turn, was connected to two half-terabyte—or 500-gigabyte—flash chips and to the two FPGAs nearest it in the server rack.Because the FPGAs were connected to each other, they created a very fast network that allowed any server to retrieve data from any flash drive. They also controlled the flash drives, which is no simple task: The controllers that come with modern commercial flash drives have as many as eight different processors and a gigabyte of working memory.Finally, the FPGAs also executed the algorithms that preprocessed the data stored on the flash drives. The researchers tested three such algorithms, geared to three popular big-data applications. One is image search, or trying to find matches for a sample image in a huge database. Another is an implementation of Google's PageRank algorithm, which assesses the importance of different Web pages that meet the same search criteria. And the third is an application called Memcached, which big, database-driven websites use to store frequently accessed information.Chameleon clustersFPGAs are about one-tenth as fast as purpose-built chips with hardwired circuits, but they're much faster than central processing units using software to perform the same computations. Ordinarily, either they're used to prototype new designs, or they're used in niche products whose sales volumes are too small to warrant the high cost of manufacturing purpose-built chips.But the MIT and Quanta researchers' design suggests a new use for FPGAs: A host of applications could benefit from accelerators like the three the researchers designed. And since FPGAs are reprogrammable, they could be loaded with different accelerators, depending on the application. That could lead to distributed processing systems that lose little versatility while providing major savings in energy and cost."Many big-data applications require real-time or fast responses," says Jihong Kim, a professor of computer science and engineering at Seoul National University. "For such applications, BlueDBM"—the MIT and Quanta researchers' system—"is an appealing solution."