Robert Triggs / Android Authority
Google’s Tensor G5, destined for next year’s Pixel 10, is perhaps the most anticipated chip in the series yet. For the first time, Google is designing the chip entirely in-house, without relying on Samsung to do most of the work. This is expected to make it more competitive as previous and even current Tensor models lag behind other chipset manufacturers. But will it actually happen? So far we haven’t heard anything about the specifications of the new chip, but that is finally changing today.
Thanks to a massive leak from Google’s gChips division, Android Authority has looked at credible documents from around the end of the Tensor G5’s design that tell us all about Google’s next-generation chip.
More weird CPU decisions
Pixel 9’s Tensor G4 had an upgraded CPU cluster compared to its predecessor – at least on paper as performance is only marginally faster in practice – just 6% faster in our multi-core tests. This is because Google dropped a core from the mid-cluster, likely in an effort to reduce power consumption.
The Tensor G5 upgrades the CPU cluster again, but not in a way you’d probably expect. Google has decided to keep the same primary core from the Arm Cortex-X4, which is an interesting choice considering the new Cortex-X925 promises some big improvements. It was also decided to shift the core clusters again: the middle cluster now has five Cortex-A725 cores, instead of three Cortex-A720 on Tensor G4, and the small cluster has been reduced accordingly to just two Cortex-A520 cores. I have put together the specifications below:
Tensor G3 (“zuma”) | Tensor G4 (“zumapro”) | Tensor G5 (“laguna”) | |
---|---|---|---|
Large cluster |
Tensor G3 (“zuma”)
1x Arm Cortex-X3 |
Tensor G4 (“zumapro”)
1x Arm Cortex-X4 |
Tensor G5 (“laguna”)
1x Arm Cortex-X4 |
Center cluster |
Tensor G3 (“zuma”)
4x Arm Cortex-A715 |
Tensor G4 (“zumapro”)
3x Arm Cortex-A720 |
Tensor G5 (“lagoon”)
5x Arm Cortex-A725 |
Small cluster |
Tensor G3 (“zuma”)
4x Arm Cortex-A510 |
Tensor G4 (“zumapro”)
4x Arm Cortex-A520 |
Tensor G5 (“laguna”)
2x Arm Cortex-A520 |
Obviously it’s hard to predict performance based on specs alone, but a change like this should give the Tensor G5 at least a decent multi-core performance boost. However, it’s disappointing to see the same Cortex-X4 being used again.
A brand new GPU
One area that particularly surprised me about the Tensor G5 while reviewing the docs was the GPU. There aren’t many vendors providing GPU IP, and Google uses Arm Mali for all its previous Tensor chips, so why would that change now? Well, I don’t know, but it was true. The Tensor G5 comes with a GPU from Imagination Technologies (or IMG) – the DXT-48-1536 running at 1.1 GHz.
Unfortunately, we don’t know much about the new GPU, aside from two interesting details: first, it will come with ray tracing support, which is new for Google chips, which typically skip such “gaming” features. The second and perhaps equally important detail is the support for GPU virtualization, which allows the use of accelerated graphics in a virtual machine. Google has been working on various virtualization-based features for a while now, so it makes sense that this is included.
I’ve collected the new GPU specs in the table below:
Tensor G3 (“zuma”) | Tensor G4 (“zumapro”) | Tensor G5 (“laguna”) | |
---|---|---|---|
GPU |
Tensor G3 (“zuma”)
Arm Mali-G715 (7 cores) |
Tensor G4 (“zumapro”)
Arm Mali-G715 (7 cores) |
Tensor G5 (“lagoon”)
IMG DXT (2 cores) |
Frequency |
Tensor G3 (“zuma”)
890MHz |
Tensor G4 (“zumapro”)
940MHz |
Tensor G5 (“laguna”)
1100MHz |
Ray tracing |
Tensor G3 (“zuma”)
Not supported |
Tensor G4 (“zumapro”)
Not supported |
Tensor G5 (“lagoon”)
Supported |
GPU virtualization |
Tensor G3 (“zuma”)
Not supported |
Tensor G4 (“zumapro”)
Not supported |
Tensor G5 (“lagoon”)
Supported |
Modest improvements for AI
![Google Pixel 9 with Pixel Studio app open and a prompt to download a new AI image model Google Pixel 9 with Pixel Studio app open and a prompt to download a new AI image model](https://www.androidauthority.com/wp-content/uploads/2024/08/Google-Pixel-9-with-Pixel-Studio-app-open-showing-a-prompt-to-download-a-new-AI-image-model.jpg)
C. Scott Brown / Android Authority
Google’s phones have always had great AI features, and AI is clearly one of the main reasons Tensor chips exist in the first place. By adding custom-designed TPUs, Google can achieve much more immersive experiences than would otherwise be possible with an off-the-shelf chip.
The Tensor G5 comes with a modestly faster TPU compared to the Tensor G4 (which itself was exactly the same as the Tensor G3). The TOPS (Trillions of Operations Per Second) value is almost 40% greater, but that doesn’t translate well to real-world performance. Google’s internal benchmarks suggest the new TPU is only 14% faster. The improved TPU also comes with a few new features for Google’s developers, such as small embedded RISC-V cores to enable operations not implemented in the hardware, as well as support for on-device training. I’ve compiled the new TPU specs below:
Tensor G3 (“zuma”) | Tensor G4 (“zumapro”) | Tensor G5 (“laguna”) | |
---|---|---|---|
TOPS |
Tensor G3 (“zuma”)
13 / 6.5 TOPS (INT8/FP16) |
Tensor G4 (“zumapro”)
13 / 6.5 TOPS (INT8/FP16) |
Tensor G5 (“lagoon”)
18 / 9 TOPS (INT8/FP16) |
Performance improvement over the previous generation |
Tensor G3 (“zuma”)
+65% |
Tensor G4 (“zumapro”)
0 |
Tensor G5 (“lagoon”)
+14% |
As we’ve already revealed, the Tensor G5 is built on TSMC’s 3nm-class N3E process node, similar to the likes of Apple’s A18 Pro. Interestingly, the die size is 121mm^2, while the Apple A18 Pro’s is only 105mm^2, making the G5 a noticeably larger chip.
While the specs of Google’s upcoming Tensor G5 don’t look super impressive on paper, it’s worth remembering that it’s still a chip that will be in a Pixel phone, with software arguably being more important than the hardware . It will be interesting to see how Google uses the new hardware to add new features to the Pixel 10 series.
What do you think of the leaked Tensor G5 specs?
202 votes