ELVEES SoC for smart devices, AR/VR, IoT, ADAS and other applications is sampling now

ELVEES reports availability of the first samples of its ELVEES Image Semantic Engine (ELISE) chip.

ELVEES companies group (Elveesion Inc., ELVEES NeoTek, and RnD Center ELVEES, JSC) reports availability of the first samples of its ELVEES Image Semantic Engine (ELISE) chip. ELISE is a unique next-generation multi-platform SoC for a wide range of existing and future global video analytics markets including augmented reality and virtual reality (AR/VR), advanced driver assistance systems (ADAS), vision based IoT systems, robotics, smart devices with navigation, and more. Companies can use the ELISE chip to develop a new generation of IP cameras with video analysis and a wide range of OEM modules, applicable across a broad spectrum of devices.

The ELISE chip design results from the company’s partnership with Imagination Technologies (IMG.L). It features IP from both Imagination and ELVEES. The 28nm ELISE SoC, manufactured in TSMC, is based on an innovative multi-platform architecture optimized for video analytics, multimedia, navigation and connectivity. Its innovative15+ core heterogeneous MIMD (multiple instruction, multiple data) architecture is designed to fulfill the needs of both high-performance mobile computing and video analytics.

Says Yaroslav Petrichkovich, CEO of RnD Center ELVEES: “Video analytics is important for many emerging markets as well as existing markets that are evolving to embrace its potential. With ELISE, we are providing a complete, state-of-the-art and highly competitive solution for these applications. Imagination’s proven, scalable IP enabled us to create a highly differentiated SoC with low risk and fast time to market.”

The heart of the ELISE design is a programmable ELVEES video processor (Velcore2), a coherent 8-core сluster with accelerator IP cores such as hardware stereo image SGBM accelerator that provides the foundation for such capabilities as up to 4K rate stereo streams for facial recognition, AR and more. Velcore2 provides effective support for most video analytics algorithms including Histogram of Oriented Gradients (HOG) and Convolutional Neural Network (CNN).

For the ELISE design, ELVEES licensed a broad range of IP cores from Imagination including arange of PowerVR multimedia cores for graphics and compute, video encode and decode and image processing – the optimized blocks work together to save power and bandwidth. In addition, the design includes multiple MIPS CPUs from entry level to high-end, which together deliver superior acceleration for compute-intensive video analytics applications and audio processing. The system incorporates Ensigma wireless communications IP for high-performance on-chip Wi-Fi and Bluetooth, and includes embedded security and encryption cores.

The ELISE design was accomplished with the multi-standard GNSS IP-core Navicore4 (GPS/GLONASS/BEIDOU/GALILEO).

Says David McBrien, EVP, sales and marketing at Imagination: “ELISE is an example of the innovation provided by the broad spectrum of Imagination’s IP cores. ELVEES has strong expertise and leading technologies for the growing video analytics market. We were delighted to work with ELVEES to make the powerful ELISE chip design a reality.”

ELISE is supported by the broad ecosystems of tools and technologies around the MIPS, PowerVR and Velcore technologies including Imagination’s PowerVR SDK for rapid graphics application development and the Codescape MIPS SDK designed to streamline programming, debugging and application optimization at all stages of SoC development. A comprehensive SDK for ELISE includes development tools, programming examples, detailed documentation and optimized libraries. Vision libraries support a rich set of vision processing functions including filtering, background processing optical flow, pattern matching, implementation, rectification, and the Khronos OpenVX™ API. In addition, users can leverage tools for: asset exporting and optimization; prototyping, developing and optimizing; recording, profiling and performance analysis; debugging and more.

ELVEES has developed several types of ELISE boards, which can be used to enable quick evaluation and software debugging for the ELISE SoC and as a reference design, ELISE based applications (ELISE evaluation board). ELISE reference 3D camera ELISE-3D, a binocular camera for embedded video analytics systems, can be used for a range of applications that require 3D scene and object analysis.

ELISE SoC engineering samples are available now.