Is the Pixel 6 outdated already?

< type="image/webp"/>“Vb” decoding=”async” loading=”eager” title=”Snapdragonm 8 Gen 1 vs Google Tensor hero” alt=”Snapdragonm 8 Gen 1 vs Google Tensor hero” src=”https://www.tipsclear.com/wp-content/uploads/2021/12/Snapdragonm-8-Gen-1-vs-Google-Tensor-hero.jpg”/>

Robert Triggs / Android Authority

The arrival of Qualcomm’s next-gen Snapdragon 8 Gen 1 inevitably calls into question whether current-gen hardware is completely outclassed or not. While we’re well past the point of “good enough” smartphone performance, new hardware drives innovation forward and it’s often the only way to experience new cutting-edge use cases. So we’re already eagerly anticipating 2022’s flagship smartphones.

Google also has new hardware in the market in the form of its first semi-custom Google Tensor SoC. Boasting in-house machine learning smarts, Google’s entry into the mobile chip market has already produced some standout features in the Pixel 6 and Pixel 6 Pro. But has Google’s silicon already been eclipsed by Qualcomm’s upcoming flagship processor? Read on to find out.

Snapdragon 8 Gen 1 vs Google Tensor specs

Snapdragon 8 Gen 1Google Tensor
CPU Config

Snapdragon 8 Gen 1:

1x Arm Cortex-X2 (3.0GHz)
3x Arm Cortex-A710 (2.5GHz)
4x Arm Cortex-A510 (1.8GHz)

Google Tensor:

2x Arm Cortex-X1 (2.80GHz)
2x Arm Cortex-A76 (2.25GHz)
4x Arm Cortex-A55 (1.80GHz)

GPU

Snapdragon 8 Gen 1:

Adreno

Google Tensor:

Arm Mali G78 MP20
848 MHz (shader clock)
996 MHz (tiler clock)

TO THE

Snapdragon 8 Gen 1:

Hexagon DSP

Google Tensor:

Google TPU

RAM support

Snapdragon 8 Gen 1:

LPDDR5 @ 3,200MHz

Google Tensor:

LPDDR5

4G/5G Modem

Snapdragon 8 Gen 1:

X65 LTE/5G (integrated)
10Gbps down (mmWave)

10CA in 5G
4×4 MIMO
Up to 256-QAM in sub-6GHz

Standalone and Non-Standalone

Google Tensor:

Exynos Modem 5123 (external)
7.35Gbps down (mmWave)
5.1Gbps down (sub-6GHz)

8CA in 5G
4×4 MIMO
Up to 256-QAM in sub-6GHz
Up to 64-QAM in mmWave
Up to 1024-QAM in 4G

Standalone and Non-Standalone

Other networking

Snapdragon 8 Gen 1:

Bluetooth 5.2
Wi-Fi 6E, Wi-Fi 6 (802.11ax), Wi-Fi 5 (802.11ac), 802.11a/b/g/n

Google Tensor:

Bluetooth 5.2
Wi-Fi 6E, Wi-Fi 6 (802.11ax), Wi-Fi 5 (802.11ac), 802.11a/b/g/n

Process

Snapdragon 8 Gen 1:

4nm (Samsung)

Google Tensor:

5nm (Samsung)

Snapdragon 8 Gen 1: Defining the next generation

< type="image/webp"/>

Browsing the spec sheet comparison there are some clear and obvious design wins for Qualcomm’s flagship processor. The CPU setup embraces the latest Armv9 architecture, complete with a more powerful Arm Cortex-X2 CPU, three Cortex-A710s, and three more efficient little Cortex-A510 cores. With the Snapdragon 8 Gen 1 touting 20% CPU performance gains and 30% energy-efficient improvements over its predecessor, due in part to the upgraded Samsung 4nm process, Qualcomm’s chip will best Tensor in benchmarks.

Although we shouldn’t discount Google Tensor’s performance here. The Arm Cortex-X1 core is still perfectly fast, and Google’s chip has two of them, which, in some workloads, could outperform the Snapdragon 8 Gen 1’s single beefy core approach.

Read more: Arm Cortex-X2, A710, and A510 deep dive

We’re anticipating a similarly competitive performance in the graphics department. Benchmarks put the large 20-core Mali-G78 Google Tensor a little ahead of the Snapdragon 888. But with a 30% graphics boost on the cards, the Snapdragon 8 Gen 1 will regain that crown. We’ve also seen some early signs that the Pixel 6 doesn’t sustain peak performance for long, while Qualcomm’s chipsets historically fair better here, although we’ll have to see what the situation is like with its latest chip and actual smartphones before drawing conclusions.

The Snapdragon 8 Gen 1 benefits from the latest Armv9 CPU cores.

Broadly speaking, we’re expecting the Snapdragon 8 Gen 1 to outperform Google Tensor in the core metrics but not by a huge amount, and Google’s chip clearly excels in some specific use cases that leverage the chip’s more unique machine learning-oriented design. Comparing machine learning chops is very difficult due to the variety of use cases and different integration strategies. Benchmarks also often struggle to capture the requirements of real-world use cases. Google leans on its own in-house TPU, while Qualcomm offers dedicated ML capabilities in its ISP, DSP, and other departments. Either way, both are very capable here, although we should note that third-party developers currently can’t tap directly into Google’s TPU to make the most of it.

There are also some key similarities on the security side. Tensor features Google’s Titan M2 security enclave for tamperproof credential storage and processing. Qualcomm now offers its own Trust Management Engine with support for Android Ready SE, which opens the door to similar use cases, including on-device ID storage and iSIM support.

See also: What is an SoC? Everything you need to know

Where the Snapdragon 8 Gen 1 does take a more clear-cut lead is in the 5G networking department. With a Release 16-class modem sporting 10 carrier aggregation and mmWave and sub-6GHz band blending, Qualcomm’s chipset offers peak downloads of 10Gbps. Google Tensor isn’t far off the mark though.