Google’s Tensor SoC preliminary benchmark data appears, lost to Apple A12 Bionic
On the just-released Pixel 6 and Pixel 6 Pro, Google abandoned the Qualcomm solution it had been using and adopted the self-developed Tensor SoC. This SoC uses a triple-cluster design in the CPU part, which includes two Cortex-X1 super-large cores, is equipped with Mali G78 GPU, and is equipped with a TPU for AI computing, and integrates a Titan M security chip. It uses Samsung 5nm process manufacturing.
Previously, Google claimed that in internal benchmark tests, Tensor SoC can provide 80% performance improvement, so the application load speed will be faster, the game response will be faster, and it can save power and have longer battery life. Google’s statement is rather vague, and there is neither a specific comparison object nor a specific project.
Recently, Twitter user @9lekt provided Geekbench 5 benchmark test data of Tensor SoC. The objects of comparison are the Pixel 6 Pro equipped with Tensor SoC and the Apple iPhone XS MAX equipped with A12 Bionic, showing that Google’s self-developed SoC is not equal to the Apple product three years ago.
Although the test results cannot fully explain the problem, after all, in the use of mobile phones, there may be differences between the optimization of the software and the conversion to the actual experience, but such results will definitely disappoint many potential users. Google is no stranger to Qualcomm’s solutions. Its self-developed chip plan shows that there may be different ideas in the smartphone plan. According to Google’s previous introduction, Tensor SoC focuses on AI acceleration performance, which may be at the expense of some CPU performance.
