Apple M4
System on a Chip (SoC) designed by Apple Inc. / From Wikipedia, the free encyclopedia
Apple M4 is an ARM-based system on a chip (SoC) designed by Apple Inc., part of the Apple silicon series, including a central processing unit (CPU), a graphics processing unit (GPU), a neural processing unit (NPU), and a digital signal processor (DSP). It was introduced in May 2024 for the iPad Pro (M4), and is the fourth generation of the M series Apple Silicon architecture, succeeding the Apple M3.[1][2][3]
General information | |
---|---|
Launched | May 15, 2024; 16 days ago (2024-05-15) |
Designed by | Apple |
Common manufacturer(s) | |
Architecture and classification | |
Application | tablet (iPad Pro) |
Technology node | TSMC N3E |
Physical specifications | |
Transistors |
|
Cores |
|
Memory (RAM) |
|
GPU(s) | Apple-designed integrated graphics (10 core) |
History | |
Predecessor(s) | Apple M3 |
According to Apple, the M4 SoC's Neural Engine is capable of 38 trillion operations per second, "which is 60x faster than Apple’s first Neural Engine in the A11 Bionic chip." The SoC is built upon a 3-nanometer process, and features 3 or 4 performance cores, 6 efficiency cores, and 10 graphics processing cores.[4]
The M4 is the world's fastest consumer SoC in single core performance according to the Geekbench benchmarking suite[5], (beating Apple's M3 Max and Intel's Core i9 desktop CPUs), and rivals Apple's M3 Pro desktop CPU in multi-core performance.[6]
It is the first iPad SoC to support AV1 decode, and has a new display controller that Apple claims is necessary to support the iPad Pro (7th generation)'s Tandem OLED display.[4]
It is rumored the Apple M4 is Apple's first SoC which uses the ARMv9 architecture for its CPU cores, ARMv9.4 to be specific.[7][8] It also could support Arm's SME2 extension,[9] which was announced in 2022[10] and is a superset of SME and SVE2,[11] to accelerate matrix operations. This is likely the cause of the majority of the estimated 7.6% Instructions Per Cycle (IPC) uplift from M3 Max, which would only be 3.0% without the new matrix extensions.[12]