News

AMD introduces world’s largest FPGA-based adaptive SoC for emulation and prototyping

Source: AMD, June 27, 2023 [1]
AMD has announced the AMD Versal™ Premium VP1902 adaptive system-on-chip (SoC), the world's largest adaptive SoC. This emulation-class, chiplet-based device streamlines the verification of complex semiconductor designs, offering 2X2 the capacity of the previous generation. With FPGA-based emulation and prototyping, it provides the highest level of performance, allowing faster silicon verification and enabling developers to shift the design cycle and begin software development before silicon tape-out.

The state of Europe’s AMR market

Source: Robotics Business Review, June 27, 2023 [2]
On the podcast this week, our guest is Waku Robotics founder and CEO Victor Splittgerber. We talk about the recent Logimat tradeshow and the state of the AMR market in Europe.

VDMA: German robotics industry to hit an all-time high in 2023

Source: The Robot Report, June 16, 2023 [3]
The German robotics and automation industry is expected to reach an all-time high in 2023, with sales expected to increase by 13% to 16.2 billion euros. This is a significant increase from the previous 9% growth forecast. The industry is expected to overcome the previous record result from 2018 and overcome supply chain bottlenecks.

Rise of the robots raises a big question: What will workers do?

Source: the Guardian, June 24, 2023 [4]
Rapid developments in AI are forcing managers and politicians alike to confront profound questions about the future of jobs

Intel looks to regain semiconductor chip leadership from TSMC; Separates manufacturing and Fabless units

Source: HPCwire, June 27, 2023 [5]
Intel is separating its manufacturing unit from fabless operations for the first time in over 50 years. The company plans to grant independence to its foundry business by early 2024, allowing it to have its own accounting books. The goal is to serve emerging markets like cloud, edge, and AI. The new fabs will make chip manufacturing and design more efficient, and Intel's chip design business units will become customers of the fabs.

Accelerating mission-critical aerospace performance with GPUs

Source: Scientific Computing World, June 25, 2023 [6]
Silicon Mechanics offers servers, storage, and high-performance computing technologies for industries like aerospace/defense. Swift data processing and analysis are crucial for decision-making, and maximizing performance is essential. GPU-accelerated computing, which uses GPUs to handle large amounts of data and execute multiple tasks simultaneously, has gained widespread application in these industries, helping agencies maximize performance.

Bluespec launches new MCUX RISC-V processor that enables developers to implement custom instructions and add accelerators

Source: HPCwire, June 27, 2023 [7]
Bluespec Inc. has introduced the MCUX RISC-V processor, allowing developers to implement custom instructions and accelerators on FPGAs and ASICs. This extension of Bluespec's MCU RISC-V processor family offers more customization opportunities and vendor choice. The MCUX is fully portable across all major FPGA architectures and ASIC technologies, resulting in cost savings and the ability to work with any FPGA vendor without licensing restrictions.

Tesla AI capacity expansion – H100, dojo d1, d2, hw 4.0, X.AI, cloud service provider

Source: SemiAnalysis, June 27, 2023 [8]
Tesla aims to become a leading AI firm, but currently lacks advanced self-driving and generative AI capabilities. However, their data collection, specialized compute, innovative culture, and AI researchers could help them excel in autonomous vehicles and robotics. Tesla's weak AI infrastructure is partly due to delays in their D1 training chip.

The uncertainties of RISC-V compliance

Source: Semiconductor Engineering, June 23, 2023 [9]
Tests and verification are required, but that still won’t guarantee these open-source ISAs will work with other software.

Papers

Real-time detection of track fasteners based on object detection and FPGA

Source: ScienceDirect, May 30, 2023 [10]
This paper presents a solution for accurate fastener detection on railway tracks, addressing limited memory and processing capacity in embedded maintenance systems. The algorithm is implemented on FPGA, and hardware acceleration strategies like loop pipelining, array partition, and ping-pong buffer optimization are applied. The improved YOLOv2 detection network and fastener detection system outperform existing systems in processing time and resources, detecting track images in 41 ms.

Modeling variability in self-adapting robotic systems

Source: ScienceDirect, June , 2023 [11]
This paper proposes a modeling approach that explicitly represents robot software variability using the MARTE::ARM-Variability metamodel. The approach is evaluated by evaluating the software architecture design of a robot navigation framework and analyzing the support provided by ROS infrastructure for runtime reconfiguration of variable aspects.

Previous Hardware Acceleration in Robotics Newsletters

Past ROS 2 Hardware Acceleration Working Group meetings

  1. AMD introduces world’s largest FPGA-based adaptive SoC for emulation and prototyping. (2023, June 27). AMD. https://www.amd.com/en/newsroom/press-releases/2023-6-27-amd-introduces-world-s-largest-fpga-based-adaptive.html ↩︎

  2. The state of Europe's AMR market. (2023, June 27). Robotics Business Review. https://www.roboticsbusinessreview.com/autonomous-mobile-robots-amrs/the-state-of-europes-amr-market/ ↩︎

  3. Wessling, B. (2023, June 16). VDMA: German robotics industry to hit an all-time high in 2023. The Robot Report. https://www.therobotreport.com/vdma-german-robotics-industry-to-hit-an-all-time-high-in-2023/?spMailingID=92472&puid=2502618&E=2502618&utm_source=newsletter&utm_medium=email&utm_campaign=92472 ↩︎

  4. Rise of the robots raises a big question: What will workers do? (2023, June 24). the Guardian. https://www.theguardian.com/technology/2023/jun/24/rise-of-the-robots-raises-a-big-question-what-will-workers-do ↩︎

  5. Bernhardt, M. (2023, June 27). Intel looks to regain semiconductor chip leadership from TSMC; Separates manufacturing and Fabless units. HPCwire. https://www.hpcwire.com/2023/06/27/intel-looks-to-regain-semiconductor-chip-leadership-from-tsmc-separates-manufacturing-and-fabless-units/ ↩︎

  6. Accelerating mission-critical aerospace performance with GPUs. (2023, June 25). Scientific Computing World. https://www.scientific-computing.com/news/accelerating-mission-critical-aerospace-performance-gpus ↩︎

  7. Bluespec launches new MCUX RISC-V processor that enables developers to implement custom instructions and add accelerators. (2023, June 27). HPCwire. https://www.hpcwire.com/off-the-wire/bluespec-launches-new-mcux-risc-v-processor-that-enables-developers-to-implement-custom-instructions-and-add-accelerators/ ↩︎

  8. Patel, D. (2023, June 27). Tesla AI capacity expansion – H100, dojo d1, d2, hw 4.0, X.AI, cloud service provider. SemiAnalysis | Dylan Patel | Substack. https://www.semianalysis.com/p/tesla-ai-capacity-expansion-h100?utm_source=substack&utm_medium=email ↩︎

  9. Mutschler, A. (2023, June 23). The uncertainties of RISC-V compliance. Semiconductor Engineering. https://semiengineering.com/the-uncertainties-of-risc-v-compliance/?cmid=5c87a7fb-5d79-4c96-8270-7cc1647e5464 ↩︎

  10. Tian, X., Tianhua, X., & Guang, W. (2023, May 30). Real-time detection of track fasteners based on object detection and FPGA. ScienceDirect. https://www.sciencedirect.com/science/article/abs/pii/S0141933123001096 ↩︎

  11. Brugali, D. (2023, June 9). Modeling variability in self-adapting robotic systems. ScienceDirect. https://www.sciencedirect.com/science/article/abs/pii/S0921889023001094 ↩︎