2024 SoC Comparison: Tensor G5 GPU Underperforms Compared to Tensor G4 in Various Tests, Achieves Up to 323% Higher Speed

Google has initiated a significant transformation within its Pixel 10 lineup, primarily by unveiling the Tensor G5 chip. Notably, this marks the company’s first venture into TSMC’s advanced 3nm manufacturing process. Initial insights into the system on a chip (SoC) indicated enhanced speed in computing tasks compared to its predecessor, the Tensor G4. However, details surrounding the graphics performance were sparse—until now. Recent benchmark comparisons have revealed some concerning statistics, yet prospective buyers should consider the underlying reasons for the Tensor G5 GPU’s performance metrics before making any decisions regarding upgrades.

Discrepancies in Performance: The Tensor G5 CPU vs. GPU Results

According to a comparison shared by @lafaiel on X, the Pixel 10 Pro’s GPU achieved a mere 3, 707 points in the Vulkan category of Geekbench 6. This result is strikingly less than what the previous Pixel 9 Pro managed, which surpassed the G5’s score by over twofold. Additionally, competitors like the iPhone 16 Pro and Galaxy S25+ demonstrated superior performance, raising questions about whether Google intentionally underperformed the GPU for cost-saving purposes. Notably, the lack of ray tracing support from the unrevealed graphics processor is a critical specification that customers should consider.

Tensor G5 GPU shows terrible results in latest performance comparison with the Tensor G4 — Benchmark results for the Pixel 10 Pro / Image credits – Geekbench

The findings indicate that the Pixel 9 Pro exhibits a remarkable performance advantage over its successor, achieving nearly a 323 percent improvement across several Geekbench 6 tests. This begs the question of what might be affecting the Tensor G5’s performance. While Google has been known to minimize transparency concerning its hardware choices, it appears that this is not the case with the Tensor G5. In the discussion thread regarding the benchmark, a keen observer noted that outdated GPU drivers could be the root cause of the Tensor G5’s uninspiring performance results.

Comparison of Tensor G5 and Tensor G4 in Geekbench 6 — Performance comparison between Tensor G5 and Tensor G4 / Image credits – Geekbench

Diving deeper into the GPU’s functionality, it appears that it only operated at a clock speed of 396MHz, rather than the expected 1, 100MHz. Further commentary from @cozy_OSS revealed that the PowerVR DXT component is ‘Vulkan 1.4 conformant’.In contrast, the drivers for the Pixel 10 Pro are currently limited to ‘Vulkan 1.1′.There is hope that forthcoming driver updates will significantly enhance performance, and as a result, we can anticipate revisiting the Geekbench 6 results soon for a comprehensive re-evaluation.

Source & Images