Week In Review: Design, Low Power

EnSilica listed on the London Stock Exchange’s AIM market under the ticker ENSI. EnSilica designs mixed signal ASICs for system developers in the automotive, industrial, healthcare, and communications markets. It also has a portfolio of core IP covering cryptography, radar and communications systems. AIM is the LSE’s market for small and medium sized growth companies. “In connection with Admission, EnSilica has raised £6 million (~$7.4 million) and will have a market capitalization of approximately £37.6 million (~$46.6 million) at the Placing Price,” the company said in a statement.

SmartSens Technology listed on the Shanghai Stock Exchange’s Science and Technology Innovation Board, commonly known as the STAR Market, under stock code 688213. SmartSens designs high-performance CMOS image sensor (CIS) chips for a range of applications including surveillance, machine vision, automotive, and cellphone cameras.

Rambus finalized its acquisition of Hardent, a provider of design services and IP. Rambus said Hardent’s silicon design, verification, compression, and Error Correction Code (ECC) expertise will provide key resources for the Rambus CXL Memory Interconnect Initiative. “Driven by the demands of advanced workloads like AI/ML and the move to disaggregated data center architectures, industry momentum for CXL-based solutions continues to grow,” said Luc Seraphin, president and CEO of Rambus. “The addition of the highly-skilled Hardent design team brings key resources that will accelerate our roadmap and expand our reach to address customer needs for next-generation data center solutions.” Headquartered in Montreal, Canada, Hardent was founded in 2002. Terms of the deal were not disclosed.

Flex Logix joined the Silicon Catalyst In-Kind Partner Program, giving companies in the Silicon Catalyst Incubator access to Flex Logix’s embedded FPGA (eFPGA) IP and software, enabling silicon reconfigurability for use in their chip designs. “We’re excited to provide Silicon Catalyst Portfolio companies the ability to leverage RTL flexibility to get to market faster. Using our eFPGA, they can tape out their chips while key algorithms are still in development as well as broaden their chip’s addressable market through chip customization for unique end application requirements,” said Andy Jaros, VP IP Sales and Marketing for Flex Logix.

NSITEXE selected Imperas Software’s ImperasDV for advanced RISC-V processor hardware design verification. In particular, NSITEXE will use it for 64bit RISC V based designs featuring vector accelerators for AI automotive applications with verification leading to the level required to achieve ISO 26262 ASIL D.

Synopsys and Analog Devices will provide model libraries for ADI’s DC/DC ICs and µModule regulators with Synopsys’ Saber system-level simulation tool. The model libraries enable accurate multi-domain simulations for powertrain designers of products such as electric vehicles, avionic machines, instrumentation equipment, and supercomputers.

X-FAB integrated Cadence’s EMX Planar 3D Solver into the X-FAB RFIC workflow. As part of its 130 nm RF SOI PDK, X-FAB plans to supply reference designs that have been characterized using the EMX Solver to customers in the communication and automotive sectors. The PDK will include a wide array of inductor models which have all been pre-characterized by the EMX Solver.

Embedded
Siemens Digital Industries Software updated its Nucleus ReadyStart embedded development solution with new features for developing embedded products based on the RISC-V architecture. It incorporates integrated software, IP, tools, and services. “Nucleus ReadyStart for RISC-V includes SMP, POSIX, Debug Agent, and security protocols from wolfSSL/OpenSSL. Users can also connect their device to the cloud with the Nucleus IoT Framework Add-on which supports several cloud services and providers such as Amazon Web Services (AWS), Microsoft Azure and MindSphere, the industrial IoT as a service solution from Siemens. These new capabilities enable RISC-V-based embedded developers to customize rich features to meet the needs of today’s advanced device manufacturers,” said Scot Morrison, vice president and general manager, Embedded Platform Solutions at Siemens Digital Industries Software.

Wireless
Keysight Technologies, FormFactor, DMPI, and Virginia Diodes Inc. teamed up on a new 170 GHz / 220 GHz Broadband Vector Network Analysis (VNA) Solution to provide on-wafer millimeter-wave component characterization under various conditions. It aims to enable the creation of highly accurate process design kits (PDKs) used to design IC models that shorten the design and verification cycle. It also enables rapid deployment of devices for the emerging 5G and 6G monolithic microwave integrated circuits (MMICs).

MediaTek launched a dual connectivity mmWave and sub-6GHz 5G chipset. Built on TSMC’s 6nm process, it supports 3CC carrier aggregation on sub-6 (FR1) spectrum and 4CC carrier aggregation on mmWave (FR2) spectrum. The Dimensity 1050 SoC integrates two Arm Cortex-A78 CPUs with speeds reaching 2.5GHz and the latest Arm Mali-G610 graphics engine. MediaTek also announced two Wi-Fi 7 platform solutions for high-bandwidth applications in the operator, retail, enterprise and consumer electronics markets.

Infineon Technologies launched a new controller for inductive wireless charging. WLC1115 is a highly integrated, Qi-compliant, and fully configurable 15 W transmitter controller IC for wireless charging solutions. It includes a USB-PD/PPS sink, DC/DC controller, gate drivers for DC/DC, a full-bridge inverter, sensing peripherals, and configurable flash memory. It supports a wide input voltage range from 4.5 V to 24 V, multipath voltage- and current-based ASK demodulation. It also comes along with a Qi v1.3.2 certified MP A11 power transmitter reference solution, targeting applications including smartphones, smart speakers, docking stations, monitor stands, as well as industrial or healthcare accessories.

Data center
Broadcom plans to acquire VMware in a cash-and-stock transaction worth approximately $61 billion. In addition, Broadcom will assume $8 billion of VMware net debt. Best known for its virtualization technology, VMware currently offers a range of cloud software and services including application modernization, cloud management, cloud infrastructure, networking, and security. Following the closing of the transaction, the Broadcom Software Group will rebrand and operate as VMware, incorporating Broadcom’s existing infrastructure and security software solutions as part of an expanded VMware portfolio. The deal is expected to close in Broadcom’s fiscal year 2023.

AMD completed its acquisition of Pensando Systems in a deal worth approximately $1.9 billion. “The data center remains one of the largest growth opportunities for AMD. The addition of the Pensando Systems team with their hardware and software portfolio will enables us to offer cloud, enterprise and edge customers a broader portfolio of leadership compute engines that can be optimized for their specific workloads,” said AMD Chair and CEO Lisa Su. “Pensando’s leadership DPU complements our data center product portfolio, enabling AMD to offer solutions that can significantly accelerate data transfer speeds while providing additional levels of security and analytics that will play a larger role in defining the performance of next-generation data centers.”

Microsoft adopted AMD’s Instinct MI200 accelerators to power large scale AI training workloads. Additionally, Microsoft plans to work with the PyTorch Core team and AMD data center software team to optimize the performance and developer experience for customers running PyTorch on Microsoft Azure and ensure that developers’ PyTorch projects take advantage of the performance and features of AMD Instinct accelerators.

Memory & storage
Hewlett Packard Labs created a new memory circuit that mimics the operation of associative learning memory in the brain, called differentiable content addressable memory (dCAM). It derives from an analog CAM (aCAM) based on ReRAM devices, which enables searching an analog value that falls within a range, while still producing a digital output. “Our research discovered that a simple modification to the analog CAM circuit could produce a differentiable CAM (dCAM), which has a continuous relationship between input and output, allowing the new cell to be used for learning through backpropagation and other known techniques,” said Giacomo Pedretti, a research scientist in the AI Research Lab at Hewlett Packard Labs. “We believe we are just scratching the surface for the potential of a dCAM. For example, a dCAM could be paired with other neuromorphic blocks (such as a dot-product-engine) to develop fully differentiable computing systems able of learning complex tasks without prior knowledge or need for heavy programming. A hybrid memory which can learn to be more like a Dot Product Engine (DPE), or a CAM could change its behavior based on the workload.”

Imec suggests two potential types of liquid memory that could be used for ultra-high-density nearline storage applications with sub-second access times. The first is called colloidal memory, which would use a colloid of at least two types of nanoparticles contained in a reservoir. An array of capillaries only slightly larger than the nanoparticles would be attached to the reservoir. Electrodes would selectively induce a type of nanoparticle to enter a capillary, and data would be encoded by the order of the nanoparticles, or bits.

The second concept, called electrolithic memory, also uses a fluid reservoir and array of capillaries. However, instead of nanoparticles, the fluid contains at least two metal ions. Read and write operations happen through electrodeposition and electrodissolution techniques. “These storage solutions could, for example, make archived ‘inactive’ data such as email archives, image & sound files, or other large documents accessible to users within seconds. From 2030 on, they might find their place between HDD and tapes, at significantly higher bit per volume but slower than 3D-NAND Flash,” said Maarten Rosmeulen, program director of storage memory at Imec.

Showa Denko started shipping 3.5-inch hard disk media to be used in 26 TB HDDs. The new HD media supports energy-assisted magnetic recording and shingled magnetic recording (SMR). It utilizes a magnetic layer design and technology to produce fine crystals of magnetic substance on the surface of the aluminum platter to reach a storage capacity of 2.6TB per platter.

Three companies held 94% of the DRAM market share in 2021, according to IC Insights. Samsung remained the world’s largest DRAM supplier with a 44% market share and sales of $41.9 billion. SK Hynix followed, with 28% of the DRAM market and sales of $26.6 billion. Micron wasn’t far behind, with 23% of the global market share and sales of $21.9 billion.

Yole Développement expects the memory market to grow to $263 billion in 2027, a CAGR of 8% from 2021. The firm anticipates DRAM to reach $158 billion, a CAGR of 9%, while NAND grows to $96 billion, a CAGR of 6%. Emerging non-volatile memories are also poised to grow to a $1.8 billion market in 2027.

Quantum computing
Quantum computing software company Quantum Computing Inc. (QCI) will acquire QPhoton, a developer of quantum photonic systems that operate at room temperature and maintain computational stability in a variety of application environments. QCI will continue to offer vendor-neutral software and support a variety of quantum computing platforms. “QCI, with QPhoton’s technology, will be launching ready-to-run anywhere, full-stack quantum systems that can deliver affordable, user-friendly solutions for real business problems to a much larger audience,” said Robert Liscouski, CEO of QCI.

The Quantum Economic Development Consortium (QED-C), managed by SRI International and supported by the National Institute of Standards and Technology (NIST), announced a $2.3 million research program to advance cryogenic technologies for quantum technologies. The program consists of four projects: FormFactor will develop a novel load lock designed to dramatically reduce the time required to test quantum chips; Northrop Grumman will work on small cryocoolers in the 3 K to 5 K range; Quantum Opus will explore two parallel paths to a low-cost and compact 2.5 K cryocooler; and Triton Systems will demonstrate a multi-stage modified Collins cycle cryocooler to provide cooling at 4K.

Check out the latest Systems & Design and Low Power-High Performance newsletters to learn about they layers of processor optimization, the potential silver lining to the talent crunch, design choices in DRAM, and more.

Find more of the week’s news at Manufacturing, Test and Auto, Security, Pervasive Computing.