Author
I have a keen interest in benchmark testing and frequently utilize them when evaluating new smartphones. Prominent synthetic benchmarks such as Geekbench and 3DMark effectively emulate typical usage and gaming scenarios, offering reliable insights into a device’s performance capabilities. With over 75 smartphones tested each year, synthetic benchmarks enable me to conveniently compare the performance of newer models against their predecessors.
A novel benchmarking tool called Evolve is available, designed to provide more than just a simple performance score. Evolve assesses performance across five distinct metrics, allowing users to identify which aspects of the GPU are efficiently handling contemporary workloads and where there may be performance bottlenecks. I have been using it on several devices, including the Find X9 Ultra, Find X9 Pro, Find X8 Pro, and the Xiaomi 15 Ultra, and it has successfully demonstrated the significant advancements in mobile GPU technology over the past two years.
Evolve Provides In-Depth GPU Insight
As mobile gaming increasingly incorporates ray tracing technology, companies like Arm and Qualcomm are intensely focusing on neural rendering methods, such as frame generation, to enhance framerates in mobile games. Evolve excels in this regard by offering a dedicated score that evaluates all hardware-accelerated ray tracing tasks — such as lighting, shadows, and reflections — executed through Vulkan Raytracing and DXR APIs. It effectively gauges how proficiently a device manages ray tracing in general. The benchmarking tool also features an Acceleration Structure score, measuring the speed at which the GPU constructs requisite data structures for ray tracing.
Moreover, Evolve includes a rasterization score assessing the hardware rasterizer, which is crucial for 3D graphics. It tests animation, visual effects, and post-processing capabilities. A driver score is also included, gauging the CPU’s efficiency. Collectively, these evaluations aim to illustrate how well a mobile GPU performs during gaming sessions. Evolve is freely accessible; however, a professional version exists that enables users to export charts and testing data, which I utilized in my evaluations.
Applying Evolve in Benchmark Testing
In my testing, I used the last two generations of Qualcomm and MediaTek chipsets with Evolve. This included the Snapdragon 8 Elite Gen 5 (Find X9 Ultra), Dimensity 9500 (Find X9 Pro), Snapdragon 8 Elite (Xiaomi 15 Ultra), and the Dimensity 9400 (Find X8 Pro).
The Evolve charts exclusively highlight GPU performance, which is the core focus of the tests. A quick summary includes the Snapdragon 8 Elite Gen 5 with the Adreno 840, Dimensity 9500 housing the Mali G1, the Snapdragon 8 Elite sporting the Adreno 830, and the Dimensity 9400 featuring the Mali-G925. Let’s delve into the scores now.
It is unsurprising to find Qualcomm leading the benchmark results; both recent generations of Adreno GPUs have showcased marked performance improvements reflected in the scores. The Adreno 840 performs 27% better in ray tracing workloads than the Mali G1, is 40% superior in rasterization, and boasts a striking 46% advantage in GPU compute power. It is essential to highlight that this does not imply that the Adreno 840 offers a 40% advantage in gaming experiences; the performance difference is much narrower in practical gaming contexts. However, the Adreno 840 provides more headroom and superior sustainability during extended gaming sessions.
One noteworthy observation is the significant year-on-year enhancement from the Adreno 830 to the Adreno 840; the latter excels by 1.47 times in rasterization tasks and a remarkable 85% in ray tracing. The performance of the Mali G1 compared to its predecessor, the G925-Immortalis, is equally intriguing; Arm claimed a twofold improvement in ray tracing capabilities for the G1, which is evident in the results. Nevertheless, the G1 did not experience a proportionate improvement in other facets, particularly not to the extent achieved by Qualcomm with the Adreno 840.
Evolve distinctly illustrates GPU time allocation for each task, helping to identify the GPU’s workload. For example, the Adreno 840 allocated 27.7% for ray tracing, over 17% for rasterization, and the majority — 45.7% — to GPU compute. In contrast, the Mali-G925 dedicated over 50% of its resources to ray tracing due to a deficiency in the necessary cores. These insights are beneficial as they reveal specific areas where a GPU struggles, providing detailed granularity.
Additionally, Evolve offers an energy score representing GPU power consumption during workloads, crucial for mobile chipsets requiring energy efficiency. Unfortunately, this feature was inaccessible on the phones I utilized. To clarify details, I consulted Jasper Bekkers, founder of Evolve, who explained that the energy score is derived from actual power data through Android’s Power Stats HAL. Some devices do not provide this information, resulting in the score being omitted for those models.
Looking ahead, I am optimistic about the performance advancements expected with the Snapdragon 8 Elite Extreme Gen 6 and Dimensity 9600 Pro, as both platforms embrace an enhanced 2nm fabrication process, with promises of generational GPU improvements. I am currently testing the Dimensity 9600 Pro — a powerhouse — and will share insights on this in the near future. Qualcomm’s flagship chipset of 2027 is also anticipated to launch soon, and I eagerly await testing devices featuring these upcoming technologies.
