We posted the poll inside our benchmark article on October 12, and it attracted over 2,800 votes to date. So did readers think Google nailed the Pixel 7’s processor? Well, 64.08% of respondents, equating to almost two-thirds of voters, voted for “it’s fine, could be better.”
We can see why people voted for this option as the chipset doesn’t seem like a major CPU and GPU upgrade over the original Tensor. And the first-gen Tensor was already slower in some ways than last year’s rival processors. The SoC should nevertheless offer more than enough grunt for most people, even if it doesn’t match rivals for peak performance.
Meanwhile, 19.99% of surveyed readers felt that Google didn’t nail the Tensor G2. This is an understandable choice given Google’s insistence on sticking with older CPU tech, an older manufacturing process, and a GPU that seems like a sideways move or a slight downgrade on paper.
Finally, 15.93% of respondents felt that Google did indeed nail the Pixel 7 processor. The processor seems to deliver good sustained performance compared to rival SoCs, while other elements like the TPU and modem have received welcome upgrades too.
Rato Ketu: Can’t wait for next year’s Tensor G3. Also, Google should consider making Tensor chips available for non-Pixel Android phones, make it a compelling option for other manufacturers to use it instead of Snapdragons or Mediateks.
Dragos Lucian: If only they used TSMC’s node instead of Samsung’s. I hope they do that in the future.
Delta Whiskey Hotel: Whether it was intentional or just a happy accident, they certainly seem to have achieved performance under load in a way that other manufacturers simply haven’t for the last few years. If Tensor G3 can put together burst performance along with sustained performance, it should be a slam dunk.
Beardednomad: Still doesn’t hold a candle to Qualcomm but overall the performance is fine for the vast majority of people. The biggest difference is in the modem which based on reviews is still a complete mess on Pixels.
Eoaoos: In other words, they’re all more than enough. But it’s good to see the Samsung & Google supports AV1 decode.
Dorian_at_ATT: Sustained performance is wonderful. There’s no need to put that much GPU power in something if it can’t do anything with it, and really what area are you going to be using the GPU for? Gaming. Especially on a phone. On a PC I can see it being used computationally, but if you’re doing that on a phone you need to look at the Tensor neural cores or you need to re-think some things about how you work.