Details about Google’s upcoming Tensor chips leaked this week, including deeper insights into the Pixel 10’s Tensor G5 processor.
First up, Android Authority published several reports based on details it obtained from documents from Google’s gChips division. To start, the publication reported that Google will drop Samsung and move to designing chips internally. As part of that shift, the company will move to TSMC for manufacturing. TSMC is currently a leader in this space, making chips for the likes of Apple and AMD.
The Tensor G5 (codenamed ‘laguna’), which will power next year’s Pixel 10 devices, will likely be the first manufactured by TSMC. Android Authority reports it will use TSMC’s 3nm class N3E process — the same Apple used for the iPhone 16 Pro’s A18 Pro and MacBook M4 chips. The Tensor G6 (codenamed ‘malibu’) is expected to use TSMC’s N3P node, the same rumoured for Apple’s A19 chip.
These changes should bring efficiency and performance improvements to the Tensor chips, though it remains to be seen how significant of an improvement we’ll get. Another report from Android Authority digs into the Tensor G5 in greater details, and it looks like Google is making some weird choices.
Minimal hardware improvements
On the CPU side, the Tensor G5 reportedly will continue using a single Arm Cortex-X4 for the ‘big’ core, which is also what we saw on the Tensor G4 and G3. (In the CPU, it’s increasingly common to use a mix of big and little cores to balance power and efficiency.) Given the new Cortex-X925 promises significant improvements, it’s a bummer to see Google stick with the X4.
However, the company is making more significant changes elsewhere in the CPU. Android Authority reports that the G5’s ‘mid’ CPU cluster will use five Cortex-A725 cores, up from three A720 cores on the G4 and four A715 cores on the G3. The G5 will also sport a smaller ‘little’ cluster with just two Cortex-A520 cores, down from four A520 cores on the G4.
It’s tough to say what this will mean for performance just from these details, but the increased mid-cluster could help boost multi-core performance. Ultimately, we’ll need to wait until the Tensor G5 lands in the hands of testers.
Elsewhere, the Tensor G5 reportedly will include a GPU from Imagination Technologies (IMG) instead of using the Arm Mali GPU like past Pixels. Specifically, the chip will feature the DXT-48-1536 GPU clocked at 1.1GHz with support for ray tracing and GPU virtualization. However, the IMG GPU only has two cores. The Mali-G715 in the G4 was clocked at 940MHz but had seven cores. It did not support ray tracing or GPU virtualization.
Finally, that brings us to the AI hardware. One of the benefits of Google’s Tensor chips is that they feature custom-design TPUs, which Google says allows it to power on-device AI and machine learning (ML) features that it couldn’t do with other hardware.
Android Authority reports that the Tensor G5 will feature a faster TPU capable of up to 18 trillion operations per second (TOPS). However, the company’s internal benchmarking shows only a 14 percent speed improvement despite a 40 percent increase in TOPS.
Another interesting note is that the new TPU features small, embedded RISC-V cores that enable running operations not implemented in hardware and support for on-device training.
What does it all mean?
It’s always a bit tough to translate technical details to real-world benefit to users, especially when we’re going off early leaks and don’t have the hardware to actually test. However, we can glean a few details. First, the Tensor G5 will likely boast improved performance and efficiency over the G4, though it will likely still fall behind the latest Qualcomm Snapdragon and Apple A-series chips.
That being said, Google’s Pixel phones also tend to work quite well even with underpowered hardware, as we’ve seen with the last few Tensor chips. Still, with Pixels increasingly costing more, it’d be nice to see Google make more gains on performance to keep up with the competition.
The G5’s TPU improvements could also open the path for some new AI features, so it’ll be interesting to see what Google does with the extra performance.
Source: Android Authority, (2)
MobileSyrup may earn a commission from purchases made via our links, which helps fund the journalism we provide free on our website. These links do not influence our editorial content. Support us here.
