Google just had its huge event to announce the Pixel 8 series of smartphones (along with some other fun gadgets). The company announced a new version of its custom Tensor processor to accompany the phone’s announcement. The Tensor G3 builds on the technology offered by the G2 in some notable ways that should be quite exciting for people who purchase Google’s latest smartphones.
What exactly does the Tensor G3 bring to the table that the Tensor G2 didn’t offer? That’s exactly what we’re going to explain.
What is Google Tensor?
Before we get into the specifics of the Tensor G3, you need to know what Google Tensor is. The name Tensor comes from Google’s open-source TensorFlow machine learning programming ecosystem.
Regarding what the chips are, they’re Tensor is Google’s in-house SoC. Google uses its own silicon instead of a MediaTek or Qualcomm chip like many other phone makers (Samsung uses its own chips in some regions, and Apple has long used its own silicon). However, unlike Apple’s in-house A-series chips, Google isn’t completely making everything itself. Instead, some of the parts are licensed.
This is probably why other chip makers spend time touting clock speeds while Google pushes its AI features and other functionality over the raw speed of the chip.
What are the differences between the Tensor G2 and G3?
What separates the Tensor G3 from its predecessor mostly comes from new features, not raw processing power. The additional processing power is apparent on the spec sheet when Qualcomm releases a new Snapdragon chip. We can look at the benchmarks and say it’s some percent faster than the last one. With Google Tensor, it’s a little less clear, as the company doesn’t publish the actually processing power of the chip.
To see the differences, we need to look at the new features, most of which center around AI, the company announced. We can also look at how much machine learning the new chip can handle, which Google says is twice as much machine learning on the devices with the Tensor G3 as compared to the Pixel 6’s Tensor.
Some new features that only work with the G3 are Google’s new Call Screen functionality, which can weed out spam calls by actually talking to the caller on your behalf. There’s also the Audio Magic Eraser, which can remove unwanted audio from a video, similar to how the existing Magic Eraser can remove things from photos. The G3 also enables screen reading on every part of the Pixel 8 and 8 Pro’s display. This isn’t new, as the G2 supported screen reading, but the G3 appears to add more functionality to it.
Generative AI is a big part of the new chip’s benefits, but many of those features aren’t out yet. Those benefits will become more apparent when the Google Bard features come to Google Assistant. Features like Best Take and Magic Editor are coming later in October, and they’ll really push the power of the G3 for photography.
What are the Tensor G3’s specs?
Unfortunately, Google didn’t announce the actual specs of the Tensor G3, which isn’t a surprise, as the company hasn’t announced the specs with previous generations of its SoC.
However, a leak from M. Brandon Lee of This Is Tech Today showcased the chip’s specs, which aren’t considered official yet. According to the leak, it’s a nine-core chip with one Cortex-X3 clocked at 2.91GHz, four Cortex-A715 medium cores at 2.37GHz, and four Cortex-A510 little cores at 1.7GHz. It also says it’ll use a Mali-G715 GPU.
Which phones use the Tensor G3 chip?
As of this writing, only the Pixel 8 and Pixel 8 Pro are slated to use the Google Tensor G3 chip, though Google will likely announce a Pixel 8a and Pixel Fold 2 that’ll use the new SoC at some point in the future.
Trending Products
