IEI introduces AI Vision Acceleration Card – Mustang-F100-A10 which designed by Intel® Arria® 10 1150 GX FPGAs
|
|
|
【Taipei, Taiwan, November 13, 2018】 |
|
IEI Mustang-F100-A10 AI acceleration card empowers the computing in artificial intelligence (AI) deep learning in vision acceleration AI applications, such as machine vision in industrial automation, Medical diagnostics, Smart retail, especially surveillance in traffic, security, and public safety management. Mustang-F100-A10 is designed with Intel® Arria® 10 FPGA and enabled with Open Visual Inference & Neural Network Optimization (OpenVINO™) toolkit. |
Open Visual Inference & Neural Network Optimization (OpenVINO™) toolkit is based on convolutional neural networks (CNN), the toolkit extends workloads across Intel® hardware and maximizes performance. It can optimize pre-trained deep learning model such as Caffe, MXNET, Tensorflow into IR binary file then execute the inference engine across Intel®-hardware heterogeneously such as CPU, GPU, Intel® Movidius™ Myriad X VPU, and Intel® Arria® 10 1150 GX FPGAs. |
 |
With the AIoT era coming, more artificial intelligence (AI) computing requests the better compute capability in shorter time and more complete contour to generate the result in AI deep learning application. FPGA stands for field programmable gate array. More, with its advantage of low power consumption and low latency, it is perfect to be implemented in AI edge computing device with excellent performance, stable quality and longer product life cycle. We are pleased to announce Mustang-F100-A10 with its featuring as a compact form factor, low power consumption, and low latency in FPGA deterministic timing and it is compatible with IEI TANK AIoT Developer Kit, industrial AI edge computer https://www.ieiworld.com/tank-aiot-development-kit/en/ |
 |
|
 |
 |
Hardware |
Software |
• |
Intel® Arria® 1150 GX FPGAs delivering up to 1.5 TFLOPs |
• |
Interface: PCIe Gen3 x 8 |
• |
Form Factor: Standard Half-Height, Half-Length |
• |
Cooling: Active fan |
• |
Operation Temperature:
5°C ~ 60°C (ambient temperature) |
• |
Operation Humidity: 5% to 90% relative humidity |
• |
Power Consumption: < 60W |
• |
Power Connector: 12V external power |
• |
DIP Switch/LED Indicator: Identify card number |
• |
Voltage Regulator and Power Supply:
Intel® Enpirion® Power Solutions |
|
• |
Operating Systems
Ubuntu 16.04.3 long-term support (LTS) 64-bit, CentOS 7.4 64bit (more OS are coming soon) |
• |
OpenVINO™ toolkit
» |
Intel® Deep Learning Deployment Toolkit |
|
• Model Optimizer |
|
• Inference Engine |
» |
Optimized computer vision libraries |
» |
Intel® Media SDK |
» |
OpenCL™ graphics drivers and runtimes.
(OpenCL™ is the trademark of Apple Inc. used by permission by Khronos) |
» |
Current Supported Topologies:
AlexNet, GoogleNet, Tiny Yolo, LeNet, SqueezeNet, VGG16, ResNet (more variants are coming soon) |
» |
Intel® FPGA Deep Learning Acceleration Suite |
|
• |
High flexibility, Mustang-F100-A10 develop on OpenVINO™ toolkit structure which allows trained data such as Caffe, TensorFlow, and MXNet to execute on it after convert to optimized IR |
|
|
|
|
|
|
|
|