Utilization of CPU, GPU and FPGAs for deep learning. I was pleased to see this article from a research team from the University of California Santa Barbara. At Viviota working with a large university in Switzerland we have built a similar operating environment to take advantage of 400 FPGA's to do massive mathematical computations in less then 50 microseconds. This is allowing researchers to study the interaction of sound waves with solid objects. The article below points to this new trend. Is anyone else seeing this trend?