Key Takeaways
1. The Google Tensor G5 achieved scores of 2,276 (single-core) and 6,173 (multi-core) in Geekbench tests, showing improvement over earlier scores but still falling short of competitors.
2. The chip features a CPU layout of one Cortex-X4 core, five Cortex-A725 cores, and two Cortex-A520 cores, with a new PowerVR DXT-48-1536 GPU.
3. The Tensor G5 demonstrates consistent performance without throttling, indicating stable clock speeds during testing.
4. Despite its improvements, the Tensor G5 is unlikely to compete with leading chips like the Snapdragon 8 Elite and Dimensity 9400 due to its use of older Arm cores.
5. Future potential may be seen if the chip transitions to TSMC’s 3 nm process, although the Exynos modem could negatively affect overall performance.
The first performance test of the Google Tensor G5 on Geekbench wasn’t very impressive, which makes sense since it was likely an engineering model. Now, we can see it tested again, this time paired with Google’s upcoming foldable device, the Pixel 10 Pro Fold.
Performance Scores
In the latest Geekbench results, the Tensor G5 achieved scores of 2,276 for single-core and 6,173 for multi-core tests. The back-end data from Geekbench indicates that the chip isn’t throttling and can sustain a consistent boost clock. Although these numbers show improvement over the initial scores of 1,323 and 4,004, they still fall significantly short compared to leading competitors like the Snapdragon 8 Elite (3,070/9,251) and the Dimensity 9400 (2,597/8,109). It does come closer to the older Dimensity 9300, which scored 2,207 in single-core and 7,408 in multi-core tests.
CPU Layout and GPU
We also get an insight into the CPU configuration of the Tensor G5, which features one Cortex-X4 prime core clocked at 3.78 GHz, five Cortex-A725 performance cores running at 3.05 GHz, and two Cortex-A520 efficiency cores at 2.25 GHz. Additionally, we see its new GPU from Imagination Technologies, the PowerVR DXT-48-1536. The Pixel 10 Pro Fold being tested has 16 GB of RAM and operates on Android 16.
Future Potential
Realistically, the Tensor G5 is unlikely to match up against the current generation giants, particularly because of its use of older Arm cores. However, switching to TSMC’s 3 nm (or 5 nm, depending on whom you ask) process might improve its power efficiency. On the flip side, the Exynos modem could negatively impact performance.
Source:
Link