Microsoft announced a new AI-based offering called Project Brainwave, designed to speed up the real-time artificial intelligence calculations.
The Project Brainwave is a hardware architecture deployed on Intel FPGAs (field programmable gate arrays). These FPGAs help in reducing the latency and lag time in making AI calculations.
It is expected to be perfect for the demands of AI computing. The users will be able to remap the hardware design to the FPGA after every improvement. This capability will be helpful for users to keep pace with new discoveries and rapidly changing AI algorithms.
Customers can reprogram the FPGAs to respond to new advances in AI. The advantage of programmable hardware is that it will be easier to reprogram and add new innovations that come along the time. The traditional systems require users to update the entire hardware, which increases cost and time.
Brainwave works on TensorFlow, the widely used framework for doing AI calculations using neural networks. Furthermore, the tech giant is working on adding the capability to support Microsoft Cognitive Toolkit, which is another popular framework for deep learning.
Microsoft said that the Project Brainwave is integrated with Azure Machine Learning to make Azure a more efficient cloud platform for AI.
The new project will also be useful for customers to do ultra-fast image recognition for several applications like scanning images and flagging false positives.
“Project Brainwave marks the start of Microsoft’s efforts to bring the power of FPGAs to customers for a variety of purposes,” said Mark Russinovich, chief technical officer for Microsoft’s Azure cloud computing platform. “I think this is the first step in making the FPGAs more of a general-purpose platform for customers.”
Announced at Microsoft Build developers conference in Seattle, the Project Brainwave is now available for preview on Azure and on the edge.