powering intelligent edge with

Netrasemi Edge AI SOCs

Embedding real-time AI on edge devices with a power-efficient Graph Stream Parallel Architecture.

Explore the Power of Edge-AI computing for your IoT solution with Netrasemi SOCs

Enabling optimal computing for smart IoT solutions

AI for smart IoT devices are driving the need for new chip architectures and computing platforms. Traditional system on chips offer very limited capabilities for the new world of embedded AI IoT solutions. Emergence of distributed IoT compute model with the need for real-time on-device AI (edge-AI) processing is exposing major gap in the existing chip supply chain. New AI chipsets in the market are also having major gap in the domain specific features that are critical for the overall efficiency of a resource limited edge platform. To become a leader in the solution space, you may need to accelerate your market entry with competitive SOCs that provide optimal computing model, unique domain-specific features, power-performance efficiency and competitive pricing. Netrasemi helps you to achieve these goals with our domain specific architecture (DSA), IP rich SOCs, AI development tools, flexible SDKs, platform reference designs and unique go-to-market strategies. We bring 10x improvements to your time to market.

A Unified Domain Specific Architecture (DSA) for Smart Sensors to Edge Servers

Embedding intelligence in small and tiny devices is the latest trend in IoT. The “edge-AI” strategy is making sensors smart, cheap, responsive, and independent with minimal data to be sent to central computers. Smart sensors, aggregators and Edge servers work as key compute elements in a world of distributed AIoT. Each of these edge devices demand unique domain specific features in an aggressively small form factor and power budget. This drives custom chip architectures and software framework for each device, making the job of a solution builders extremely difficuilt. Netrasemi brings in a family of SoCs powered with a unique domain specific architecture that scales from tiny edge devices to edge servers. Our SoCs are powered by a family of high-performance acceleration cores, unique heterogeneous compute fabric and a UCIe based Die-to-Die (D2D) interconnect. This architecture allows us to enable true graph-stream processing without the power performance boundaries of a silicon die. The architecture is supported with a common software framework which allows our customers to build their distributed computing solution with a unified hardware-software architecture.

Faster Time to market for AI solutions

Our NetraSoC architecture comes with both ASIC (high volume) and FPGA technology targets. FPGA platforms are targeted for customers having agile go-to-market plans, smaller volumes and special customization needs. We have SoC-FPGA reference designs with AI/ML and domain-specific features.

Flexible APIs, SDKs, reference designs

NetraSoC designs come flexible software API libraries, software development kits (SDKs), NETRA SmartEdge studio development environment and application reference design kits (RDKs) to enable a faster time to market for our customers.

Graph stream architecture

Netrasemi graph heterogeneous parallel processing fabric integrates a variety of hardware acceleration cores ensuring seamless data flow and performance without major CPU load. This allows multi-data analytics at the best power efficiency suitable for edge-AI.

Compute Cores

NetraSoC family support multi-core CPU variants with Linux and RTOS support and security features. Our SoC family also include a rich portfolio of hardware accelerator cores, peripheral IP cores, high speed memory interfaces and low-power SoC features. It helps to achieve maximum solution flexibility, power efficiency, features and performance for high complexity workloads like video analytics.

Deep Neural Processor (NPU)

Netrasemi's Deep neural processor (Tej-NPU) IP core helps to achieve best-in-class AI/ML and smart vision performance. The high energy efficiency offered by our NPU architecture and supporting graph stream compute fabric are ideally suited for edge computing applications.

WHAT WE DO?

Making edge AI product development Efficient, Simple and Economical.

Netrasemi is an Edge AI semiconductor technology company building System-on-chips (SOC) products. We have a rich portfolio of In-house silicon IPs that enable edge AI with extreme power efficiency and performance. We are bringing a family of AI/ML capable SOCs and reference designs for solving 100+ AI/ML use cases. The chipsets feature 4 to 100 TOPs performance with best-in-class power efficiency backed by a strong Chiplet-based DSA & SOC roadmap. Our key target segments include Surveillance, Perimeter security, Smart sensors, Smart infrastructure, Machine vision & industry 4.0, Robotics, Drones, Smart Home & Office, Smart retail and Autonomous vehicles.

OUR TECHNOLOGY PARTNERS

Blog

Keep informed about what is happening in the technology domain.