Wave Computing’s dataflow architecture is the fundamental technology behind the company’s machine learning compute appliance. It is built upon a revolutionary dataflow computing technology that eliminates the traditional CPU/GPU co-processor structure and associated performance and scalability bottlenecks. This allows Wave Computing’s solutions to exploit data and model parallelisms present in deep learning models, such as convolutional and recurrent neural networks.

Wave Computing’s dataflow systems utilize Dataflow Processing Units (DPUs), which contain thousands of interconnected dataflow Processing Elements (PEs). The performance and scalability of these systems make them ideal for organizations using machine learning to easily develop, test, and deploy deep learning models for frameworks such as TensorFlow. To help data scientists speed time to results, the Wave Computing systems include complete software solutions and stacks: the WaveFlow SDK, the WaveFlow Agent Library, WaveFlow Execution Engine, and the Wave Machine Learning Framework Interface.

Read the Whitepaper