Homepage > Google’s Pixel 6 Tensor processor looks extremely weird

Google’s Pixel 6 Tensor processor looks extremely weird

Google is set to launch its first in-house smartphone chipset, the Tensor SoC, in its upcoming Pixel 6 and Pixel 6 Pro phones. What if the last report of XDA is correct, the supposed processor configuration of the Tensor could be very, very strange, even by Google’s standards.

So far, Google has greatly increased the AI ​​performance of the Tensor SoC; but it did not reveal any information about the core specs of the chip’s processor and GPU. Google’s Rick Osterloh would only say The edge that “the standard things that people watch will be very competitive and the AI ​​things will be totally differentiated.”

Some aspects of the components of the Tensor have already emerged. One earlier XDA report notes that the Pixel 6 will likely use a standard Arm Mali-A78 GPU design (which Samsung uses on its flagship Exynos 2100), while Reuters reports that Google will source the 5G modem from Samsung.

But the CPU was still a mystery, until today, when XDA released a report based on both a Geekbench score and a source claiming to have a real Pixel 6 Pro. The report claims that the processor configuration on the Tensor will consist of two Cortex-X1 performance cores clocked at 2.802 GHz, two Cortex-A76 performance cores clocked at 2.253 GHz, and four Cortex-A55 efficiency cores.

Google's new Pixel 6 phones, standard and Pro

Image: Google

If you’ve been following the major smartphone flagship chips, this is a very strange list to see, a list that mixes powerful new cores with weaker old ones. Let’s take a step back to explain why:

When we look at most smartphone SoCs, there are generally two main parts: performance processor cores and efficiency processor cores. Arm-based designs tend to mix them up in big.LITTLE configurations, to allow devices to boost their performance by using the more powerful “big” cores for intensive tasks like gaming, while performing tasks. less demanding (like checking your email) on “small” efficiency cores to extend battery life.

A typical Arm-based design might include four performance cores (like the Cortex-A78) and four efficiency cores (like the Cortex-A55). But last year, Arm added an even more powerful new performance option for chipmakers: the Cortex-X1.

So the best smartphones of 2021 tend to offer a triple-cluster design: the Snapdragon 888 uses partially customized versions of a single Cortex-X1, three Cortex-A78s, and four Cortex-A55 cores, while the Exynos 2100 from Samsung uses a similar setup. . Tensor, on the other hand, would offer two Cortex-X1 cores, two Cortex-A76 cores, and the usual four Cortex-55 cores.

Which makes it a very odd version of a triple cluster design. By including not one, but two Cortex-X1 performance cores, Tensor could theoretically allow it to outperform even the best chips from Qualcomm and Samsung, on paper, if not for the second half of the rumor, which is that Google is also using two the old Cortex-A76 cores … which, quite simply, doesn’t make sense.

As XDA points out, the Cortex-A76 was introduced in 2018 and represents two full generations behind the Cortex-A78 design used in the flagship chips of 2021. There’s no immediately logical reason why Google would use the old design, either; the A78 is both faster and more efficient than its older counterpart, making it a extremely strange choice to include in the Tensor processor cluster, especially if Google is already going all out with two X1 cores.

It’s possible that Google is simply hiding its processor design in the Geekbench score, although the report notes that this would be unlikely.

For now, however, the mystery of the Pixel 6’s Tensor chip has grown even stranger. And it will likely stay that way until Google reveals more information when the phones arrive later this fall.

.
Similar Posts:

Leave A Reply

Please enter your comment!
Please enter your name here

Stay on Top - Get the daily news in your inbox

Recent Articles

Most Popurlar

Trending this Week

Similar Posts: