Xilinx made several announcements Tuesday, the first day of its Adapt 2021 conference, including new devices in its Versal adaptive computing acceleration platform that are graded for the aerospace and defense sectors, a new robotics stack for its Kria system-on-module family, and more.
For starters, the company is launching two new versions of its Versal ACAP products with the addition of the space-grade XQR Versal portfolio and the defense-grade XQ ruggedized Versal portfolio. These products are being positioned at sectors of growing importance for many semiconductor companies, although Xilinx’s most recent earnings report in late July showed evidence of declining space and defense sales for the company.
The space-grade portfolio includes the new Versal AI Core and Versal AI Edge series devices that are targeted at on-board data processing and machine learning inference applications in satellite and space 2.0 environments. These radiation-tolerant devices, which support full reconfiguration while in orbit, follow closely on the recent unveiling of radiation-tolerant, space-grade FPGAs by Xilinx. They are scheduled to be available late in the second quarter of 2022.
The defense-grade XQ devices are heterogeneous multicore compute platforms equipped with adaptable AI and machine learning capabilities, and built to withstand harsh environments. The latest portfolio, targeted for initial availability in the first quarter of next year, includes the XQ AI Core, XQ AI Edge, XQ Prime and XQ Premium families.
Meanwhile, in the automotive sector, Xilinx is announcing new research on LiDAR sensors this week. The company, which supplies silicon for LiDAR products for automotive and other applications, teamed with Strategy Analytics on the research, which found that about $1 billion is being invested in up to 100 different companies developing high-resolution LiDAR sensors, and that automotive market adoption in particular is accelerating.
The research comes after critics such as Tesla CEO Elon Musk have questioned the wisdom of using the potentially costly technology, but Xilinx said in an email to Fierce Electronics that it’s simply trying to highlight what’s really happening in the market in the automotive industry and elsewhere.
“The research was done to showcase how FPGAs and Adaptive SoCs are being used to further LiDAR innovation in real world deployments,” the company said. “Automotive aside we see LiDAR being used in a wide range of applications in the field including medical, land management, archeology, and agriculture.
Xilinx said the capabilities of its FPGAs and adaptive SoCs, such as advanced signal processing, point cloud pre-processing, and point cloud ML acceleration, bring tremendous value to LiDAR sensor architectures. Mark Fitzgerald, director for Strategy Analytics’ Autonomous Vehicle Service, will be at the Adapt 2021 even to discuss the research and provide a further market outlook.
Among other Adapt 2021 announcements:
- Xilinx is expanding its Kria system-on-module portfolio, initially unveiled last April. beyond vision AI and is now equipping roboticists with hardware-accelerated functions and tools via the new Kria robotics stack (KRS). Built around the Robot Operating System (ROS), a framework for robot application development, KRS is comprised of an integrated set of utilities helping speed up ROS 2 and Gazebo, empowering faster, secure, and more complete robot behavior, Xilinx said. Eventually, robotic-focused accelerated applications built with KRS will be made available via the Xilinx App Store.
- Microsoft has announced that Azure is now utilizing its NP-VM FPGA-as-a-Service infrastructure, powered by Xilinx Alveo accelerators, to dramatically improve the performance of Apache Spark on the Azure Synapse analytics platform. Alveo acceleration delivers up to a 40x improvement in data parsing CSV and JSON file formats, eliminating a common bottleneck in Spark performance, Xilinx said.
- Xilinx also has released the Xilinx Video Transcoding SDK, making it easier and faster than ever before for streamers to build new or migrate existing services to take advantage of Xilinx acceleration.
- Also in video, Xilinx is announcing the Vitis Video Analytics SDK, a video analytics software development kit providing a complete software stack for intelligent video analytics applications on Xilinx platforms, deployed edge-to-cloud. The end-to-end video analytics solution enables video decoding, pre-processing, AI inference, tracking and post-processing for whole application acceleration. The SDK intends to provides the foundational layer for video analytic applications such as understanding vehicle and pedestrian traffic patterns in smart cities, health and safety monitoring in hospitals, self-checkout and analytics in retail, and detecting component defects at manufacturing facilities.
- Xilinx also has expanded its video and imaging IP portfolio for UltraScale+ and Versal devices, enabling higher-quality 8K video capture and 8K display with HDMI 2.1 and DisplayPort (DP) 1.4. Combining HDMI 2.1 or DP 1.4 with its 8K scaler and mixer IPs allows customers to build a complete, end-to-end 8K video pipeline. To provide gamers with immersive gaming experiences, Xilinx video and imaging portfolio now includes High Dynamic Range (HDR), High Frame Rates and Variable Refresh Rate formats and adaptive sync across the UltraScale+ and Versal product families.
- In the realm of wired and wireless communications, Xilinx said it has begun shipping its new Zynq RFSoC DFE in volume to multiple radio customers worldwide including a top wireless system vendor. Zynq RFSoC DFE is a new class of radio platforms that integrates hardened digital front-end (DFE) application-specific blocks for 5G NR performance and power savings while also offering integrated programmable adaptive logic to enable a futureproof solution for evolving 5G 3GPP standards and O-RAN radio architectures.