Apple announced a suite of four new iPhones today, and with them, new processors to power the devices. Additionally, the company also has new networking and cellular chips.

The phones include the iPhone 17, iPhone Air (a new 5.6-mm thin design), iPhone 17 Pro, and iPhone 17 Pro Max.

The base iPhone 17 will be powered by the A19 system on a chip, while the rest of the lineup relies on the more powerful A19 Pro. These chips are likely built on TSMC's latest N3P process node, which is also expected to be used for Apple's upcoming M5 chips for iPads and Macs.

The A19 boasts six CPU cores and five GPU cores. The processor has four efficiency cores and two performance cores, while the GPU continues advancements in hardware-accelerated ray tracing, mesh shading, and MetalFX upscaling.

The A19 Pro will power the new iPhone Air alongside the iPhone 17 Pro line. Like the A19, its six-core CPU utilizes two performance and four efficiency cores. The performance cores have better branch prediction and increased front-end bandwidth, while the new efficiency cores have a 50% increase in last-level cache.

In terms of raw numbers, Google says the CPU inside the Tensor G5 delivers a 34% speed increase compared to the Tensor G4. For AI-driven tasks, the improvements are even more substantial, with a 60% boost in performance. However, it's worth noting that this claim comes from Google, and there's no information provided about the data formats used.

It appears the chip uses TSMC's N3P (a performance-enhanced 3nm-class process technology) fabrication process, the most advanced foundry technology that currently exists. The motivation for Google's move may be that TSMC has consistently delivered stronger yields and more refined transistor designs at 3nm-class, enabling higher performance at lower power.

TSMC's N3P manufacturing process is an optical shrink of the N3E platform, offering tangible gains in performance, efficiency, and density while retaining full design compatibility, which is not the case for Google's IP.

In terms of performance, N3P provides about a 5% boost in speed at the same power levels, or alternatively a 5–10% reduction in energy use when maintaining the same clock rates. This balance gives designers a lot of flexibility in terms of either pushing for higher performance or optimizing for longer battery life and reduced thermal output, depending on product goals, which was perhaps Google's goal. The chip designer has likely achieved success, as the G5 powering the Pixel 10 is now in mass production.

By: DocMemory
Copyright © 2023 CST, Inc. All Rights Reserved