English
You are here:
Processor IP for AI vision
The DesignWare® ARC® EV71, EV72, and EV74 Embedded Vision Processor IP provides high performance, low power, area efficient solutions for a standalone computer vision and/or AI algorithms engine or as an accelerator for vision-enabled SoCs. The EV7x family is designed for power sensitive vision applications ranging from low power facial recognition, to augmented reality, to high-performance automotive autonomous vehicle vision. The EV7x family integrates a high-performance 32-bit scalar core, a 512-bit vector DSP, an optional vector floating point unit and an optional deep neural network (DNN) engine for complete computer vision and deep learning algorithm (CNN/RNN) coverage to perform fast and accurate object detection, classification, and scene segmentation. An optional AES-XTS encryption engine protects sensitive coefficient and graph topology data as well as user’s biometric data.
Features
- Optimized for high frame-rate and video resolution embedded vision and imaging applications Integrates 32-bit scalar core, 512-bit vector DSP, and optional DNN accelerator
- DNN accelerator delivers up to 35 TOPS performance
- ASIL B or D Ready EV7xFS products integrate safety-critical hardware features
- IEEE 754-compliant vector floating point unit option offers single or half precision operations and advanced math functions
- AES encryption option protects valuable data such as training and sensor data
- High productivity MetaWare EV Development Toolkit supports OpenCV, OpenVX and OpenCL C, Tensorflow, Caffe, and ONNX standards
Deliverables
- The DesignWare EV71, EV72, and EV74 Processors and optional DNN accelerator are delivered as Verilog HDL in the ARChitect IP Library.
- The HDL is configured and output from the ARChitect IP Configurator tool. To test that the product performs as expected, a basic testbench of Customer Confidence Tests (CCT) is included.
Applications
- Automotive driver assistance systems (ADAS) and autonomous driving
- Smart home / IoT
- Augmented/mixed reality
- Robotics
- Surveillance
- Drones
- Facial recognition payment
- Multi-function printers
- Digital still cameras
Video Demo of the Processor IP for AI vision
Synopsys has announced the EV7x Embedded Vision Processor IP family with deep neural network (DNN) accelerator offering performance up to 35 TOPS. The EV7x can scale from 880 to 14,080 MACs with improved utilization, performance, bandwidth, and power over EV6x. The EV7x also offers vision engine enhancements to improve SLAM and DSP performance, optional AES encryption for security and protection of coefficients and image data, and improved LSTM and RNN capabilities. Learn more at www.synopsys.com/ev7x
View EV74 processor IP for AI vision applications with 4 vector processing units full description to...
- see the entire EV74 processor IP for AI vision applications with 4 vector processing units datasheet
- get in contact with EV74 processor IP for AI vision applications with 4 vector processing units Supplier
vision processor
- Edge AI Processor Architecture for Imaging & Computer Vision
- Intelligent Vision Processor
- Imaging and Computer Vision Processor
- Neural Network Processor for Intelligent Vision, Voice, Natural Language Processing
- ARC EV74FS safety processor for vision applications with four vector processing units, ASIL B/C/D
- EV72 processor for embedded vision applications with two vector processing units