Ever since Apple started to develop its own smartphone processors, it has consistently offered the fastest system-on-chips for handsets, and more recently, these SoCs have even challenged CPUs for PCs when it comes to benchmark scores. The new six-core Apple A19 Pro does just that: it beats its predecessor, it leaves no chances for its arch-rival Snapdragon 8 Elite, and even conquers desktop-grade CPUs in the single-thread Geekbench 6 benchmark. In addition, the processor seems to pack the highest-performing smartphone GPU, offering performance comparable to that of GPUs for client PCs and tablets.
11% - 12% higher CPU performance
| Row 0 - Cell 0 | A19 Pro | A18 Pro | A17 Pro | A16 Bionic | Snapdragon 8 Elite | Snapdragon 8 Gen 3 |
General specifications | 2P+4E, up to 4.26 GHz | 2P+4E, up to 4.0 GHz | 2P+4E, up to 3.77 GHz | 2P+4E, up to 3.46 GHz | 2P+6E, up to 4.47 GHz | 5P+3E, up to 3.01 GHz |
Single-Thread | 3895 | 3505 | 2950 | 2641 | 2862 | 1959 |
Multi-Thread | 9746 | 8658 | 7279 | 6989 | 9481 | 4989 |
The latest A19 Pro smartphone CPU from Apple scores 3,895 points in single-thread Geekbench 6 tests, outpacing its predecessor by 11% and Qualcomm's Snapdragon 8 Elite by 36%. In addition, the new chip leaves behind all stock processors for client devices, including Apple's own M4 (by 5.3%) and AMD's mighty Ryzen 9 9950X (by 11.8%). Given Apple's focus on performance efficiency, it is not surprising that the new SoC beats everything in single-thread workloads.
The new A19 Pro application processor also scores 9,746 points in the multi-thread Geekbench 6 test, which is 12% higher compared to A18 Pro. However, the smartphone SoC still cannot beat CPUs for desktops and notebooks in multi-thread workloads, which is not surprising.
While an 11% - 12% generation-to-generation performance increase looks fairly solid, it is lower compared to the improvements of the A18 Pro compared to the A17 Pro (circa 18%).
| Row 0 - Cell 0 | A19 Pro | M4 | M3 | Ryzen 9 9950X | Core i9-14900KS |
General specifications | 2P+4E, up to 4.26 GHz | 4P+6E, up to 4.40 GHz | 4P+4E, up to 4.05 GHz | 16P/32T, 4.30 GHz - 5.75 GHz | 8P+16E/32T, 3.20 GHz - 6.0 GHz |
Single-Thread | 3895 | 3697 | 3076 | 3482 | 3362 |
Multi-Thread | 9746 | 13778 | 11863 | 23584 | 23445 |
Apple's A19 Pro SoC features two high-performance cores operating at up to 4.26 GHz (+6.5%) and featuring improved branch prediction (higher performance in branch-heavy workloads and better power efficiency) and increased front-end bandwidth (which points to higher instructions-per-cycle, but does not indicate how many instructions the core can decode per cycle) as well as four energy-efficient cores that now boast with a 50% larger last level cache compared to the predecessor.
The A19 Pro processor is made by TSMC on its N3P fabrication process, which is an optical shrink of N3E that enables a 4% higher transistor density as well as a 5% performance increase at the same power or a 5% - 10% power consumption reduction at the same frequency compared to N3E.
Given the capabilities of the fabrication process, a 6.5% clock speed boost looks quite solid. The CPU also has some microarchitectural improvements, so its performance advantages over its predecessor go beyond the frequency improvement. However, given the fact that Apple uses a vapor chamber cooling system and an aluminum unibody chassis for its iPhone 19 Pro, it is surprising that the company did not increase CPU clocks more significantly to get higher peak performance. Perhaps the company decided to focus on workloads that are branch-heavy and/or benefit from higher IPC more than from sole frequency; it looks like these enhancements do not significantly improve performance in Geekbench 6.
37% higher GPU performance
At the presentation of its A19 Pro, Apple did not reveal anything about improvements to its GPU, but said that it still has six clusters. Based on Geekbench 6 results published so far, the A19 Pro GPU is a whopping 37% faster than its predecessor. The GPU scores 45,657 points, which is comparable to the GPU performance of M2 or M3 in iPad Air. It is also comparable to the performance of AMD's Radeon 890M integrated GPU.
| Row 0 - Cell 0 | A19 Pro | A18 Pro | A17 Pro | A16 Bionic | Snapdragon 8 Elite | Snapdragon 8 Elite |
Configuration | 6-cluster GPU | 6-cluster GPU | 6-cluster GPU | 5-cluster GPU | Adreno 830 | Vulkan | Adreno 830 | OpenCL |
Metal Score | 45657 | 33183 | 28429 | 23951 | 23839 | 17971 |
Background Blur | 23489 | 14458 | 12443 | 9760 | Row 3 - Cell 5 | Row 3 - Cell 6 |
Background Blur | 97.2 images/sec | 59.8 images/sec | 51.5 images/sec | 40.4 images/sec | Row 4 - Cell 5 | Row 4 - Cell 6 |
Face Detection | 31986 | 24306 | 20492 | 16855 | Row 5 - Cell 5 | Row 5 - Cell 6 |
Face Detection | 104.4 images/sec | 79.4 images/sec | 66.9 images/sec | 55.0 images/sec | Row 6 - Cell 5 | Row 6 - Cell 6 |
Horizon Detection | 40348 | 31675 | 27208 | 23537 | Row 7 - Cell 5 | Row 7 - Cell 6 |
Horizon Detection | 1.26 Gpixels/sec | 985.7 Mpixels/sec | 846.7 Mpixels/sec | 732.4 Mpixels/sec | Row 8 - Cell 5 | Row 8 - Cell 6 |
Edge Detection | 42578 | 35145 | 30801 | 28064 | Row 9 - Cell 5 | Row 9 - Cell 6 |
Edge Detection | 1.58 Gpixels/sec | 1.30 Gpixels/sec | 1.14 Gpixels/sec | 1.04 Gpixels/sec | Row 10 - Cell 5 | Row 10 - Cell 6 |
Gaussian Blur | 81074 | 34102 | 27105 | 28077 | Row 11 - Cell 5 | Row 11 - Cell 6 |
Gaussian Blur | 3.53 Gpixels/sec | 1.49 Gpixels/sec | 1.18 Gpixels/sec | 1.22 Gpixels/sec | Row 12 - Cell 5 | Row 12 - Cell 6 |
Feature Matching | 15603 | 12352 | 10670 | 7571 | Row 13 - Cell 5 | Row 13 - Cell 6 |
Feature Matching | 615.1 Mpixels/sec | 486.9 Mpixels/sec | 420.7 Mpixels/sec | 298.5 Mpixels/sec | Row 14 - Cell 5 | Row 14 - Cell 6 |
Stereo Matching | 124982 | 106998 | 92574 | 73904 | Row 15 - Cell 5 | Row 15 - Cell 6 |
Stereo Matching | 118.8 Gpixels/sec | 101.7 Gpixels/sec | 88.0 Gpixels/sec | 70.3 Gpixels/sec | Row 16 - Cell 5 | Row 16 - Cell 6 |
Particle Physics | 92533 | 83372 | 74580 | 63440 | Row 17 - Cell 5 | Row 17 - Cell 6 |
Particle Physics | 4072.5 FPS | 3669.3 FPS | 3282.3 FPS | 2792.1 FPS | Row 18 - Cell 5 | Row 18 - Cell 6 |
The Apple A19 Pro GPU shows the highest performance advantages over its predecessor in Background Blur (relevant for depth-of-field in games, real-time multi-layer compositing, video background blur, etc.) and Gaussian Blur (relevant for post-processing, vision processing, FPU operations). Still, the new GPU is faster than its predecessor quite significantly across the board.
Follow Tom's Hardware on Google News, or add us as a preferred source, to get our up-to-date news, analysis, and reviews in your feeds. Make sure to click the Follow button!
