News

How the EU chips act could build "Innovation capacity" in Europe

Source: IEEE Spectrum, March 8, 2023 [1]
The European Commission wants Europe to increase its share of global semiconductor production to 20% by 2030 through the European Chips Act, which will approve appropriations for R&D, incentivize manufacturing, and take steps to make the supply chain more secure.

How Efinix is conquering the hurdle of hardware acceleration for devices at the edge

Source: EETimes, March 7, 2023 [2]
Efinix FPGAs are a microarchitecture-agnostic RISC-V instruction set architecture that surpasses other AI chipsets in terms of reconfigurability, power consumption, size, speed, and cost. They are built on the innovative quantum compute fabric made up of exchangeable logic and routing (XLR) cells, allowing for a high-density fabric in a small device package.

Figure promises first general-purpose humanoid robot

Source: IEEE Spectrum, March 3, 2023 [3]
Figure is a robotics startup that is unveiling the world's first commercially viable general purpose humanoid robot. It will have the ability to think, learn, and interact with its environment and is designed for initial deployment into the workforce to address labor shortages and lead the way in eliminating unsafe and undesirable jobs. The company has hired 40 engineers from institutions such as IHMC, Boston Dynamics, Tesla, Waymo, and Google X, most of whom have prior experience with humanoid robots or other autonomous systems.

Intel pushes out hybrid CPU-GPU compute beyond 2025

Source: The Next Platform, March 6, 2023 [4]
Intel can enter the GPU compute space due to the IT market's need for at least three competitors. Nvidia claims to have invented the modern Graphics Processing Unit, but their cards lacked good performance and low cost. Nvidia embraced general purpose GPU compute as academics were hacking GPUs to offload parallel calculations from CPUs to the GPU shader engines, and it has been the driving force in GPU compute in the datacenter.

Intel’s new Agilex 7 FPGAs deliver industry’s fastest transceivers

Source: Intel, March 6, 2023 [5]
Intel launched Intel Agilex® 7 FPGAs with F-Tile, equipped with the fastest field-programmable gate array (FPGA) transceivers available on the market and designed to help customers address challenges across the most bandwidth-intensive areas of the data-centric world, including data centers and high-speed networks.

Marvell's DSP dilemma? Networking’s tectonic shift led by Broadcom, Nvidia, Arista networks, Microsoft, Meta, Macom, and more

Source: SemiAnalysis, March 8, 2023 [6]
Marvell is facing a tectonic shift in the networking industry, with competitors such as Broadcom, Nvidia, Arista Networks, Microsoft, Meta, Macom, and many others rallying to design Marvell out. Marvell is fighting back, but the biggest question is where technology roadmaps, deployments, market share, and revenue share go.

Source: NVIDIA Blog, March 6, 2023 [7]
NVLink is a high-speed interconnect for GPU and CPU processors in accelerated systems, propelling data and calculations to actionable results.

So you want to do robots, Part 2: What do you need to invent?

Source: General Robots, February 9, 2023 [8]
First let’s be clear on the goal. You want to make a robot that changes the world. It is going to be used in all kinds of places, free humanity from drudgery and scarcity and be super freaking cool. The key to making this happen is to make robots work in human spaces.

Maximizing performance with massively parallel hash maps on GPUs

Source: NVIDIA Technical Blog, March 8, 2023 [9]
Hash maps are a popular data structure for information storage, but are seldom discussed in the context of GPU-accelerated computing. This post walks through the fundamentals of hash maps and how their memory access patterns make them well suited for GPU acceleration.

Papers

A new algorithm for real-time scheduling and resource mapping for robot operating systems (ROS)

Source: MDPI, January 24, 2023 [10]
ROS-2 has been developed to address real-time and fault constraints, but current implementations lack strong real-time scheduling and optimization of response time. This article presents a heuristic scheduling algorithm to improve resource mapping and system scheduling for robotics applications.

GPU acceleration of the ATLAS calorimeter clustering algorithm

Source: iopscience, March , 2023 [11]
The ATLAS Collaboration is evaluating the use of Graphical Processing Units as hardware accelerators to improve the performance of the Topological Clustering algorithm, which is used to reconstruct calorimeter showers by grouping cells according to their signal-to-noise ratio.

Previous Hardware Acceleration in Robotics Newsletters

Past ROS 2 Hardware Acceleration Working Group meetings

  1. Moore, S. K. (2023, March 8). How the EU chips act could build "Innovation capacity" in Europe. IEEE Spectrum. https://spectrum.ieee.org/eu-chips-act-imec ↩︎

  2. How Efinix is conquering the hurdle of hardware acceleration for devices at the edge. (2023, March 7). EETimes. https://www.eetimes.com/how-efinix-is-conquering-the-hurdle-of-hardware-acceleration-for-devices-at-the-edge/ ↩︎

  3. Ackerman, E. (2023, March 3). Figure promises first general-purpose humanoid robot. IEEE Spectrum. https://spectrum.ieee.org/figure-humanoid-robot ↩︎

  4. Morgan, T. P. (2023, March 6). Intel pushes out hybrid CPU-GPU compute beyond 2025. The Next Platform. https://www.nextplatform.com/2023/03/06/intel-pushes-out-hybrid-cpu-gpu-compute-beyond-2025/ ↩︎

  5. Intel’s new Agilex 7 FPGAs deliver industry’s fastest transceivers. (2023, March 6). Intel. https://www.intel.com/content/www/us/en/newsroom/news/new-agilex-7-fpgas-deliver-industrys-fastest-transceivers.html ↩︎

  6. Patel, D. (2023, March 8). Marvell's DSP dilemma? Networking’s tectonic shift led by Broadcom, Nvidia, Arista networks, Microsoft, Meta, Macom, and more. SemiAnalysis | Dylan Patel | Substack. https://www.semianalysis.com/p/marvells-dsp-dilemma-networkings?utm_source=substack&utm_medium=email ↩︎

  7. Merritt, R. (2023, March 6). What is NVLink? NVIDIA Blog. https://blogs.nvidia.com/blog/2023/03/06/what-is-nvidia-nvlink/ ↩︎

  8. Holson, B. (2023, February 9). So you want to do robots, Part 2: What do you need to invent? General Robots | Benjie Holson | Substack. https://generalrobots.substack.com/p/so-you-want-to-do-robots-part-2-what ↩︎

  9. Juenger, D., Iskos, N., Wang, Y., Hemstad, J., Hundt, C., & Sakharnykh, N. (2023, March 8). Maximizing performance with massively parallel hash maps on GPUs. NVIDIA Technical Blog. https://developer.nvidia.com/blog/maximizing-performance-with-massively-parallel-hash-maps-on-gpus/ ↩︎

  10. Chaaban, K. (2023, January 24). A new algorithm for real-time scheduling and resource mapping for robot operating systems (ROS). MDPI. https://www.mdpi.com/2076-3417/13/3/1532 ↩︎

  11. Fernandes, N. (2023, February 15). GPU acceleration of the ATLAS calorimeter clustering algorithm. iopscience. https://iopscience.iop.org/article/10.1088/1742-6596/2438/1/012044/pdf ↩︎