Wave’s Compute Appliance is Redefining How Machine Learning is Done

  • 2.9 PetaOps per second of performance

  • More than 2TB of high-speed memory

  • Up to 256,000 processing elements per appliance

  • Scales up to four appliances per data center node

  • Initially supporting TensorFlow

 
 
 

See how Wave Computing’s native dataflow architecture eliminates the performance and scalability bottlenecks of CPUs and GPUs.

 
 

Find out if you qualify for one of the limited
positions in Wave’s Early Access Program.