DMP Released Next-Generation AI Accelerator IP "ZIA A3000 V2" - Industry-leading PPA efficiency to propel the future of edge AI
November 11, 2024 -- Digital Media Professionals Inc. (Headquarters: Nakano-ku, Tokyo, Chairman, President & CEO: Tatsuo Yamamoto, hereinafter referred to as DMP) has announced the launch of its next-generation Neural Processing Unit IP (NPU IP) “ZIA A3000 V2” (hereinafter referred to as A3000). Targeting the expanding edge AI semiconductor market, this new product will be available starting November 2024. A3000 combines scalable high performance, low power consumption, and compact size to drive the advancement of edge AI.
1. Key Features of A3000
(1) Scalable Performance and Precision Support
- Performance can be easily adjusted by changing the number of processing units and cores.
- Achieves inference processing capabilities exceeding 40 TOPS, enabling image processing and execution of complex AI algorithms.
- In a multi-core setup, each core can handle inference processing for different AI models
- Supports FP16, INT8, INT4, and FP4, allowing optimal processing by mixing high-precision and low-precision/high-speed computations.
(2) Compact Size
- Reduces area by over 50% compared to similar-class products.
- Supports 96 to 2048 MAC units per core, enabling performance expansion in multi-core configurations.
- Flexible IP core configuration to meet various chip size requirements.
(3) Optimized Architecture for Edge AI Devices
- Proprietary architecture ensures high performance, power efficiency, and compact size.
- Delivers high processing capability even under the strict constraints of edge devices.
- Supports adaptive power management, dynamically saving power when the IP core is idle, making it ideal for long-operating applications like security cameras.
- Optimized for processing ONNX instruction sets for edge vision AI tasks like object detection, segmentation, and face recognition.
(4) Comprehensive Development Environment
- Offers a comprehensive development environment to maximize the productivity of edge AI SoC developers.
- Supports major AI frameworks such as PyTorch, TensorFlow, and ONNX, facilitating easy porting of existing AI models.
- Advanced profiling tools provide detailed analysis of performance and memory usage by layer, helping identify and optimize bottlenecks.
- Includes rich sample source code and documentation for rapid development.
- FPGA environment available for precise performance and accuracy estimation before the actual chip is completed (Optional).
A3000 System Block
2. Broad Application Fields
A3000 serves as a core technology in next-generation edge AI applications across various fields, particularly supporting technological innovations in the following areas:
- Mobility: Advanced Driver Assistance Systems (ADAS), Driver Monitoring Systems (DMS)
- Security: High-performance AI camera systems
- Robotics: Industrial and service robots
- Smart City: Traffic management systems
- Smart Home Devices
3. Future Outlook
“A3000 combines our semiconductor technology capabilities with our expertise and experience in AI solution development. It uniquely offers both high performance and low power consumption required for edge AI devices. We aim to bring A3000 to advanced semiconductor developers worldwide to support the development of next-generation AI devices. By leveraging A3000, we intend to showcase Japan’s cutting-edge semiconductor technology to the world and contribute to the global development of the edge AI ecosystem,” said Tatsuo Yamamoto, Chairman, President and CEO of Digital Media Professionals.
DMP will continue to pursue innovations in edge AI computing and GPU technologies, contributing to the revitalization of Japan’s semiconductor industry. Through the provision of A3000, we aim to create and deliver new value to customers and society, striving to become a global leader in the edge AI solutions market.
About Digital Media Professionals Inc. (DMP) https://www.dmprof.com/en/
DMP is an R&D-driven fabless semiconductor vendor that originally specialized in licensing hardware IPs and software IPs based on its proprietary 2D/3D graphics technology for embedded devices, along with providing graphics LSI that incorporates these IPs since its founding in 2002. In recent years, to become the world’s leading “AI Computing Company”, DMP has provided solutions through a broad portfolio including AI processor IPs, hardware/software products and services, and a self-established AI ecosystem.
|
DMP Inc. Hot IP
Related News
- LeapMind's Ultra Low-Power AI accelerator IP "Efficiera" Achieved industry-leading power efficiency of 107.8 TOPS/W
- Foxconn Partners with Socionext and Hailo to Launch Next-Generation AI Processing Solution for Video Analytics at the "Edge"
- DMP adopted for NEDO project of "Survey of issues for finding ideas regarding Technology Development for AI Chip and Next-generation Computing for High-efficiency and High-speed Processing"
- Cadence Launches New Tensilica DNA 100 Processor IP Delivering Industry-Leading Performance and Power Efficiency for On-Device AI Applications
- DMP AI Processor IP "ZIA DV700" Enables AI at the Edge
Breaking News
- 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
- CXL Fabless Startup Panmnesia Secures Over $60M in Series A Funding, Aiming to Lead the CXL Switch Silicon Chip and CXL IP
- Cadence Unveils Arm-Based System Chiplet
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 |