Submit a review

Google Pixel 8: GPU tests show low performance, high efficiency

Google has once again teamed up with Samsung to develop the Tensor G3 chipset in the new Pixel 8 and Pixel 8 Pro.
Google Pixel 8 Pro test

Google has once again teamed up with Samsung to develop the Tensor G3 chipset in the new Pixel 8 and Pixel 8 Pro. The G3 is manufactured using Samsung’s 4nm process and features a 9 core CPU (1x Cortex-X3 + 4x A715 + 4x A510) and an ARM Mali-G715 GPU. Google has not yet revealed any details about this GPU. But it seems to have 7 computing units, just like the Tensor G2 from last year with its Mali-G710. How efficiently this GPU works has now come to light in several tests.

3DMark Wildlife Extreme

Starting with 3DMark Wildlife Extreme, the chipsets are rated based on the score divided by the average power consumption – the higher the better. The Tensor G3 doesn’t perform too well. It is about as fast as an Apple A14 and slower than the Snapdragon 8 Gen 1. MediaTek’s Dimensity 9200 also uses the Mali-G715 GPU, but with 11 processing units. In contrast to the G3, the Dimensity 9200 is manufactured by TSMC using their 4nm process. The Dimensity far outperforms the Tensor in terms of performance, but is still almost on par in terms of efficiency.

GFXBench Aztec Vulkan/Metal

In the GFXBench Aztec Vulkan/Metal benchmark test at 1440p, the Dimensity 9200 also outperforms the Tensor G3. It manages almost twice the frame rate, but also uses almost twice as much power, so the efficiency is basically the same. It is somewhat disappointing that the Tensor G3 can only make up little ground in this test compared to its predecessor G2.

GFXBench 3.1

Finally, another GFXBench test with version 3.1 shows that the gap between the Google chip and the MediaTek processor is smaller at 1080p resolution. Nevertheless, if you take power consumption into account, the two chipsets are on par here too.

According to the reviewer, Samsung’s 4nm LPP node is 1.5 to 2 generations behind the 4nm TSMC node. Therefore, the Tensor G3 cannot keep up with the current flagship platforms in the more demanding tests. Not to mention the Snapdragon 8 Gen 3 and the Dimensity 9300, both of which will be unveiled later this year.

The disadvantage in the manufacturing process doesn’t bode well for Samsung’s own Exynos 2400, even though it has a completely different GPU. Samsung is working with AMD on graphics, bringing their RDNA 3 technology into play. This advantage could keep the Exynos in the game. The Tensor G4 will likely also come from Samsung again as Google still needs the expertise of the Exynos team. However, there are rumors that Google is working on an in-house design that won’t be dependent on Samsung for 2025.


Thomas Dietrich
We will be happy to hear your thoughts

Leave a Reply

Register New Account
Compare items
  • Total (0)