Dolphin Design wins an Embedded Award for Tiny Raptor, its Energy-Efficient Neural Network AI Accelerator
Embedded World, Nuremberg, Germany --June 21, 2022 -- Dolphin Design, an advanced chip design company and leader in Edge Computing and AI platforms for the Internet of Things (IoT), receives the Embedded World Award 2022 in the start-up category for Tiny Raptor. This IP represents a breakthrough in the neural processing segment with a fully programmable product designed to run neural networks with high energy efficiency using the Near-Memory Computing architecture. With Tiny Raptor, IoT chip designers will be able to improve the latency and energy efficiency of their products.
Dolphin Design’s CTO, Vincent Huard, received the award during Embedded World.
Author: NuernbergMesse / Frank Boxler
Tiny RAPTOR is a power-efficient neural processing IP platform specialized in sound and vision, featuring near-memory computing with a smart DMA, local memory, and up to 128 processing elements in its most powerful version.
Tiny RAPTOR is a fully configurable accelerator designed to execute deep neural networks (DNN) in an energy-efficient way and to reduce the inference time needed to run Machine Learning (ML) Neural Networks (NN). Tiny RAPTOR fits particularly well with Dolphin Design‘s MCU subsystem CHAMELEON as well as with any other MCU subsystem.
TinyRaptor configurable architecture includes SmartDMA, local memory and a scalable array of processing elements. Credits: Dolphin Design
Tiny RAPTOR is a complete Neural Processor solution to deploy AI at the very edge combining Software and Hardware approaches, resulting from more than three years of development,
Starting from their traditional learning frameworks, customers can seamlessly quantize their algorithms (both Byte and sub-Byte quantization) and deploy them to Dolphin Design’s dedicated Hardware.
This very specific software/hardware co-development enables very high computing efficiency (more than 50%) and optimized Data Reuse (more than 90%) thanks to its unique Near-Memory Computing approaches, reaching unmatched energy efficiency results in TinyML benchmarks.
This delivers a low energy benchmark of 32uJ and a latency of 10ms for Visual Wake Words detection and 12uJ with a latency of 3.5ms for keywords in the MLperf Tiny inference benchmarks from MLcommons. These outstanding performances are measured on Dolphin Design’s demonstrator VEP in 22FDX from GlobalFoundries.
Tiny Raptor fits particularly well with both sound and vision use cases such as speech recognition, noise cancellation, sound recognition, face identification, object detection, and image classification. It enables AI market-favorite applications like surveillance, smart camera, wearable, TWS, smart speakers, and IoT sensor fusion.
Philippe Berger, CEO of Dolphin Design, said: “This award recognizes the hard work we have been doing for three years now, introducing a comprehensive list of coherent processing IP platforms enabling superior AI and Edge computing performances. Ultimate energy efficiency and ease of use brought by those new processing IPs are confirming we are on the right track to deliver on our promises to our customers to do much more with less energy and effort.”
Cyril Menon, Operations Senior Executive Vice President at Soitec and Chairman of the Dolphin Design Board, said: “This award is confirming Dolphin Design is now recognized as a leader in the Edge AI and TinyML market. It is marking the beginning of a great journey that should bring Dolphin technology to more and more smart devices embedding speech and voice recognition or image classification. We are proud to continue to support the development of the processing activity at Dolphin Design in France and Singapore, with their brand new design center dedicated to AI and hosted on Soitec premises, which was announced in January.”
Meet us at Embedded World
To learn more, you can contact us or stop by our booth (3A-225) at Embedded World to see demos including our toolchain and TinyML benchmarks running on silicon. We will also take part in the pitch contest on June 22, from 10:00 to 11:00 GMT+2, at the Exhibitor Forum Hall 2: Booth 2-520
About Dolphin Design
Dolphin Design is an innovative and fast-growing semiconductor company.
It employs more than 180 people, including 140 engineers who design key functions of an integrated circuit – called “IP blocks or platforms” – or even a complete integrated circuit – called ASIC or System-on-Chip (SoC). Our company has unique know-how in optimizing energy efficiency, with unbeaten low level of energy requirements (<15uJ) for KWS in Voice Controlled systems, which earned us the honor of being nominated for the innovation award at Embedded World and being promoted as one of the 1000 companies worldwide to receive the Solar Impulse label.
|
Dolphin Design Hot IP
Related News
- Imagination announces PowerVR Series3NX Neural Network Accelerator, bringing multi-core scalability to the embedded AI market
- VeriSilicon's Neural Network Processor IP Embedded in Over 100 AI Chips
- UNISOC and Imagination carry out strategic cooperation on AI based on IMG Series3NX neural network accelerator
- Siemens simplifies development of AI accelerators for advanced system-on-chip designs with Catapult AI NN
- SoftBank and EdgeCortix Partner to Jointly Realize Low-latency and Highly Energy-efficient 5G Wireless Accelerators
Breaking News
- Ubitium Debuts First Universal RISC-V Processor to Enable AI at No Additional Cost, as It Raises $3.7M
- TSMC drives A16, 3D process technology
- Frontgrade Gaisler Unveils GR716B, a New Standard in Space-Grade Microcontrollers
- Blueshift Memory launches BlueFive processor, accelerating computation by up to 50 times and saving up to 65% energy
- Eliyan Ports Industry's Highest Performing PHY to Samsung Foundry SF4X Process Node, Achieving up to 40 Gbps Bandwidth at Unprecedented Power Levels with UCIe-Compliant Chiplet Interconnect Technology
Most Popular
- Cadence Unveils Arm-Based System Chiplet
- CXL Fabless Startup Panmnesia Secures Over $60M in Series A Funding, Aiming to Lead the CXL Switch Silicon Chip and CXL IP
- Esperanto Technologies and NEC Cooperate on Initiative to Advance Next Generation RISC-V Chips and Software Solutions for HPC
- Eliyan Ports Industry's Highest Performing PHY to Samsung Foundry SF4X Process Node, Achieving up to 40 Gbps Bandwidth at Unprecedented Power Levels with UCIe-Compliant Chiplet Interconnect Technology
- Arteris Selected by GigaDevice for Development in Next-Generation Automotive SoC With Enhanced FuSa Standards
E-mail This Article | Printer-Friendly Page |