Google launched the Pixel 7 series alongside the Pixel Watch. At the center of the new Pixel smartphones and the teased Pixel Tablet is the new Tensor G2 chipset.
The second-generation Tensor chip is a minor upgrade from the company’s first system-on-a-chip.
Google Tensor G2
The G2 shares much of the chip layout namely 2 x“Big” Arm Cortex X1 cores and 4 x “Small” Arm Cortex A55 cores. One of its upgrades is in the 2 x “Medium” cores which are now Arm Cortex A78 cores instead of the A76. Additionally, the clock speeds on the Big and Medium cores are slightly higher.
Its Mali-G710 MC10 GPU is also said to offer performance and efficiency boosts over the Arm Mali-G78 MP20. Moreover, the new GPU is said to give a 35% boost for machine learning tasks.
The Tensor G2 also has the next-generation TPU AI accelerator which is able to run learning tasks 20% more efficiently and 60% faster.
Among the features that the new chipset feature include Photo Unblur which uses the company’s advanced AI to remove blur and noise from old photos taken with any device. This is a feature exclusive to the new Pixel 7 series and can be used on photos taken with the new Pixel phones.
The chipset also introduces Cinematic Blur mode, Super-Res Zoom, and Night Sight photos.
Additionally, there are voice capabilities including faster Assistant interaction, voice translation, and voice typing, to name a few. Thanks to the new chipset, you can also describe emojis while going through Assistant voice typing, like asking for an “inverted smiley”.
On paper, it looks like a pretty interesting upgrade to the Tensor SoC but also looks like a pretty small update. We’ll have to wait for benchmarks and real-life tests to see just how the Tensor G2 performs.