Neuromorphic computing is yet another branch of computer science, attempting to mimic the brains architecture in a digital format, to achieve computation, and fundamentally alter how computation is done. Guided by brain-like ‘spiking’, a Neuromorphic computer draws from the biological capabilities of computing data, and is a fundamentally different way to architect our future computers. The interdisciplinary field, began with the implementation of silicon circuits to mimic biological architectures, but has spawned into its own complex field of designing the architecture.
Designing a Neuromorphic computer has been the greatest challenge yet, as the complexity of biology cannot be fully mirrored into our silicon world. It is for instance a decision that needs to be made by whomever is designing a Neuromorphic computer in how deep they wish to dive into each part of the computer. Should the scope for example be to mimic a neuron as exact as possible? Should the complexity be even lower, such as on the chemical level, or should we forego the inner workings of each neuron and look at the bigger picture, and focus on ‘spiking’ as the primary computational focus?
Intel seems to have made a choice, after announcing their latest Neuromorphic computer, dubbed Pohoiki Springs, based on their Loihi architecture as they call it. A much more detailed article about the latest Intel release can be found here
Leave a Reply
You must be logged in to post a comment.