Wave Computing to donate training scheme to TensorFlow

March 22, 2018 // By Peter Clarke
Wave Computing Inc. (Campbell, Calif.), a company with a history in asynchronous logic that is now focused on deep learning, has said it will contribute variable fixed-point training technology called Versipoint to the TensorFlow development environment.

TensorFlow is a symbolic math library developed by Google for dataflow programming and used for machine learning applications such as neural networks.

Wave claims that Versipoint speeds up the training of neural networks without the need for floating-point hardware such as is found in GPUs. Versipoint is part of Wave Computing's dataflow acceleration systems.

Versipoint is used to reduce storage and processing requirements and allows for changes to occur dynamically at runtime. It includes runtime software, schedulers, pre-developed and pre-compiled software  and ships as part of a Wave computer for either on-premise or data center environments.

Related links and articles:


News articles:

Wave Computing offers machine learning platform

Google's second TPU processor comes out

Machine learning interview with Jem Davies of ARM

Vous êtes certain ?

Si vous désactivez les cookies, vous ne pouvez plus naviguer sur le site.

Vous allez être rediriger vers Google.