In earlier benchmark tests, the A17Pro's single-core and multi-core results were unsatisfactory compared to the A16Bionic (an improvement of only about 10%), but in the latest Metal GPU running score data leak, Apple's flagship SoC, which powers the iPhone 15 Pro and iPhone 15 Pro Max, successfully achieved better graphics processing results. According to the latest data, the new chip's performance is 20% ahead of its predecessor, most likely due to the addition of GPU cores.
access:
Apple Online Store (China)
A new metal GPU leak appeared on Geekbench6, which was tested on an iPhone 16,2 and scored 27158 points. In comparison, the A16 Bionic graphics card used in last year's iPhone 14 Pro and iPhone 14 Pro Max and this year's iPhone 15 and iPhone 15 Plus is 20% slower, but it is worth noting that it uses a 5-core GPU instead of a 6-core GPU.
In fact, if the A17 Pro were powered by the same number of GPU cores, its performance results would probably be just as disappointing as the single-core and multi-core results. According to research by Vadim Yuryev (who runs the popular YouTube tech channel MaxTech), the A17Pro has a single-core performance of 4526 points, while the A16Bionic has a single-core performance of 4455 points. Doing the math, the performance improvement must be attributed to the additional GPU.
Overall, the A17 Pro's graphics scores are in line with the numbers published in Apple's marketing materials. However, the A17Pro’s performance in ray tracing is significantly better than its predecessor. During Apple's keynote, the company mentioned that its latest SoC supports hardware-accelerated ray tracing, which is four times faster than the A16 Bionic.
However, it should be pointed out that A16Bionic does not support hardware-accelerated ray tracing and can only rely on software to improve visual effects, which resulted in significant frame drops when Apple showed performance demonstrations. Although the A17 Pro is mass-produced using TSMC's most advanced 3nm process, it should perform better in both categories, which shows that you can't always rely on improving the manufacturing process to get immediate performance gains.