- Apple has announced its latest M5 chipset with the next-gen GPU that brings a Neural Accelerator into each GPU core.
- The Apple M5 features a 10-core GPU, up to 10 CPU cores, a 16-core Neural Engine, and up to 32GB unified memory.
- The M5 unified memory can go up to 153 GB/s to quickly process AI workloads.
Apple has introduced its latest M5 chip and it’s powering the new 14-inch MacBook Pro, iPad Pro, and Vision Pro. The new M5 silicon is built on TSMC’s 3rd-gen 3nm process node. The Apple M5 chip comes with up to 10 CPU cores, 10 GPU cores, and a 16-core Neural Engine. The unified memory (up to 32GB) bandwidth is 153 GB/s on the M5 — a 30% increase over the M4.
Apple Adds a Neural Accelerator in Each GPU Core
The most important bit is that Apple has included a dedicated Neural Accelerator (like Nvidia Tensor cores) in each GPU core, like we saw on this year’s A19 chipsets. This results in over 4x faster GPU compute performance for AI workloads, compared to the previous Apple M4 chipset. On the graphics front, the 10-core M5 GPU delivers 30% faster performance vs the Apple M4.
The new M5 GPU brings third-generation ray tracing and second-generation dynamic caching for smoother gameplay and realistic gameplay. In ray-traced games, the graphics performance sees up to 45% improvement. Apple has also announced seamless integration with Core ML, Metal 4, Metal Performance Shaders, and Tensor APIs for Neural Accelerator programming.
Moving to the Apple M5 CPU, the 10-core CPU features 4 performance cores and 6 efficiency cores. Apple says the M5 has the “world’s fastest performance core,” and delivers 15% faster multithreaded CPU performance over the M4. The 16-core Neural Engine has seen performance improvements and it’s more power efficient than before.
Apple is debuting the M5 on the new MacBook Pro, iPad Pro, and Vision Pro. You can pre-order them starting today.
