Category Archives: Device Architecture

Ever-growing data generation driven by mobile devices, the cloud, the IoT , and big data, as well as novel AI applications, all part of the megatrends, requires continuous advancements in memory technologies. Emerging NVM takes benefit of this dynamic ecosystem.

After more than 15 years in development, PCM, one of the emerging NVM technologies, has finally taken off thanks to the strong involvement of two leading companies, Micron and Intel, announces Yole Développement (Yole). The growth mainly arises from stand-alone applications. “Although momentum is building around emerging NVM for embedded applications, stand-alone memories will be the dominant market, which will be mainly driven by SCM enterprise and client applications,” comments Simone Bertolazzi, PhD, Technology & Market Analyst at Yole.

The market research and strategy consulting company Yole proposes today a technology & market survey dedicated to the emerging non-volatile memory technologies and markets, Emerging Non-Volatile Memory.

Yole and its partners System Plus Consulting and Knowmade, deeply investigate the memory business. The Group set up this year valuable memory services and reports to deliver world class research, data and insight. The emerging NVM report is part of them.

“With our memory activities including a dedicated webcasts program covering DRAM & NAND and emerging NVM, Yole Group of Companies provides valuable expertise and knowledge to its clients and allow them to understand the evolution of this competitive industry,” asserts Emilie Jolivet, Director, Semiconductor & Software from Yole.

The emerging NVM report is a comprehensive analysis of the semiconductor memory ecosystem with the following technologies (STT-) MRAM, RRAM and PCM, plus an introduction to standard memory, flash NAND, DRAM, NVDIMMs. It provides a deep understanding of the NVM applications and details the related market forecasts until 2023. NVM technologies are well described with the companies involved. In this new report, Yole’s Semiconductor & Software team highlights the competitive landscape with supply chain, market positioning and market shares analysis.
What is the status of the emerging NVM business? Yole Group of Companies invite you to enter in the memory world.

Since its latest edition, Yole’s analysts point out today market evolution and technical innovations. According to Yann de Charentenay, Senior Technology & Market Analyst at Yole, DRAM scaling will continue in the next five years, though at slower pace. NAND density will keep increasing thanks to continuous advancements in 3D integration approaches. And emerging NVM will not replace NAND and DRAM but they will rather complement them in “combined” memory solutions. In addition, SCM will be the main emerging NVM market and will be dominated by 3DXPoint for the next 5 years.
From a technology point of view, (STT-) MRAM is gaining momentum for embedded MCU applications since all big foundries are getting involved in this area. Stand-alone RRAM will try to catch market share to PCM on SCM applications. And emerging NVM sales will grow by more than one order of magnitude in the next three years, thanks to SCM applications.

In parallel, Yole’s team identified an increased foundry involvement in (STT-) MRAM and RRAM market segment. Key players such as GlobalFoundries, TSMC, UMC, SMIC and Samsung Foundry Services develop a strong expertise with related capabilities to offer attractive services. This trend is showing a growing foundries’ interest in memory business. As an example, the leading semiconductor company, TSMC announced possible acquisition of a memory company. Moreover, analysts point out the growing number of players including Chinese companies.

In the stand-alone business, emerging NVMs will not replace DRAM and NAND but will be used in combination with them inside memory modules, e.g. SSDs, DIMMs, and NVDIMMs. In 2023, PCM will maintain its lead in the stand-alone memory market thanks to the increasing adoption of 3D XPoint as an enterprise and client SCM. It is worth noting that Samsung and Toshiba took a different strategic path by developing 3D NAND-based SCM solutions such as Z-NAND (Samsung) and XL-Flash (Toshiba, showcased in August 2018). However, these technologies will be used in enterprise SSDs and will not compete with DDR4-compatible Optane DIMMs, which we expect will represent more than 50% of overall 3D XPoint sales.

RRAM was expected to be the first stand-alone technology to compete with 3D XPoint, but it has suffered repeated delays due to technical challenges. We presume that RRAM could return in the race for SCM after 2020, and possibly start competing with NAND for mass storage applications. STT-MRAM, thanks to its high speed and high endurance, is promising for enterprise storage SCM. However, its success will be much lower compared to stand-alone PCM due to higher costs, greater fabrication complexity, and challenging scalability.

Compared to stand alone, the embedded emerging NVM market is relatively small, representing ~3% of the emerging NVM market in 2017. The market is dominated today by RRAM, since only a few RRAM based MCUs are available on the market. However, all top foundries are now getting ready with 28/22nm technology processes for STTMRAM whereas RRAM adoption has been delayed by approximately two years by SMIC and UMC.

Therefore, we expect that STT-MRAM will be the first to take-off in the coming years and will lead the embedded emerging NVM market, especially MCUs, which represent the most important embedded segment. Emerging memory will first replace eFlash, which is facing major scaling challenges due to rising fabrication complexity/costs for technology nodes ≤ 28nm. The adoption of STT-MRAM as an embedded cache memory (SRAM or eDRAM) in high-end processors and mobile application processors (AP) will occur later due to more strict scalability requirements (≤ 14nm).

AI on the edge is the most innovative application for embedded emerging NVM. Crossbar recently demonstrated various AI applications, i.e. face recognition, through the use of RRAM chips. We expect that such RRAM-based AI devices will enter the market after 2021.

Yole Group of Companies leverage decades of industry experience while partnering with its clients to make sure they are consistently well-informed on this dynamic memory market. These years were indeed impressive, not only in terms of revenues, but also in pricing and capital expenditure. Mike Howard, VP of DRAM & Memory Research and Walt Coon, VP of NAND & Memory Research at Yole describe in a dedicated interview published last week, the memory ecosystem and its players, highlighting the latest technology advancements and the future evolutions of the market: click Memory business: what’s next?.

Synopsys, Inc. (Nasdaq: SNPS) announced today another milestone in its longstanding partnership with imec, a research and innovation hub in nanoelectronics and digital technologies, with the successful completion of the first comprehensive sub-3 nanometer (nm) parasitic variation modeling and delay sensitivity study of complementary FET (CFET) architectures. With the potential to significantly reduce area versus traditional FinFETs, CFET is a promising option to maintain area scaling beyond 3nm technology.

In 3-nm and 2-nm process technologies, the magnitude of variation increases significantly for middle of line (MOL) parameters, as well as interconnect, due to high resistance of metal lines, vias, and surface scattering. Therefore, modeling parasitic variation and sensitivity is a critical factor in bringing CFET to mainstream production.

Prediction at early stages of process development will allow foundries to create more robust and variation-tolerant transistors, standard cells, and methodologies for metal interconnect. Using the QuickCap® NX 3D field solver, in a close collaboration between Synopsys R&D and imec research teams, allowed for fast and accurate modeling of parasitics for a variety of device architectures and to identify the most critical device dimensions and properties. This allowed the optimization of CFET devices for better power/performance trade-offs. As part of a comprehensive set of tools that includes Raphael™ TCAD extraction to StarRC™ parasitic extraction for the largest system-on-chips (SoCs), QuickCap NX effectively helps process engineers understand the sensitivity of circuit performance to variations in process parameters and improves modeling accuracy by establishing golden reference values.

“This work has allowed us to accurately model and analyze cell and interconnect variation at advanced processes and architectures, such as Complementary FET,” said Anda Mocuta, director, Technology Solutions and Enablement at imec. “Our collaboration with Synopsys continues a legacy of successful collaborations that enable us to search for technological breakthroughs below 3 nanometers. The capabilities of Synopsys tools, such as QuickCap NX, have been key to our joint research on variability.”

“Imec is at the forefront of research into semiconductor technology. Our collaboration with imec to develop variation-aware solutions down to 2 nanometer processes will benefit the entire semiconductor industry,” said Antun Domic, chief technology officer at Synopsys. “Utilizing the flexibility of Synopsys’ QuickCap NX 3D parasitic extraction interface, engineers can better target and significantly reduce the number of trials needed to optimize circuit performance in the presence of process variation and reduce circuit sensitivity. This significantly reduces the overall turnaround time for device and circuit optimization.”

The Global Semiconductor Alliance (GSA) is proud to announce the award recipients honored at the 2018 GSA Awards Dinner Celebration that took place last evening in Santa Clara, California. For almost a quarter century, the GSA Awards have recognized the achievements of top performing semiconductor companies in several categories ranging from outstanding leadership to financial accomplishments, as well as overall respect within the industry.

Individual Awards:

Dr. Morris Chang Exemplary Leadership Award
The GSA’s most prestigious award recognizes individuals, such as its namesake, Dr. Morris Chang, for their exceptional contributions to drive the development, innovation, growth and long-term opportunities for the semiconductor industry. This year’s recipient is Dr. Lisa Su, President and CEO of Advanced Micro Devices (AMD).

Rising Women of Influence Award
This newly initiated award recognizes and profiles the next generation of women leaders in the semiconductor industry that are believed to be rising to top executive roles within their organizations. This year’s award was presented to Vanitha Kumar, Vice President of Software Engineering at Qualcomm Technologies, Inc.

Company Awards:

Most Respected Public Semiconductor Companies
GSA members identified the winners in this category by casting ballots for the industry’s most respected companies, judged for their vision, technology and market leadership. Below are this year’s recipients:

Most Respected Public Semiconductor Company Achieving Greater than $5 Billion in Annual Sales:

NVIDIA Corporation

Most Respected Public Semiconductor Company Achieving $1 Billion to $5 Billion in Annual Sales:

Marvell Semiconductor

Most Respected Public Semiconductor Company Achieving $500 Million to $1 Billion in Annual Sales:

Silicon Labs

Most Respected Emerging Public Semiconductor Company Achieving $100 Million to $500 Million in Annual Sales:

Nordic Semiconductor

Most Respected Private Company:

SiFive Inc.

Best Financially Managed Semiconductor Companies
T

hese awards are derived from a broad evaluation of the financial health and performance of public fabless and IDM semiconductor companies. Below are this year’s recipients:

Best Financially Managed Company Achieving up to $1 Billion in Annual Sales:

Holtek Semiconductor Inc.

Best Financially Managed Semiconductor Company Achieving Greater than $1 Billion in Annual Sales:

Micron Technology, Inc.

Start-Up to Watch
GSA’s Private Awards Committee, comprised of successful executives, entrepreneurs and venture capitalists, chose the winner by identifying a promising startup that has demonstrated the potential to positively change its market or the industry through innovation and market application. This year’s winner is Movandi.

As a global organization, the GSA recognizes outstanding companies headquartered in the Europe/Middle East/Africa and Asia-Pacific regions having a global impact and demonstrating a strong vision, portfolio and market leadership. Two awards were presented in this category:

Outstanding Asia-Pacific Semiconductor Company

Samsung Electronics Co., Ltd.

Outstanding EMEA Semiconductor Company

Infineon Technologies AG

Analyst Favorite Semiconductor Company
Two analyst pick awards were presented based on technology and financial performance as well as future projections:

NVIDIA Corporation was chosen by Rajvindra Gill, Managing Director at Needham & Company, LLC

Advanced Micro Devices (AMD) was chosen by Mark Lipacis, Managing Director at Jefferies, LLC
This year’s ceremony was attended by close to 1500 global executives in the semiconductor and technology industries.

There is often a pronounced symmetry when you look at the lattice of crystals: it doesn’t matter where you look – the atoms are uniformly arranged in every direction. This behavior was also to be expected by a crystal, which physicists at the Helmholtz-Zentrum Dresden-Rossendorf (HZDR), the University of Warsaw and the Polish Academy of Sciences produced, using a special process: a compound from an indium arsenide semiconductor, spiked with some iron. The material, however, did not adhere to perfect symmetry. The iron formed two-dimensional, lamellar-shaped structures in the crystal that lent the material a striking property: it became magnetic. In the long term, the result could be vital in understanding superconductors.

By using lasers, scientists from Germany and Poland were able to create a remarkable compound of indium arsenide and iron. Surprisingly, the compound — the black stripes in this image — formed lamellar-shaped structures in the surface of the crystal along one crystalline axis. Credit: HZDR / S. Zhou

“Using the possibilities of our Ion Beam Center, we fired fast iron ions at a crystal made of indium arsenide, a semiconductor made of indium and arsenic,” says Dr. Shengqiang Zhou, physicist at the HZDR Institute of Ion Beam Physics and Materials Research. “The iron penetrated approximately one hundred nanometers deep into the crystal surface.” The iron ions remained in the minority – they constituted only a few percent in the surface. The researchers then fired light pulses at the crystal using a laser. The flashes were ultra-short so that only the surface melted. “For much less than a microsecond, the top one hundred nanometers were a hot soup, whereas the crystal underneath remained cold and well ordered,” Zhou says, describing the result.

The crystal surface cooled again just a blink of an eye after the laser bombardment. Something unusual had happened: the surface had essentially reverted back to the indium arsenide lattice structure. The cooling, however, was so rapid that the iron atoms did not have sufficient time to find and occupy a regular lattice state in the crystal. Instead, the metal atoms joined forces with their peers to form remarkable structures – small two-dimensional lamellae, arranged in parallel.

“It came as a surprise that the iron atoms were arranged in this manner,” says Zhou. “We were thus able to create such a lamellar structure for the first time globally.” When the experts examined the newly created material more closely, they determined that it had become magnetic due to the influence of iron. The researchers from Poland and Germany also managed to theoretically describe the process and simulate it on the computer. “The iron atoms arranged themselves into a lamellar structure because this was energetically the most favorable state they could take in the brief period of time,” says Prof. Tomasz Dietl from the International Research Center MagTop at the Polish Academy of Sciences, summarizing the result of the calculations.

The result could be relevant in, for example, understanding superconductors – a class of materials that can conduct electricity entirely without loss. “Lamellae-like structures can also be found in many superconducting materials,” explains Zhou. “Our compound could therefore serve as a model system and help in better understanding superconductor behavior.” This could perhaps also serve to optimize their properties: in order for superconductors to work, they must currently be cooled to comparatively low temperatures of, for example, minus two hundred degrees Celsius. The aim of many experts is to increase these temperatures gradually – until they find a dream material, which loses its electrical resistance even at normal ambient temperatures.

The Semiconductor Industry Association (SIA), representing U.S. leadership in semiconductor manufacturing, design, and research, today announced worldwide sales of semiconductors reached $41.8 billion for the month of October 2018, an increase of 12.7 percent from the October 2017 total of $37.1 billion and 1.0 percent more than last month’s total of $41.4 billion. Monthly sales numbers are compiled by the World Semiconductor Trade Statistics (WSTS) organization and represent a three-month moving average. Additionally, a newly released WSTS industry forecast was revised upward and now projects annual global market growth of 15.9 percent in 2018 and 2.6 percent in 2019.

“The global semiconductor industry posted solid year-to-year growth in October and is on pace for its highest-ever annual sales in 2018, but growth has moderated in recent months,” said John Neuffer, president and CEO, Semiconductor Industry Association. “Although strong sales of DRAM products continue to boost overall market growth, sales in all other major product categories also increased year-to-year in October, and all major regional markets posted year-to-year gains. Double-digit annual growth is expected in 2018, with more modest growth projected for 2019.”

Regionally, year-to-year sales increased in China (23.3 percent), the Americas (14.1 percent), Europe(7.0 percent), Japan (5.5 percent), and Asia Pacific/All Other (3.7 percent). Compared with last month, sales were up in the Americas (2.8 percent), Asia Pacific/All Other (1.8 percent), Japan (0.4 percent), and Europe (0.2 percent), but down slightly in China (-0.4 percent).

Additionally, SIA today endorsed the WSTS Autumn 2018 global semiconductor sales forecast, which projects the industry’s worldwide sales will be $477.9 billion in 2018. This would mark the industry’s highest-ever annual sales, a 15.9 percent increase from the 2017 sales total of $412.2 billion. WSTS projects year-to-year increases across all regional markets for 2018: the Americas (19.6 percent), Asia Pacific (16.0 percent), Europe (13.2 percent), and Japan (9.6 percent). In 2019, growth in the semiconductor market is expected to moderate, with annual sales projected to increase by 2.6 percent. WSTS tabulates its semi-annual industry forecast by convening an extensive group of global semiconductor companies that provide accurate and timely indicators of semiconductor trends.

The pattern of arrangement of atoms in a crystal, called the crystal lattice, can have a huge effect on the properties of solid materials. Controlling and harnessing these properties is a challenge that promises rewards in applications such as novel sensors and new solid-state devices. An international research collaboration, including researchers from Osaka University, has reported the induction of an interesting type of magnetic order, called helimagnetism, in a cobalt oxide material by expanding its lattice structure. Their findings were published in Physical Review Materials.

This is a schematic illustration of the helimagnetic-ferromagnetic transition driven by the lattice expansion/compression in the cubic perovskite Sr1-xBaxCoO3. Credit: S. Ishiwata and H. Sakai

Magnetic behavior results from the order of the magnetic moments of the many individual atoms in a material. In helimagnetism, instead of the magnetic moments being aligned–as they are in permanent magnets, producing ferromagnetism–the moments arrange themselves in a helical pattern. This behavior is generally only observed in complicated lattice structures where different types of magnetic interactions compete with each other, therefore the report of induced helimagnetism in a simple cubic cobalt oxide structure, is highly significant.

“We have shown emergent helical spin order in a cubic perovskite-type material, which we achieved simply by expanding the lattice size,” study first author Hideaki Sakai says. “We were able to control the size of the lattice expansion by using a high-pressure technique to grow a series of single crystals with particular chemical compositions. Changing the amount of different ions in our materials provided us with sufficient control to investigate the magnetic properties.”

Systematically replacing strontium ions in the structure with larger barium ions caused the lattice to continually expand until the regular ferromagnetic magnetic order present at room temperature was disrupted, resulting in helimagnetism. These experimental findings were successfully supported by calculations.

“The fact that we were able to largely reproduce our findings by first principles calculations verifies that the magnetic interactions in the materials are highly sensitive to the lattice constant,” Sakai says. “The more we can understand about the magnetic behavior of crystalline materials, the closer we move towards translating their properties into useful functions. We hope that our findings will pave the way for novel sensor applications.

The control of magnetic order simply by changing the lattice chemistry, as demonstrated by this research, provides a foundation for investigating the properties of many other crystalline materials.

SEMI announced today that the Industry Strategy Symposium (ISS) 2019 will take place January 6-9 at Half Moon Bay’s Ritz-Carlton Hotel with the theme “Golden Age of the Semiconductor: Enabling the Next Industrial Revolution.” ISS is the year’s first executive check-in, bringing together leading analysts, researchers, economists, and technologists for insights on the forces impacting the semiconductor industry. The annual symposium offers executives a unique platform for identifying growth opportunities and gaining industry intelligence to help them ensure that their business plans and forecasts are based on up-to-the-minute market conditions. Registration for ISS 2019 is now open.

Major developments are transforming the extended supply chain — artificial intelligence (AI), intelligent vehicles, augmented and virtual reality, and seemingly limitless connectivity within the cloud. Collaboration across an expanding ecosystem and advanced technical innovations are giving rise to advanced electronics that continue to raise performance and power-consumption requirements and drive heterogenous integration.

“It can be hard to lead to where you’ve not been, particularly at this most dynamic time in business,” said David Anderson, president of SEMI Americas. “Executives from across the supply chain historically have asked for this timely ISS program in order to gauge what the year ahead looks like. To help management get the very latest picture, and to compare notes to most confidently tune their operations and sales, ISS is a tremendous level set for the new year.”

ISS 2019 is the first of seven conferences comprising SEMI’s inaugural Technology Leadership Series of the Americas. Aligned from coast to coast over one year, the series is designed to foster critical discussions on the short- and long-term influences and opportunities to the $2 trillion emerging markets.

ISS 2019 highlights include:

Keynotes

  • Dr. Jo de Boeck, chief strategy officer at IMEC, will share how nanotechnology’s “magic” will enable advanced applications for SMART mobility, SMART cities, infotainment and healthcare
  • Dr. Ann Kelleher, senior vice president for the technology and manufacturing group at Intel
  • Jim Talent, former U.S. Senator, will discuss the evolving U.S.-China strategic relationship

Economic trends and market perspectives affecting the chip industry

  • Executives and economists from Harvard Kennedy School, Gartner, Hilltop Economics, Linx, Amazon, Bank of America, Rockwell Automation, Ericsson, Selexis, Enthought and VLSI Research

Technology, manufacturing and industrial revolution discussions

  • Executives and CTOs from Google, TEL, Micron, Intel Labs, Applied Materials, Xperi, McKinsey, DECA Technologies, Carbon and Brewer Science

For more information about ISS, click here.

Scientists from Jülich together with colleagues from Aachen and Turin have produced a memristive element made from nanowires that functions in much the same way as a biological nerve cell. The component is able to both save and process information, as well as receive numerous signals in parallel. The resistive switching cell made from oxide crystal nanowires is thus proving to be the ideal candidate for use in building bioinspired “neuromorphic” processors, able to take over the diverse functions of biological synapses and neurons.

Computers have learned a lot in recent years. Thanks to rapid progress in artificial intelligence they are now able to drive cars, translate texts, defeat world champions at chess, and much more besides. In doing so, one of the greatest challenges lies in the attempt to artificially reproduce the signal processing in the human brain. In neural networks, data are stored and processed to a high degree in parallel. Traditional computers on the other hand rapidly work through tasks in succession and clearly distinguish between the storing and processing of information. As a rule, neural networks can only be simulated in a very cumbersome and inefficient way using conventional hardware.

Systems with neuromorphic chips that imitate the way the human brain works offer significant advantages. Experts in the field describe this type of bioinspired computer as being able to work in a decentralised way, having at its disposal a multitude of processors, which, like neurons in the brain, are connected to each other by networks. If a processor breaks down, another can take over its function. What is more, just like in the brain, where practice leads to improved signal transfer, a bioinspired processor should have the capacity to learn.

“With today’s semiconductor technology, these functions are to some extent already achievable. These systems are however suitable for particular applications and require a lot of space and energy,” says Dr. Ilia Valov from Forschungszentrum Jülich. “Our nanowire devices made from zinc oxide crystals can inherently process and even store information, as well as being extremely small and energy efficient,” explains the researcher from Jülich’s Peter Grünberg Institute.

For years memristive cells have been ascribed the best chances of being capable of taking over the function of neurons and synapses in bioinspired computers. They alter their electrical resistance depending on the intensity and direction of the electric current flowing through them. In contrast to conventional transistors, their last resistance value remains intact even when the electric current is switched off. Memristors are thus fundamentally capable of learning.

In order to create these properties, scientists at Forschungszentrum Jülich and RWTH Aachen University used a single zinc oxide nanowire, produced by their colleagues from the polytechnic university in Turin. Measuring approximately one ten-thousandth of a millimeter in size, this type of nanowire is over a thousand times thinner than a human hair. The resulting memristive component not only takes up a tiny amount of space, but also is able to switch much faster than flash memory.

Nanowires offer promising novel physical properties compared to other solids and are used among other things in the development of new types of solar cells, sensors, batteries and computer chips. Their manufacture is comparatively simple. Nanowires result from the evaporation deposition of specified materials onto a suitable substrate, where they practically grow of their own accord.

In order to create a functioning cell, both ends of the nanowire must be attached to suitable metals, in this case platinum and silver. The metals function as electrodes, and in addition, release ions triggered by an appropriate electric current. The metal ions are able to spread over the surface of the wire and build a bridge to alter its conductivity.

Components made from single nanowires are, however, still too isolated to be of practical use in chips. Consequently, the next step being planned by the Jülich and Turin researchers is to produce and study a memristive element, composed of a larger, relatively easy to generate group of several hundred nanowires offering more exciting functionalities.

The semiconductor manufacturing industry is fighting to attract, educate, and retain the necessary talent for its continued growth. A significant workforce gap of up to 10,000 global positions stretches the industry’s ability to meet the world’s already demanding technology needs. To solve this challenge, SEMI, the global electronics manufacturing association, is launching an audacious and innovative campaign to raise industry awareness and attract students and recent graduates that don’t yet know about the immense opportunities available to them in semiconductor manufacturing.

Semiconductors are the brains and memory of all modern electronics. Their incredible processing power has made breakthroughs possible in communication, transportation, and medicine, powering everything from smartphones to space travel. Whether you’re driving a car, surfing the internet or using a computer, semiconductors drive technological innovation. Global semiconductor revenue has grown by over $100 billion in the last four years and is projected to surpass $0.5 trillion by 2019.

The campaign, You’re Welcome, speaks to how fundamental, yet underappreciated, this technology is. It includes a suspenseful, action-filled movie trailer that shows what happens when scientists, engineers, and mathematicians make semiconductors to save the world from the brink of disaster. The video also takes viewers behind-the-scenes of a semiconductor facility, or fab, which brings together cutting-edge STEM fields to develop the world’s most advanced technology. The campaign’s website provides information about the value and production of semiconductors, as well as a career guide that showcases the wide variety of opportunities available with companies such as Intel, Samsung, Applied Materials, Tokyo Electron, and the more than 2,000 SEMI member companies.

The campaign is just one piece in SEMI’s comprehensive workforce development plan. The plan engages students as early-on as 4th grade, inspires and motivates them through high school and college, and provides pathways to professional careers, building a pipeline to fill the short-term and long-term needs of the industry. Through the You’re Welcome campaign, SEMI is addressing the increasingly urgent workforce need by taking a completely new, never-before-seen approach to talent recruitment by leveraging high-interest areas of entertainment, media and storytelling to excite students about the industry’s role in society.

Leti, a research institute at CEA Tech, has proven that RRAM-based ternary-content addressable memory (TCAM) circuits, featuring the most compact structure developed to date, can meet the performance and reliability requirements of multicore neuromorphic processors.

TCAM circuits provide a way to search large data sets using masks that indicate ranges. These circuits are, therefore, ideal for complex routing and big data applications, where an exact match is rarely necessary.  TCAM circuits allow searching for stored information by its content, as opposed to classic memory systems in which a memory cell’s stored information is retrieved by its physical address. They shorten the search time compared to classic memory-based search algorithms, as all the stored information is compared with the searched data in parallel, within a single clock cycle.

But conventional SRAM-based TCAM circuits are usually implemented with 16 CMOS transistors, which limits storage capacity of TCAMs to tens of Mbs in standard memory structures, and takes up valuable silicon real estate in neuromorphic computing spiking neural-network chips.

The breakthrough of the CEA-Leti project replaced SRAM cells with resistive-RAM (RRAM) in TCAM circuits to reduce the number of required transistors to two (2T), and to two RRAMs (2R), which is the most compact structure for these circuits produced to date. In addition, the RRAMs were fabricated on top of the transistors, which also consumed less area. This suggests such a 2T2R structure can decrease the required TCAM area by a factor of eight compared to the conventional 16-transistor TCAM structure.

But while using RRAMs in TCAM circuits significantly reduces both silicon chip area needed and power consumption, and guarantees similar search speed compared to CMOS-based TCAM circuits, this approach brings new challenges:

  • Circuit reliability is strongly dependent on the ratio between the ON and OFF states of the memory cells. RRAM-based TCAM reliability could be affected by the relatively low ON/OFF ratio (~10-100) with respect to the 16-transistor structure (~), and
  • RRAMs have a limited endurance with respect to CMOS transistors, which can affect the lifespan of the system.

Overcoming these challenges requires trade-offs:

  • The voltage applied during a search operation can be decreased, which improves system reliability. However, this also degrades system performance, e.g. slower searches, and
  • The limited endurance can be overcome by either decreasing the voltage applied during each search, or increasing the power used to program the TCAM cells beforehand. Both increase system endurance, while slowing searches.

The work, presented Dec. 4 at IEDM 2018 in a paper entitled, “In-depth Characterization of Resistive Memory-based Ternary Content Addressable Memories”, clarifies the link between RRAM electrical properties and TCAM performance with extensive characterizations of a fabricated RRAM-based circuit.

The research showed a trade-off exists between TCAM performance (search speed) and TCAM reliability (match/mismatch detection and search/read endurance). This provides insights into programming RRAM-based TCAM circuits for other applications, such as network packets routing.

“Assuming many future neuromorphic computing architectures will have thousands of cores, the non-volatility feature of the proposed TCAM circuits will provide an additional crucial benefit, since users will have to upload all the configuration bits only the first time the network is configured,” said Denys R.B. Ly, a Ph.D. student at Leti and lead author of the paper. “Users will also be able to skip this potentially time-consuming process every time the chip is reset or power-cycled.”