Neural network every day are becoming more common. But their work requires very much energy. And if a couple of years ago, someone said that the neural network will be able to work in a smartphone, many would have thought it was a joke. However, thanks to the efforts of experts from the Massachusetts Institute of technology (MIT) came to light AI microprocessor for portable gadgets.
According to Engadget, the experts at MIT have developed a processor, the power consumption of which 95% less than existing analogues with extremely high performance. Thanks to a new processor you can use a resource – and energy-consuming applications such as neural networks even on handheld devices.
In addition to reduced energy consumption, faster microprocessor from MIT used smartphonetoday today 3-7 times. This was achieved thanks to the new architecture using the scalar product, allowing to calculate the right connection for the whole data set without sending intermediate results to RAM and back. About the new chip even supported by the President of IBM’s Dario Gil:
“The new chip uses a very effective energy approach. This is a new word in the implementation of convolutional operations with huge amounts of memory. This will enable you to create more complex and precise neural networks for wearable devices Internet of things”.
What do you think about such a rapid development of neural networks? Voice your opinion in our new telegram chat.
The new processor will enable to run the neural network on smartphones
Vladimir Kuznetsov