Apple’s new M4 chip is set to be integrated into the upcoming iPad Pro models, which will feature OLED displays. The focus of this advanced processor is to enhance the performance of tasks related to artificial intelligence in iPados.
During a presentation, Apple highlighted the capabilities of the M4 chip in artificial intelligence. With a new neural engine, the M4 can handle up to 38 trillion operations per second, marking a significant improvement compared to the NPU of the 2017 A11 chip, which was 60 times slower. Despite falling short of the 45 trillion operations of the Snapdragon X Elite, Apple claims that the M4 offers similar performance to the latest PC chips while consuming only a quarter of the energy.
The M4 chip, manufactured using 3rd-generation 3 Nm technology, boasts a 10-core processor with four high-performance cores and six energy-efficient cores, making it 50% more powerful than its predecessor, the M2, found in the previous iPad Pro models. The 10-core graphics system in the M4 supports dynamic caching, shadow grid, and ray tracing, with the new graphics engine boasting a fourfold increase in rendering speed compared to the M2.
Although the exact number of NPU cores in the M4 was not disclosed, previous chips like the M1 and M2 ULTRA series featured 16 NPU cores, with performance improvements seen in each generation. With the M4 now offering a significant leap from 18 TOPS in the M3 to 38 TOPS, it is evident that Apple is making strides in this area. However, the total TOPS value may not always accurately reflect the actual efficiency of AI tasks due to their varying complexities.
The announcement of the M4 chip comes just before a special Microsoft event in Seattle on May 20, focusing on Windows and Surface AI. It is anticipated that Microsoft will unveil new ARM-based devices, with the company confident in Windows on ARM’s ability to outperform Apple’s MacBook Air M3 laptops in terms of CPU performance and tasks accelerated by the new Snapdragon X Elite processors from Qualcomm, offering up to 45 TOPS of NPU productivity.