Apple A18 Pro Vs A17 Pro: Apple Delivers Meaningful Upgrades

The newly launched iPhone 16 Pro series is powered by Apple’s latest A18 Pro chipset. Apple says the A18 Pro is the most powerful mobile chipset in the world. Indeed, it is built on TSMC’s much-improved N3E process node and packs an updated CPU, but what new upgrades does it bring over its predecessor, the Apple A17 Pro? To find out, we have compared A17 Pro and A18 Pro chipsets, including their CPU, GPU, and Neural Engine performance.

A17 Pro vs A18 Pro: Specs Comparison

A17 Pro vs A18 Pro: CPU

Starting with the CPU, the Apple A18 Pro is built on TSMC’s 2nd-generation 3nm process node, which is N3E. The A18 Pro packs six CPU cores including two performance cores that can reach up to an unprecedented 4.04GHz frequency, a first for a mobile chipset. And the chipset also includes 4x efficiency cores.

The most interesting part about the A18 Pro is that Apple has adopted the Armv9.2 architecture with this chipset. While it doesn’t have SVE (Scalable Vector Extensions) support for high-computing workloads, A18 Pro brings two SME (Scalable Matrix Extension) units for faster performance in AI/ML workloads. Apple is calling it ML accelerators.

As for last year’s A17 Pro, it was built on TSMC’s 1st generation 3nm (N3B) process node with the same six-core CPU setup. The 2x performance cores can reach up to 3.78GHz and the 4x efficiency cores can boost the frequency up to 2.11GHz.

Overall, Apple says the A18 Pro CPU is 15% faster than the A17 Pro CPU. Not only that, the A18 Pro CPU consumes 20% less power than the A17 Pro CPU. The Geekbench score of A18 Pro has appeared and it scores 3,409 in single-core and 8,492 in multi-core. Comparing it with A17 Pro does bring a performance difference of around 15%.

A17 Pro vs A18 Pro: GPU

Moving to GPU, Apple has not made substantial changes to the GPU microarchitecture this year. The 6-core GPU on the Apple A18 Pro is almost the same as last year except that the GPU frequency has been increased. In addition, the faster LPDDR5X memory (7500 MT/s) that offers 17% faster memory speed helps the GPU achieve better graphics performance.

Not to mention, the new aluminum internal frame and thermal substructure will enable the GPU to sustain peak performance for much longer. Overall, Apple says the 6-core GPU on the A18 Pro offers 20% faster graphics performance compared to the A17 Pro 6-core GPU. In addition, hardware-accelerated Ray Tracing is now 2x faster than the A17 Pro.

A17 Pro vs A18 Pro: Neural Engine

As for Neural Engine, Apple says the 16-core Neural Engine on both A18 Pro and A17 Pro can deliver up to 35 TOPS (trillion operations per second). It seems there is no change to the Neural Engine on the iPhone 16 Pro and 16 Pro Max. However, with 17% faster memory, the Neural Engine can deliver faster performance in AI/ML tasks. The new Neural Engine on A18 Pro is capable enough to deliver Apple Intelligence features locally, on the device.

A17 Pro vs A18 Pro: Connectivity

Finally, on the connectivity front, Apple is likely using Qualcomm’s Snapdragon X75 5G modem on the Apple A18 Pro. Last year’s A17 Pro included the Snapdragon X70 5G modem, the same modem found on the Snapdragon 8s Gen 3.

The peak download/upload speed is the same for both modems, however, the X75 modem promises better coverage and efficiency. For local connectivity, the Apple A18 Pro is equipped with Wi-Fi 7 and Bluetooth 5.3. The A17 Pro supports Wi-Fi 6E and Bluetooth 5.3.

A17 Pro vs A18 Pro: Closing Thoughts

To sum up, the latest A18 Pro chipset offers meaningful upgrades over the year-old A17 Pro in the CPU and GPU departments. With the adoption of Armv9.2 architecture and the addition of SME units in the CPU, AI/ML performance would be much faster. Most importantly, Apple has worked on improving the thermal performance by adding a thermal graphite substructure that allows the CPU and GPU to sustain peak performance for a longer period.

So, as a whole package, the iPhone 16 Pro series, powered by the A18 Pro chipset, turns out to be a substantial upgrade as you get faster performance without compromising on efficiency. However, to reach a definitive conclusion, we need to run various benchmarks and measure performance. Stay tuned for future updates!