What is a neural net processor?
Neural net processors reduce the requirements for brain-like computer processing from whole networks of computers that excel in complex applications such as AI, machine learning or computer vision down to one multi-cored chip.
Neural network processor for AI
Many implementations of convoluted neural networks (CNN) are currently software-modelled and can span many computers. These current implementations may use many CPUs or, for faster parallel processing, GPUs or even purpose-designed FPGA cards.
Much as computers were reduced from room-sized monstrosities, it is now possible to make processors that function like the human brain in a single package, as in the case of IBM's 4096 core TrueNorth, a single chip that mimics one million human neurons and 256 million synapses. That design can reduce the power requirements for neural net processing down to one-tenth of what had formerly been required.
Other designs, like silicon chip design company Synopsys's, implement embedded vision co-processors along with neural net processors for greater efficiency in computer vision tasks.
For some, the name neural net processor may summon a sense of dread, evoking Arnold Schwarzenegger in his role as a killer cyborg in The Terminator, describing his brain.