Wave Computing Announces Early Access Program For Its Super-Fast Machine Learning Appliance

[Lucian Armasu]

An Early Access Program is now available for Wave Computing's machine learning solution, which is a compute appliance that can deliver high performance and scalability for training neural networks.

Wave Computing Announces Early Access Program For Its Super-Fast Machine Learning Appliance : Read more

 

[genz]

Funny enough, this reads like the APU theory AMD was jumping on (combine CPU/GPU to work together with low latency, just like their core) then the HBM theory AMD was jumping on (give everything massive throughput to the same memory space so that they don't have to swap memory from GPU space to CPU space) then the Infinity Fabric thing AMD was jumping on (connect everything with fat pipes in a web instead of a ring bus or other bottlenecked system).

Just an observation. Cool case too.