Google’s Tensor chip is just the first step towards great Pixel phones

Google’s Pixel 6 is shaping up to be the company’s most ambitious smartphone in years, thanks in large part to the new custom-designed Tensor processor, which aims to propel Google to the forefront of the smartphone market through to the power of Google’s years of machine learning experience. . And Google needs it: Despite the popularity of Pixels in tech circles, its phones just aren’t popular sellers in the US, barely moving the needle compared to behemoths like Samsung and Apple or even gamers. smaller like Oppo or Xiaomi.

Tensor is Google’s big bet, focused on the AI-boosting TPU that promises to improve photos and videos, search, closed captioning, text-to-speech, and more. That’s a tall order for any chip, let alone one that focuses primarily on machine learning as a standout feature – but although the Tensor SoC may not yet be able to launch Google to heights comparable to iPhone, it could be a crucial first step towards transforming its Android as well. – met a leading competitor.

Image: Google

The rest of the SoC is a mystery at the moment, but it looks like Google will be using third-party designs for things like the processor, GPU, and modem, meaning the Pixel 6 will likely feel quite similar to any other. Android smartphone powered by a Qualcomm or Samsung processor for most tasks, instead of some kind of revolutionary upgrade comparable to Apple’s A-series iPhones.

Google hasn’t given much information on what the actual Tensor architecture will look like for things like the CPU, GPU, modem, or other major SoC components beyond the TPU. But based on the rumors and the fact that Google isn’t taking the opportunity to brag about any major customizations or its progress here, it’s likely that most of Tensor’s hardware stack will be outsourced designs. Qualcomm and Samsung are already doing something similar – the Snapdragon 888 uses partially customized versions of Arm’s Cortex-X1, A78, and A55 designs, while Samsung’s Exynos 2100 uses Arm designs for its processor and GPU.

XDA Specifically notes that the Tensor will likely be a combination of Arm’s Cortex-A78, Cortex-A76, and Cortex-A55 processor cores and Arm’s standard Mali GPU. Which means the difference between Tensor and, say, a Snapdragon 888 or Exynos 2100 might not be that big for things like overall CPU or GPU performance. It’s a good category, especially if Google is really trying to create a flagship device.

Image: Google

But as Google’s Rick Osterloh said The edge, “The standard items that people will watch will be very competitive and the AI ​​items will be totally differentiated.” However, what makes Tensor special and unique is not the speed at which it can run games or the efficiency of its processor and battery.

That means Tensor probably won’t be the magic bullet Android fans were hoping for either: a custom chip, designed by Google, designed specifically for Android and the Pixel’s hardware to deliver the kind of performance and power that Apple does. has been able to deliver on the iPhone for years using a similar strategy.

But the good news is that Tensor is just a first generation product; It’s easy to point the finger at Apple’s custom chips and demand that Google do the same, but it’s important to remember that Apple’s first A-series iPhone chips started out the same way. The A4 and A5 chips featured standard Arm designs for processor cores (with a few more optimizations and improvements), before Apple moved on to fully custom designs in subsequent generations.

Google is rumored to be working with Samsung, which will both manufacture the chips – which makes sense, given that it is one of only two companies, along with TSMC, to currently manufacture 5nm chips (a service it provides to many companies, including Qualcomm).

Samsung is also rumored to be working more closely with Google on the actual chip design itself, leveraging its Exynos hardware and software in a capacity that goes beyond simple manufacturing. We’ll have to wait for Google to reveal more about Tensor to know the extent of this partnership, but that could mean Google will get additional help in more substantial custom chip improvements than if it had to reinvent the wheel from scratch.

The idea of ​​a specially designed machine learning chip in a Pixel phone isn’t a new idea for Google either. The company has already installed dedicated internal AI chips in its phones, like the Pixel Neural Core and Pixel Visual Core in the Pixel 2, PIxel 3, and Pixel 4 models, making the Tensor less a whole new advancement and more of a refinement. of his past work.

Image: Google

Tensor’s TPU appears to be the next step, presumably offering more powerful AI functionality than either of these previous two chips. It’s also more tightly integrated with the phone, with Google noting that it’s capable of doing things like routing image data directly through the TPU.

These benefits may be worth developing a custom SoC – again, we’ll have to wait and see how Google actually implements these AI features and how substantial they are compared to a standard Android flagship. (A sentiment that, in general, applies to how Google places a heavy emphasis on TPU as the main differentiator from the Pixel 6, something that seems like a tough thing to sell to customers.)

As it stands, however, the main difference between Tensor and a Snapdragon 888 seems to be that it will just be a few fancy neural network tricks. But, just like with Apple’s first internal chips, Tensor may just be the first step in a long journey towards a more personalized Google chip. And that’s a far more exciting concept than subtly better machine learning demos.

Related: