New Features: At the recent Microsoft Build conference, Microsoft introduced the Project Brainwave-based Azure machine learning hardware acceleration model and integrated it with the Microsoft Azure Machine Learning SDK for preview. Customers can use Azure’s massively deployed Intel® FPGA (Field Programmable Gate Array) technology to provide their models with industry-leading artificial intelligence (AI) inference capabilities.
“As a holistic technology provider, we support artificial intelligence by working closely with Microsoft. Artificial intelligence applies to everything from training to inference, from speech recognition to image analysis. Intel has a wide range of hardware, software and tools. Combined to meet the needs of these workloads.”
Daniel McNamara, Vice President of Intel Corporation and General Manager of Programmable Solutions Group
Significance: Data scientists and developers can easily use Deep Neural Networks (DNN) to handle a variety of real-time workloads in the world’s largest accelerated computing cloud, covering applications in manufacturing, retail, and medical fields. Whether in cloud computing or edge computing, they can use Intel FPGAs to train models and deploy them to Project Brainwave.
Importance: Project Brainwave uses Intel FPGAs to provide real-time artificial intelligence, unleashing the potential of programmable hardware and demonstrating the future of artificial intelligence. The FPGA-based architecture is characterized by high throughput and economical efficiency: For example, ResNet 50 can be run—an industry-standard DNN that requires nearly 8 billion calculations—without the need for batch processing. AI customers do not need to choose between high performance and low cost.
How it works: By using the Azure Machine Learning SDK for Python, customers can retrain ResNet 50-based models and their data to specifically handle image recognition tasks.
For real-time AI workloads, dedicated hardware accelerators are required due to the high computational power. Intel FPGAs support Azure to configure hardware for tasks to achieve peak performance.
Doug Burger, a Distinguished Engineer at Microsoft, said: “Customers can now use Intel FPGAs and Intel Xeon technologies to use Microsoft’s range of AI breakthrough technologies in the cloud and on the edge. These new capabilities support the integration of AI into real-time processing processes to leverage The power of Microsoft Azure and Microsoft AI drives business transformation.” Users can further refine the FPGA or completely change its purpose according to the specific requirements of the Azure workload. The Azure architecture developed with Intel FPGAs and Intel® Xeon® processors supports innovation through accelerated AI, based on the user’s customized software and hardware configuration provisions. Customers can access Project Brainwave Public Preview. A limited preview of Project Brainwave is now available, allowing users to access Microsoft designed Intel FPGA-based systems that act as Azure IoT Edge appliances and connect to the Azure IoT Hub. Disclaimer This newsletter is for reference purposes only. Intel does not provide any express or implied warranties in this newsletter. © 2018 Intel Corporation. Intel and the Intel logo are trademarks of Intel Corporation in the United States and/or other countries. Other names and brands may be the assets of other owners.
