On October 15, Apple officially unveiled its next-generation M5 chip, built on TSMC's advanced N3P process. The M5 packs a 10-core CPU and a 10-core GPU, both gaining two additional cores over the previous M-series entry-level processors. Each GPU core now integrates a neural accelerator, designed to boost AI workloads significantly.
"The M5 marks a major leap forward in Apple silicon's AI capabilities," said Johny Srouji, Apple's Senior Vice President of Hardware Technologies. "With neural accelerators built directly into the GPU, a faster neural engine, the world's fastest CPU cores, and increased unified memory bandwidth, M5 delivers remarkable performance for MacBook Pro, iPad Pro, and Apple Vision Pro."
In detail, the 10-core CPU combines six performance cores and four efficiency cores, mirroring the structure of the M4. Despite the same configuration, Apple claims a 15% boost in multi-threaded performance compared with its predecessor.
The 10-core GPU delivers up to 30% faster graphics performance than the M4, while integrated neural accelerators enhance AI computation up to four times. It also introduces third-generation hardware-accelerated ray tracing, with Apple stating that ray tracing in supported apps now runs 45% faster.
Meanwhile, the 16-core Neural Processing Unit (NPU) remains, offering powerful AI performance with minimal power consumption. The unified memory bandwidth has been upgraded to 153 GB/s, a 30% increase over the M4's 120 GB/s.
Apple highlighted that the NPU works seamlessly with the CPU and GPU neural accelerators to fully optimize AI-driven tasks. For example, users can transform 2D photos into spatial scenes or generate Personas faster in Apple Vision Pro. Tools like Apple Intelligence and Image Playground also benefit from these performance upgrades.
While the M5 Pro and M5 Max chips weren't launched alongside the base M5, references found in macOS "Tahoe" suggest both high-end versions are in development and expected to debut later.