AMD plans $135M for R&D in Ireland; Aurora supercomputer one step closer; RF and microwave design; Microsoft goes topological for quantum push.
AMD plans to spend $135 million in Ireland over four years to boost its adaptive computing segment, formerly Xilinx. The investment will fund R&D projects for next generation AI, data center, networking, and 6G communications infrastructure. The company will also add up to 290 engineering and research positions.
Argonne National Laboratory installed the final blade of its Aurora supercomputer. Built by Intel and Hewlett Packard Enterprise (HPE), Aurora will be theoretically capable of delivering more than two exaflops of computing power. Next, early users will stress test the system prior to deployment, including using it to train large-scale open source generative AI models for science. The supercomputer, which comprises 63,744 Intel Data Center GPU Max Series and 21,248 Intel Xeon CPU Max Series processors, will support a wide range of scientific research using simulations, data analytics, and AI.
The rapidly growing number of features and options in chip design means more elements in the design flow need to be choreographed much more precisely. Some steps have to shift further left, while others need to be considered earlier in the planning process even if the work happens later in the flow. Otherwise, problems can crop up anytime in the design cycle that are more costly to fix, and schedules may be delayed that can leave some resources idle while others are scrambling to meet deadlines.
Revenue of the global top ten IC design houses remained about the same in the first quarter of 2023 as it was in the prior quarter, with a 0.1% increase putting total revenue at $33.86 billion, according to TrendForce.
Keysight Technologies updated its radio frequency and microwave (RF/uW) design software suite, adding new millimeter wave (mmWave) and sub-terahertz (sub-THz) frequency capabilities for 5G mmWave product design and future 6G wireless communications development. Updates to PathWave Advanced Design System (ADS) 2024 include faster 3D-EM and 3D-Planar meshing and solvers; design sign-off directly from ADS for LVS, LVL, DRC, and ERC for MMICs; electrothermal validation of dynamic device operating temperatures under different bias and waveform conditions; and custom workflow support. Additionally, instruments and software from Keysight received Qualcomm Development Acceleration Resource Toolkit (QDART) validation for the Qualcomm QRU100 5G RAN Platform, which enables Open RAN Radio Unit (O-RU) and gNodeB (gNB) vendors to verify products using the Qualcomm 5G RAN Platforms throughout the design and production workflow.
Quadric announced that its general purpose neural processing unit (GPNPU) IP supports vision transformer (ViT) machine learning inference models. ViT models can be used for image and vision processing in embedded systems and repeatedly interleave MAC-heavy operations with DSP/CPU centric code. The company’s Chimera core family intermixes integer MAC hardware with a general purpose 32-bit ALU functionality.
Programming processors is becoming more complicated as more and different types of processing elements are included in the same architecture. While systems architects may revel in the number of options available for improving power, performance, and area, the challenge of programming functionality and making it all work together is turning out to be a major challenge. It involves multiple programming tools, models, and approaches from various IP providers.
Infineon Technologies introduced an embedded security solution. Optimized for the PSoC and AIROC family of products, the software offers four categories of security configurations for customer requirements in IoT applications based on increasing security capabilities and preconfigured product-security categories to satisfy regulatory and industry-standard levels.
Xpeedic updated its suite of tools for design of RF devices. The system-level design and simulation platform features a new filter synthesis algorithm and supports parametric padstack and permittivity, SNP-based LC matching in Smith Chart, bondwire simulation, and hierarchy design for schematic and layout.
In a perfect world, a semiconductor device would be verified to operate according to its complete specification, and continue to operate correctly over the course of its useful life. The reality, however, is this is becoming much more difficult. Chips are being deployed in more mission-critical and safety-critical applications. Device geometries make random faults more likely. Aging can cause chips to degrade over time. And it has become very lucrative to force devices to do things outside their intended functionality.
DB GlobalChip deployed the Cadence Spectre FX Simulator, integrated with Spectre AMS Designer, to verify its crucial analog and mixed-signal IP, including for its analog transistor-level designs and to include digital control logic to account for analog/digital interaction effects.
Codasip selected SmartDV to provide peripheral design IP, enabling customers to license a selection of SmartDV peripheral IP that has been validated for compatibility and ease of integration under a single license agreement and contract.
Keysight’s PathWave Test Executive for Manufacturing Developer Version now supports automated RF calibration and verification testing for the Autotalks cellular vehicle-to-everything (LTE V2X) and dedicated short-range communications (DSRC) chipsets.
Ansys and OneSky Systems are teaming up on software to train and validate neural networks with mission-driven simulation for advanced air mobility and unmanned aircraft systems.
Microsoft laid out its plans for quantum computing, starting with a new benchmark it calls reliable Quantum Operations Per Second (rQOPS). The metric counts operations that remain reliable for the duration of a practical quantum algorithm. The company estimates a quantum supercomputer capable of simulating simple models of correlated materials will need one million rQOPS.
In the first step toward reaching this goal, Microsoft announced creation of a brand-new qubit it claims has inherent stability at the hardware level. It is based on semiconductor-superconductor heterostructure nanowire that can create and control a topological phase of matter characterized by Majorana particles. Next, it will work toward building a hardware-protected qubit.
On the software side, Microsoft introduced a product that combines HPC and AI to speed up certain chemistry simulations. It uses a GPT-4 model which has been enhanced by feeding it additional data in the areas of quantum computing and chemistry. The company also debuted an AI-based assistant to aid in writing code for quantum chemistry problems.
IBM and UC Berkeley put classical and quantum computers head-to-head in complex materials simulation. IBM used error mitigation techniques with its Eagle quantum processor composed of 127 superconducting qubits on a chip to generate large, entangled states that simulate the dynamics of spins in a model of material and accurately predict properties such as its magnetization. To verify the accuracy, a team at UC Berkeley simultaneously performed these simulations on advanced classical computers located at Lawrence Berkeley National Lab’s NERSC and Purdue University. With the help of the error mitigation techniques, the quantum computer continued to turn out accurate results as the scale of the model increased, while classical approximation methods eventually faltered, according to the researchers. Results were checked against an exact brute-force classical calculation. More details on the error mitigation is available in a blog.
IonQ announced commercial availability of its latest trapped ion quantum computer, with 32 physical qubits and 29 algorithmic qubits (IonQ’s metric for the size of circuits that can be run). Forte replaces a key control system from previous system generations with an acousto-optic deflector, which the company says results in more accurate qubit control signals and improved quantum gate operation accuracy. The technology allows for control lasers to be individually addressed to individual qubits for increased software configurability.
Archer Materials optimized its carbon-based qubit material to mimic a high vacuum environment, preserving quantum functionality at room temperature in air, by using ALD and PECVD to encapsulate the qubit material with nanometer and micrometer thin films of metal oxides and other semiconductors.
Researchers at Chalmers University of Technology developed open-source, freely available software for advanced simulations and analyses of superconductive quantum components at the mesoscopic level (micrometers down to nanometers), which is small enough for quantum properties to become apparent but large enough to be applied in practice. “We are specifically interested in unconventional superconductors, which are an enigma in terms of how they even work and what their properties are. We know that they have some desirable properties that allow quantum information to be protected from interference and fluctuations. Interference is what currently limits us from having a quantum computer that can be used in practice. And this is where basic research into quantum materials is crucial if we are to make any progress,” said Mikael Fogelström, professor of theoretical physics at Chalmers.
The UK government is putting £75 million (~$96 million) into quantum computing. UK Research and Innovation (UKRI) will put £45 million (~$58 million) behind projects working on quantum technologies for position, navigation, and timing; software enabled quantum computation; and feasibility studies in quantum computing applications. Meanwhile, the National Quantum Computing Centre (NQCC) will deploy £30 million (~$38 million) to fund prototype quantum computers.
X. Sharon Hu was selected to receive the 2023 Marie R. Pistilli Women in Electronic Design Award. The award honors her research and development contributions in low-power system design, circuits and architectures for emerging technologies, real-time embedded systems, and hardware-software co-design, as well as her mentorship of women early in their career path.
Check out the latest Low Power-High Performance and Systems & Design newsletters for these highlights and more:
If you’d like to receive Semiconductor Engineering newsletters and alerts via email, subscribe here.
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
Leave a Reply