Artificial intelligence and machine learning are slowly taking over the world. The applications are numerous. From energy optimization in data centers to taking better pictures, the artificial intelligence of today is the electricity of tomorrow.
The most exciting progress is happening in the artificial neural network field. Artificial neural network algorithms try to imitate the human brain. But computers don’t work like the brain, we have to build specialized hardware for it.
Nvidia has a graphic card series specialized in parallel computation while Google has done something similar with their TPUs.
In 2017, Apple released the A11 chip that has a “Neural Engine”. The Neural Engine helps to use neural networks and machine learning in a more energy-efficient manner. This has a tremendous advantage over using a CPU or a GPU (graphic card).
What is Apple Neural Engine?
A Neural Engine is hardware, a piece of metal, that is responsible for accelerating and optimizing machine learning and neural network tasks for speed and energy efficiency.
It can be used to accelerate video analysis, voice recognition, and image processing. It should be also great to run machine learning models that have been trained previously.
Can third-party apps use the neural engine?
At first, the neural engine was used for Apple’s internal applications like face id and animated emojis (Animoji). Apple has mentioned that third-party apps could use the neural engine but initially, this technology was mainly used for Apple’s proprietary applications.
Core ML, an apple framework for machine learning, can now support the use of Neural Engine when possible.
Is Apple Neural Engine faster than the GPU and CPU?
Yes! It is faster and more energy-efficient. But the real advantage is in a further separation of concerns.
Using the Apple Neural Engine on resource-intensive tasks, like real-time video analysis, will leave the iPhone cold and responsive. Additionally, the battery won’t be drained as quickly as using the GPU or…