Robert Triggs/Android Authority
When Google introduced its first Tensor chipset with the Pixel 6, its intentions were quite clear: to be independent from Qualcomm’s Snapdragon (or any other chipset vendor) and create experiences that would otherwise be impossible. While we’re still waiting for the Pixel 8 series and its accompanying third-generation Tensor G3 processor, Google’s chips already deliver some impressive features that would likely be impossible on other chipsets, particularly AI and In Image Processing.
However, all this was somewhat undermined by the fact that Tensor chips have so far been designed in collaboration with Samsung. Although they contain Google-made parts, key aspects of their design are derived from Samsung’s Exynos chips.
Unfortunately, this will be the case for the upcoming Tensor G3 and future Tensor G4. Thanks to a Google insider, Android Authority Now Google’s Tensor plans for 2024 and beyond are known.
Google Tensor G4 (and beyond)
As reported Information (paid), Google initially released its 2024 Pixel series with a “fully custom” Tensor SoC codenamed “Redondo” (also known internally as “RDO”) built on a TSMC process node. Had planned to do. However, due to the missed deadline, the chip was too late to be included in the 2024 Pixel 9 series (as chip development is a lengthy process, usually starting several years before release).
Google has instead decided to build the SoC for testing purposes, as its successors, which will likely make it into 2025 Pixel devices, will share many of its design elements. As we’ve learned, Google has designed a development board called “ChallengerDeep” to work on the chip.
However, that’s still the case for the Pixel 9. With Redondo out of the picture, Google needs another chip to become the Tensor G4. The solution to this problem is a new chip codenamed “Zuma Pro”. Note that the Tensor G3 – the SoC that’s coming to the Pixel 8 series – is known as “Zuma”.
As planned, the Pixel 9 will not feature Google’s Redondo chip, instead it will use the Zuma Pro.
The new chip is still semi-custom, co-designed with Samsung’s System LSI division, and will likely be a small upgrade from the initial plan. This is similar to how Tensor G2 (codenamed “Whitechapel Pro”) was a minor update compared to the original Tensor (“Whitechapel”). The chip currently runs under a development board codenamed “RipCurrent 24” (also known as “RipCurrent Pro”), while the Tensor G3 simply uses “RipCurrent”.
Google is already working on a next-generation, fully-custom Tensor chip codenamed “Laguna Beach” (or simply “Laguna”) for the 2025 Pixel. The chip’s development board is codenamed “DeepSpace”.
You may be confused about what “fully custom” or “semi-custom” means in the context of SoC design. To answer this, we need to look at the history of Tensor chips and how they will change after this switch.
How to Design a Custom SoC (Fast)
Hadley Simmons/Android Authority
Co-designing the SoC with Samsung
Google probably first planned to design its own SoC something like this: several years ago, it wanted to combine the great work of its existing hardware teams (like the Pixel Visual Core, Titan security chip, and Edge TPU) into a single smartphone. Wanted to integrate into. Chip that will be specifically tailored to its needs and requirements.
At the time, the Google chip team (known as gChips) was quite small, so it would be difficult (if not impossible) to design an entire SoC from scratch. Instead, Google decided to find someone who could handle the parts of the design process that it did not have the capacity for, while its team worked on components that directly impacted the Pixel user experience.
Samsung LSI reportedly established a new “custom SoC” division in early 2020 to provide close design support to new chipset customers. The company was looking to leverage its vast library of IP blocks, relying on its proven track record across a range of products, ranging from CPUs to 5G modems. We don’t know exactly when Google approached Samsung (it would likely have been 2019 or earlier), but the nature and timing of Tensor and public awareness of Samsung’s custom chip venture is pretty good.
Initially, Google worked on smaller IP blocks and required an experienced chip designer to put the rest together.
Google would have given a list of its project requirements, which would have included the custom hardware it wanted to include and specific optimizations – for example, Tensor’s more unique 2+2+4 CPU core layout. Samsung began working on its part of the contract, using IP from its vast library to build the remaining SoCs for specific requirements. Meanwhile, Google began preparing to deliver its custom hardware parts (such as the Titan chip, Edge TPU, custom ISP elements, or a custom AV1 decoder) to Samsung for final integration. After that, Samsung began working on the physical design of the chip and eventually manufactured it in one of its fabs.
After that, Samsung helped Google with the first phase of work on the chip (called “the bringing-up”) and also helped them resolve issues and bugs. The company also provided software to Google for some of the design elements present in the chip.
Thanks to all this, Google was able to bring its first custom chip to market in much less time and with the smaller team it took to create Tensor independently.
Becoming (completely?) independent
Over the past few years, gChips has significantly expanded its team and experience. It has also expanded its IP block library. For example, the Tensor G2 introduced a DSP called GXP. Google is increasingly able to keep up with Samsung.
So, what exactly changes between Redondo and Laguna Beach? Google will now be responsible for all parts of the design process previously handled by Samsung.
It must first source the necessary hardware IP blocks, whether by designing them in-house or licensing them from a third party. There are many companies providing this type of services, such as Arm, Cadence, Synopsys, or… Samsung. It’s quite likely that Google will still use some of the IP originally used in Exynos chips in its “fully custom” chips, but the nature of the design will be far more customizable.
By moving away from Samsung, Google will have more control over future chipsets.
Google’s design team also has to deal with the verification and physical design of the chips, two extremely difficult aspects. Not only that, but it will be responsible for dealing with chip fabs (like TSMC) and other miscellaneous steps they face in the process. A lot of the software will need to be written from scratch, and Google will no longer have Samsung to ask for support.
What will this look like in practice? It’s unlikely we’ll see drastic changes in upcoming Tensor chips in the short term. The move to fully custom is an effort for freedom and will allow Google to customize its chips beyond those currently available, but it’s no miracle – it’s simply a logical next step in Tensor’s evolution. . However, one thing we do know for sure is that it will allow them to switch to TSMC’s foundry, which offers far more efficient manufacturing nodes than anything currently offered by Samsung. .
We’re excited to see what Google has in store for its future Tensor chips. While Redondo being delayed is undoubtedly disappointing, it will ultimately lead to an improved Tensor G5, which should be great for future Pixel buyers.
notes