The A13 Fusion of Apple is faster and consumes less energy

Revealed on stage at Apple 's iPhone 11 event, the A13 would have the fastest processor and graphics processor of any smartphone or tablet, thus enhancing the power offered by the Apple iPhone 11. A12 used in the 2018 models. The processor and graphics processor of the 64-bit chip would be 20% faster than the variants of the A12, with a variety of elements allowing it to perform more than thousand billion operations per second, some related to machine learning.

Learning power and machine

The A13 is composed of many sections, but the three main ones are the processor, the graphics processor and the neuron engine. The central unit consists of two performance cores and four cores of efficiency, each being used depending on the workload. The GPU contains four optimized cores for the metal, while the neural engine contains eight others.

A pair of machine learning accelerators, used to perform matrix multiplication, a calculation frequently used in machine learning, is also buried. Apple says the A13 will perform this calculation six times faster than the A12 Fusion. It is these accelerators that allow the processor to reach the milestone of billions of operations.

A selection of areas improved by Apple in the Bionic A13

Due to the load balancing of the learning controller designed by Apple, machine learning models can be scheduled on the processor, graphics processor and neural engine depending on what would offer the best performances. The controller also does this while balancing the need to stay as efficient as possible, which helps to reduce the amount of energy used.

Because the controller makes any decision, at any time, away from the developer to process machine learning models, it also simplifies the development process.

Reduced power and longer battery life of the iPhone

In addition to providing more processing power than the A12, A13 Bionic also strives to reduce the amount of energy needed to perform the calculations in the first place. For this generation, this has helped Apple to achieve several hours of battery life in the iPhone 11 Pro, instead of the usual improvements of an hour.

Part of the savings comes from changes in the way it produces chips in the first place. Taking advantage of its partner's latest business processes for creating 7-nanometer chips, described as an "improved 7nm 2nd generation enhanced transistor," Apple has adapted each transistor to its performance and power.

Meanwhile, work has led Apple to compress 8.5 billion transistors on the A13, compared with 6.9 billion on the A12.

In addition to the transistors themselves, while being more selective compared to what is used to perform calculations, Apple has also been working to improve the architecture.

The processor, graphics processor and neural engine are all more powerful, but energy efficient

The use of hundreds of voltage areas on the chip gives Apple more control over the type of power used and when. By activating only sections at a time when they will be used for processing, while leaving unused areas without electricity, the amount of energy used in a calculation is drastically reduced to the bare necessities.

At an even smaller level, using hundreds of thousands of smaller domains allows for granular control of what is being powered, ensuring that only the smallest amount of logic in the chip is used for a process.

The work has resulted in vast energy savings. The two processor performance cores consume 30% less power, the four-core efficiency saves 40%, the four-core graphics processor also saves 40% and the eight-core Neural Engine enable efficiency increased energy by 15%.