Apple Silicon has made waves when it comes to SoCs (Systems on a Chip). It has created a standard for power efficiency and performance that the industry lacked. Now, with its fourth generation, the Apple M4 is looking to create an even larger gap with improved GPU architecture and the fastest Neural Engine in the company’s repertoire.
The Apple M4 chip is built using the company’s second generation 3nm architecture. It comes with a 10-core CPU, a 10-core GPU and the fastest Neural engine that Apple has created yet. The M4 consists of 28 billion transistors built with 3nm technology that allows for better power efficiency.
The 10-core CPU consists of four performance cores and six efficiency cores that help forge a delicate balance between performance and power efficiency. These cores are able to process higher workloads and better predict branches and have a wider decode and execution engines. Also featured on both cores are next-generation Machine Learning (ML) accelerators. Thanks to this, the M4 is able to deliver 1.5x the performance of the M2.
The M4’s 10-core GPU is built on the next-generation graphics architecture that was first featured in the M3. This architecture gives a significant boost to performance through Dynamic Caching. This process allocates memory dynamically in hardware and in real time resulting in an substantial increase in GPU utilisation. The new GPU also enables hardware-accelerated ray tracing, a first on the iPad line up. This enables more realistic shadows and reflections in games like Diablo Immortal as well as other graphically rich experiences. The GPU is also able to better process and reproduce graphically intense scene and workloads thank to hardware-accelerated mesh shading. The performance of the GPU is increased by up to four times on apps like Octane.
The Neural Engine on the Apple M4 is able to process 38 trillion operations a second. That’s 60 times faster than the A11 Bionic which featured the first generation Neural Engine. This combined with the ML accelerators, the CPU and the GPU make this processor Apple’s best processor for AI workloads. This is utilised in iPadOS with features like Live Captions which can generate captions in real time or visual look up that help identify items in videos or pictures.
The M4 also brings hardware acceleration for Apple’s AV1 code to the iPad for the first time. The media engine of the M4 also supports codecs like H.264, HEVC and ProRes.
