Week In Review: Design, Low Power

AMD closes Xilinx acquisition; RF heterogenous integration; new AI processors; accelerator in memory.


AMD completed its acquisition of Xilinx. The all-stock deal ended up being valued at approximately $50 billion due to a rise in AMD’s share price (the deal was valued at $35 billion when announced). The Xilinx business will become the newly formed Adaptive and Embedded Computing Group (AECG), led by former Xilinx CEO Victor Peng, and will continue its FPGA, adaptive SoC, and software roadmaps along with expanding to solutions that incorporate AMD CPUs and GPUs. AMD CEO Lisa Su noted that the acquisition would allow the company to “capture a larger share of the approximately $135 billion market opportunity we see across cloud, edge and intelligent devices” for adaptive computing solutions.

Tools & IP
Cadence will provide services, tools, and Assembly Design Kits (ADKs) to Qorvo and their customers as part of the State-of-the-Art Heterogeneous Integrated Packaging (SHIP-RF) project, a U.S. Department of Defense initiative for design and manufacturing of next-generation RF and mixed-signal systems that are delivered as both packages and modules. As part of Qorvo’s Design Center team, Cadence will develop methodologies and flows to implement design and simulation capabilities for 2.5D and 3D heterogeneous integration.

LG Electronics signed a five-year enterprise license agreement with Ansys, which will enable it to integrate virtual R&D simulations based on computer-aided engineering into the early stages of product development for technologies and sectors like healthcare, sensors, equipment, and materials.

Synopsys reported first quarter 2022 financial results with revenue of $1.27 billion, an increase of 30.9% from the same quarter last year. “Synopsys began fiscal year 2022 with excellent momentum, supported by strength across all product groups and geographies. We delivered financial results at or above guidance targets, returned $250 million to shareholders through stock repurchases, and are raising our outlook for the year,” said Aart de Geus, chairman and co-CEO of Synopsys.

Keysight reported first quarter financial results for 2022 with revenue of $1.25 billion, an increase of 5.9% compared to the same period last year. “The Keysight team delivered a strong start to the year, exceeding the high end of our revenue and EPS guidance,” said Ron Nersesian, Keysight’s chairman, president, and CEO. “The Electronics Industrial Solutions Group delivered double-digit order and revenue growth for the sixth consecutive quarter. Record orders were driven by strong demand for automotive and semiconductor solutions, as well as broad general electronics applications.”

AI hardware
Aspinity launched its first analog machine learning chip. The company said its AML100 reduces always-on system power to under 100µA by determining data relevancy at the sensor while the data is still analog and keeping the digital components in low power mode until important data is detected. It is comprised of an array of independent, configurable analog blocks that are fully programmable within software to support a range of functions, including sensor interfacing and ML. It targets battery applications including voice recognition, security, predictive maintenance, and biomedical monitoring.

SK Hynix unveiled its PIM (Processing in Memory) technology. Its first product using the technology is the GDDR6-AiM (Accelerator in Memory) that adds computational functions to GDDR6 memory chips, which process data at 16Gbps. The company said that the combination of GDDR6-AiM with CPU or GPU instead of a typical DRAM makes certain computation speed 16 times faster. SK Hynix said it runs on 1.25V and by reducing data movement to the CPU and GPU, it can reduce power consumption by 80%. It targets machine learning, high-performance computing, and big data computation and storage.

STMicroelectronics introduced its Intelligent Sensor Processing Unit (ISPU), which combines a DSP suited to run AI algorithms and a MEMS sensor on the same silicon. The processor offers a full-precision floating-point unit, uses a fast four-stage pipeline, operates from 16-bit variable-length instructions, and includes a single-cycle 16-bit multiplier, with a four cycle interrupt response. ST’s sensors with ISPUs will be packaged in standard 3mm x 2.5mm x 0.83mm packages and will be pin compatible with ST predecessors for upgrades. “It advances the sensor’s features to speed decision-making by reducing data transfers, enhancing privacy by keeping data local, while reducing size and power consumption, which cuts costs,” said Andrea Onetti, Executive Vice President, MEMS Sub-Group, STMicroelectronics. “Moreover, the ISPU is easily programmable with commercial AI models and can ultimately operate with all of the leading AI tools.”

AlphaICs started sampling its Gluon edge AI inference co-processor. At 8 TOPS, it focuses on classification and detection neural networks, and the company said it reached 32 FPS/Watt for Yolo-V2 and 22 FPS/Watt for VGG-19. It targets vision market segments, such as surveillance, industrial, retail, Industrial IoT, and edge gateway manufacturers.

QuickLogic announced availability of its PolarPro 3 family of low power, SRAM-based FPGAs. The family has power consumption as low as 55uA and up to 2,000 effective logic cells of SRAM-based FPGA fabric, 64Kbits of SRAM, and up to 46 configurable I/Os. It targets a variety of functions in handheld, wearable, mobile, IoT and other battery-powered applications including interfaces, level shifting, small CPU cores, low speed serial protocols, and GPIO muxing.

Infineon launched a new Trusted Platform Module. The OPTIGA TPM SLB 9672 features a firmware update mechanism with a 256-bits key length, along with an additional check based on post-quantum cryptography (PQC) using XMSS signatures, allowing it to still be updated if the standard algorithms are no longer trusted. It includes an expanded non-volatile memory to store new features such as additional certificates and cryptographic keys. Security evaluation and certification are performed by independent bodies according to the Common Criteria and FIPS requirements.

Samsung Research and Keysight inked a memorandum of understanding to develop test and verification technologies for 6G wireless systems, including accelerating the development of AI-enabled air interfaces that leverage massive multiple input multiple output (MIMO) antenna technology.

Power devices
Infineon plans to invest more than €2 billion (~$2.3B) to expand its wide bandgap manufacturing capacity, building a third module at its site in Kulim, Malaysia. The new module will supply products based on silicon carbide and gallium nitride. Construction will begin in June and the fab will be ready for equipment in summer 2024. The first wafers will leave the fab in the second half of 2024.

Transphorm will list on Nasdaq next week under the ticker TGAN. It is currently listed on the OTCQX market. The company designs and manufactures high voltage, high reliability gallium nitride (GaN) power conversion products. “This uplisting should raise the visibility of Transphorm in the capital markets and allow for increased sponsorship from worldwide institutional investors as well as increased liquidity in the trading volume of our stock,” said Primit Parikh, president and co-founder of Transphorm.

Power Integrations added two new AEC-Q100 qualified, 1700-volt rated ICs to its InnoSwitch3-AQ family. Delivering up to 70 watts of output power, the new ICs incorporate a silicon carbide (SiC) primary switching MOSFET and are targeted for use in 600- and 800-volt battery and fuel-cell electric passenger vehicles, as well as electric buses, trucks, and a wide range of industrial power applications.

Quantum computing
IonQ and the Pacific Northwest National Laboratory (PNNL) teamed up to create barium qubits for IonQ’s quantum computers. The collaboration focused on creating a manufacturing pipeline to supply barium qubits in perpetuity, allowing the company to scale its manufacturing operations. Additionally, the teams were able to shrink the barium source material to a microscopic scale, reducing the size of core system components.

Leave a Reply

(Note: This name will be displayed publicly)