Wave is Building a Family of Deep Learning Computers that are Faster  than GPUs with Better  Energy Efficiency than FPGAs


  • Powered by an advanced Dataflow Processing Unit (DPU) architecture with tens of thousands of interconnected processing elements.


  • Designed to support massive amounts of internal memory and external memory bandwidth.


  • Purpose-built, DPU-based hardware and software supports TensorFlow and other dataflow-based computing frameworks for Deep Neural Networks.


  • Real-time, re-programmable dataflow architecture that is tuned to accelerate data driven processing on a massive scale.


  • Committed to helping you scale your computing to match the scale of your growing data.