- AI IP for Cybersecurity monitoring - Smart Monitor
- ARC EV Processors are fully programmable and configurable IP cores that are optimized for embedded vision applications
- Enhanced Neural Processing Unit for safety providing 32,768 MACs/cycle of performance for AI applications
- EV74 processor IP for AI vision applications with 4 vector processing units
- EV7x Vision Processors
- EV7xFS Vision Processors for Functional Safety
- More Products...
IP-SOC DAYS 2025 IP-SOC DAYS 2024 IP-SOC DAYS 2023 IP-SOC DAYS 2022 IP-SOC DAYS 2021 IP-SOC 2024 IP-SOC 2023 IP-SOC 2022 IP-SOC 2021
|
|||||||
![]() |
|

At Embedded World 2023, Dolphin Design showcases AI-based vision applications at sub-mW level that fit in less than 1MB RAM
- See the 2025 Best Edge AI Processor IP at the Embedded Vision Summit
- Cadence Unveils Millennium M2000 Supercomputer with NVIDIA Blackwell Systems to Transform AI-Driven Silicon, Systems and Drug Design
- Cadence Accelerates Physical AI Applications with Tensilica NeuroEdge 130 AI Co-Processor
- Kyocera Licenses Quadric's Chimera GPNPU AI Processor IP
- The future of AI runs on the GPU
- Perforce Partners with Siemens for Software-Defined, AI-Powered, Silicon-Enabled Design (May. 16, 2025)
- Semidynamics: From RISC-V with AI to AI with RISC-V (May. 16, 2025)
- TSMC Board of Directors Meeting Resolutions (May. 16, 2025)
- Arm Evolves Compute Platform Naming for the AI Era (May. 16, 2025)
- Secafy Licenses Menta's eFPGA IP to Power Chiplet-Based Secure Semiconductor Designs (May. 15, 2025)
- See Latest News>>
Mar. 14, 2023 –
Grenoble --March 14, 2023. Dolphin Design, a leader in silicon IPs for Very Edge Computing embedding AI, is demonstrating at Embedded World 2023 its total solution for AI-based vision SoCs, which comprises a scalable and highly energy-efficient NN accelerator IP, an easy-to-use SDK, up to a library of efficient and ultra-compact AI models. The technology breakthrough from Dolphin Design is showcased through a device-like smart camera demonstrator called CamCube. It proves feasible to run AI-based vision applications, such as gesture recognition and tracking of people and hands, at a record-breaking budget of less than 1 mW with 20 frames per second, while fitting in less than 1MB of on-chip memory.
“Ultra-compact AI models for vision combined with highly energy-efficient silicon hardware not only meet the pressing demand from devices makers, but it also paves the way for innovation in connected devices!” says Philippe Berger, CEO of Dolphin Design.
|youtubevideo:SIV1Aqj7J5s|
CamCube in action
“We built the CamCube device demonstrator with two goals in mind. The first one is to inspire and show silicon designers that bringing intelligent vision into a power-friendly and cost-effective ambient computing SoC is now possible, but also easy & fast. The second one is to showcase to devices makers aiming at wearable applications that solutions exist to turn their bright ideas into solutions for the mass market.”, adds Philippe Berger.
CamCube is a small and low-cost hardware module that detects people, tracks hands, performs gesture recognition, and enables voice interfaces. It can operate on battery for days and a standard USB-C port enables powering and integrating it with existing embedded system.
It embeds a demonstration chip specifically designed by Dolphin Design to prove, on silicon, the possibility of achieving unprecedent energy-efficiency through the combination of the company’s silicon IPs. The gesture recognition demonstration, which fits on the 1 MB RAM on-chip memory, runs within sub-mW power budget at 20 inferences per second. This is also made possible through the partnership with Neovision, an AI Engineering Consulting company, specializing in Deep Learning & Computer Vision. Neovision provides a new class of ultra-compact, but efficient, deep learning models for vision that leverages the unique capabilities of Raptor, the Dolphin Design’s NN accelerator IP.
The demochip in the CamCube leverages Dolphin Design’s silicon IP design platforms, in particular Raptor, the company’s ultra-low power Edge AI accelerator IP which was awarded in 2022 by an Embedded Award, that is capable to run high-performance machine learning models.
Raptor is a pure digital and scalable silicon IP. It re-invents energy efficiency through a technology-independent near-memory computing architecture that truly minimizes data movement to save system-level energy in comparison with traditional Edge AI architectures. Dolphin Design’s engineering team co-designed Raptor hardware and its SDK to drastically increase AI computing efficiency for allowing ambient computing to finally take off.
“Join us for a meeting with our team at Embedded World in Nuremberg (Germany), from the 14th to 16th of March. You will get a real demonstration and lively explanations on our unique IP recipe to help you design smart and extremely energy-efficient SoCs. We are located in the Embedded Vision area, hall 2 stand 452”, concludes Philippe Berger.
About Dolphin Design
DOLPHIN DESIGN, a subsidiary of Soitec, is a fast-growing IP provider and Turnkey Design Service company. Their unique know-how in optimizing energy efficiency earned Dolphin Design the honor of being one of the 1,000 companies worldwide to receive the Solar Impulse label. Dolphin Design enables companies to realize their vision in different verticals such as edge IoT, AIoT, multimedia, AI and 5G, or automotive and aerospace markets wherever energy efficiency and/or low power consumption is a necessity.
Our mantra to our customers is simple: “Tell us your biggest dream. Dare the impossible. We make it happen”.
For more information, please visit https://www.dolphin-design.fr/