Google blocked Pixel 8, Pixel 8 Pro reviewers from using popular benchmarks to test the Tensor G3 chip, new owners too
In a highly unusual move, it has been revealed that Google has blocked reviewers of the Pixel 8 and Pixel 8 Pro from installing popular benchmarks including Geekbench and 3D Mark. Geekbench is a popular benchmark that assesses CPU performance in smartphone silicon, while 3DMark is a benchmark that assesses the GPU performance of a chip. Running benchmarks like these is a typical part of the review process and the review to help formulate their overall assessment of a device based on both objective data and qualitative analysis based on real world usage.
Google's narrative about the Tensor G3 has been to emphasize that it has been designed with efficiency and AI in mind rather than outright performance -- making it clear it is acutely aware of its relative performance shortcomings. While blocking access to the tests via the Google Play Store will have affected the ability for some reviewers to run the benchmarks, others with a little more technical knowledge have been able to by-pass this limitation using a method known as side-loading. Consequently, Geekbench, 3DMark and other benchmark results for the Tensor G3 SoC powering the Pixel 8 and Pixel 8 Pro have surfaced regardless. These results have not been flattering.
In the first Geekbench 6 result to surface during the embargo period, the Tensor G3 as fitted to the Pixel 8 Pro achieved a single-core score of 1760 while it reached a score of 4442 in the multi-core test. This compares unfavorably with the Qualcomm Snapdragon 8 Gen 2 (soon to be superseded by the even faster and more efficient SD8G3), which produces a single-core score of 2003 and 5427 in the multi-core test. The marked discrepancy is despite the fact that both chips use very similar Arm-based CPU architectures. In fact, the Tensor G3 actually enjoys the benefit of an additional mid-core (9 cores total versus 8 cores) and a better overall mid-core architecture. In addition to 2x Cortex-A715 cores, Qualcomm actually opted for a couple of slightly older Cortex-A710 cores for 32-bit backwards compatibility -- the Tensor G3 exclusively uses four of the newer Cortex-A715 cores.
The performance disparity can be explained by Google's partnership with Samsung, which gave it the opportunity to develop more customized chips, but has also locked them into using Samsung's problematic 4 nm node technology. Not only does Samsung's 4 nm node produce much less efficient chips, it also can't match the transistor density of TSMC's superior N4P node used to fabricate the Snapdragon 8 Gen 2. Similar performance shortcomings, are apparent in the first 3D Mark Wild Life Stress results testing the Tensor G3's Arm Mali-G715 GPU which also surfaced during the embargo period, despite Google's best efforts to block them. Efficiency tests have also recently been conducted showing the Tensor G3 well down the list of the most efficient performers, and more comparable to older chips like the Snapdragon 888.
It is not unusual during the launch phase of any major smartphone for tech-focused segments of social media to light up with benchmarks results, even during the embargo period suggesting that Google’s highly unusual tactics have been effective – at least to an extent. Tactics that Google appears determined to continue, it seems. In a perhaps even more extraordinary move, Google extended the block to new owners of the Pixel 8 and Pixel 8 Pro, preventing them from installing benchmarks as well. As mentioned earlier, this will not stop users from sideloading the apps, but it will undoubtedly help mitigate against the wider spread of information regarding the performance and efficiency question marks hanging over Pixel 8 and Pixel 8 Pro.
While Pixel fans will rightly argue, smartphones don’t necessarily need to have the most outright performance to be serviceable. However, the performance of the Tensor G3 puts it more in line with smartphones priced at the mid-range segment of the market – not the high-end market dominated by Apple and Samsung that sell similarly priced flagship phones with chips that are faster, more efficient, and even possess faster hardware-based AI/ML acceleration. Of course, Google’s strengths remain its leadership in software-based AI features, its excellent cameras and high-end design and build quality, which Google is clearly doing its utmost to keep users focused on.
However, blocking reviewers and ordinary users from downloading and installing popular benchmarking apps is suggestive of a company not prepared to stand by the marketing claims it has made about the Tensor G3. Otherwise, it wouldn't be trying to impede reviewers and users from being able to test those claims themselves.
Source(s)
Shreyas via Twitter/X
Own