Wave Computing to Contribute its Versipoint Technology For Use Across the Deep Learning Industry

 

CAMPBELL, Calif., March 21, 2018 -- Wave Computing®, the Silicon Valley company that is revolutionizing artificial intelligence (AI) and deep learning with its dataflow-based solutions, announced today that it will contribute its variable fixed-point training technology, called Versipoint™, to the Tensorflow open source development community. Wave’s Versipoint technology enables data scientists to quickly train neural networks without the need for energy-consuming floating point hardware, such as in a GPU. Developed by Wave’s leading deep learning team and implemented as part of the company’s dataflow acceleration systems, Versipoint helps Wave’s systems deliver greater compute efficiency and better overall deep learning performance.

Dr. Debajyoti Pal is Wave’s Vice President of Machine Learning and an IEEE Fellow who is renowned in the fields of adaptive signal processing, digital communications and control theory. Dr. Pal commented, “Wave’s Versipoint technology is another example of how we are advancing the forefront of deep learning while giving back to the AI community. We believe that broadened access to our more efficient fixed point training and inferencing technology will help developers better commercialize their multi-layer, deep learning models across a larger class of AI applications.”

“Wave Computing's generous offer to release its Versipoint technology to the TensorFlow community will bring more efficient machine learning training to a wider group of researchers and practitioners,” said Kevin Krewell, Principal Analyst at TIRIAS Research.

  Sample results of Wave’s Versipoint techniques for training a deep neural network. These show more efficient training using Wave’s 16-bit Versipoint technology compared to 32-bit floating point technology in the same amount of training epochs.

Sample results of Wave’s Versipoint techniques for training a deep neural network. These show more efficient training using Wave’s 16-bit Versipoint technology compared to 32-bit floating point technology in the same amount of training epochs.

How Wave Computing’s Versipoint Technology Works

Wave’s Versipoint technology is a computational framework that is powerful, yet flexible enough to support any type of deep neural network. Versipoint allows more efficient, non-floating point hardware to be used for training neural networks by reducing storage and processing requirements by 2x or more. Enabled by the programmable and reconfigurable nature of Wave’s dataflow architecture, Versipoint technology allows for a wide variety of mode changes to occur dynamically at runtime.

Versipoint technology is part of Wave’s complete system solution that includes runtime software, schedulers, pre-developed and pre-compiled software modules and more, all of which is “under the hood” and ships as part of a Wave Computing AI system for either on-premise or data center environments.

 

About Wave Computing

Wave Computing, Inc. is the Silicon Valley company that is revolutionizing artificial intelligence and deep learning with its dataflow-based solutions. The company’s vision is to “follow the data” and bring deep learning to customers’ data wherever it may be—from the datacenter to the edge of the cloud. Offering its solutions to customers globally, Wave Computing has been named Frost & Sullivan’s 2018 “Machine Learning Industry Technology Innovation Leader,” and has been recognized by CIO Application Magazine’s as one of the “Top 25 Artificial Intelligence Providers.”
 


Wave Computing, the Wave Computing logo and Versipoint are trademarks of Wave Computing, Inc. TensorFlow, the TensorFlow logo and any related marks are trademarks of Google Inc. All other trademarks are used for identification purposes only and are the property of their respective owners. ©2018 Wave Computing, Inc.  All rights reserved.