Google's Tensor inside of Pixel 6, Pixel 6 Pro: A Look into Performance & Efficiency
by Andrei Frumusanu on November 2, 2021 8:00 AM EST- Posted in
- Mobile
- Smartphones
- SoCs
- Pixel 6
- Pixel 6 Pro
- Google Tensor
Conclusion & End Remarks
Google’s newest Pixel 6 and 6 Pro are definitely most interesting devices, as in many ways they represent Google most competitive and value-rich phones the company has been able to make in years. While today’s article isn’t focusing on the device itself – more on that in a later review, including more in-depth camera coverage, what we did have a deeper look today was at the new chip powering the phones, the new Google Tensor.
The company notes that the primary reason they saw the need to go with a customized silicon approach, was that current merchant silicon solutions didn’t allow for the performance and efficiency for machine learning tasks that the company was aiming for in their devices. This performance and efficiency is used to enable new use-cases and experiences, such as the many ML features we see shipped and demonstrated in the Pixel 6 series, such live transcribing, live translation, and image processing tricks, all that run on the Tensor’s TPU.
While Google doesn’t appear to want to talk about it, the chip very clearly has provenance as a collaboration between Google and Samsung, and has a large amount of its roots in Samsung Exynos SoC architectures. While yes, it’s a customised design based on Google’s blueprints, the foundation means that some of the defining characteristics of Exynos chips is still found on the Tensor, particularly power efficiency is one area of the SoCs that are very much alike in, and that also means that the Tensor falls behind, much like the Exynos, against Qualcomm’s Snapdragon solutions when it comes to battery life or efficiency.
Google’s CPU setup is a bit different than other SoCs out there – a 2+2+4 setup with X1 cores, A76 cores and A55 cores is unusual. The two X1 cores are fine, and generally they end up where we expected them, even if there’s a few quirks. The A76 cores, ever since we heard those rumours months ago that the chip would feature them, made no sense to us, and even with the chip in our hands now, they still don’t make any sense, as they clearly fall behind the competition in both performance and efficiency. Who knows what the design process looked like, but it’s just one aspect of the chip that doesn’t work well.
GPU performance of the Tensor seems also lacklustre – while it’s hard to pinpoint wrong-doings to the actual SoC here, Google’s choice of going with a giant GPU doesn’t end up with practical advantages in gaming, as the phones themselves have quite bad thermal solutions for the chip, not able to properly dissipate the heat from the chip to the full body of the phones. Maybe Google makes more use of the GPU for burst compute workloads, but so far those were hard to identify.
So that leads us back to the core aspect of the Tensor, the TPU. It’s the one area where the SoC does shine, and very clearly has large performance, and likely also efficiency advantages over the competition. The metrics here are extremely hard to quantify, and one does pose the question if the use-cases and features the Pixel 6 comes with were really impossible to achieve, on say a Snapdragon chip. At least natural language processing seems to be Google’s and the Tensor’s forte, where it does have an inarguably large lead.
One further aspect that isn’t discussed as much is not related to the performance of the chip, but rather the supply chain side of things. We of course have no idea what Google’s deal with Samsung looks like, however both new Pixel 6 phones are devices that seemingly are priced much more aggressively than anything we’ve seen before from the company. If this is related to the SoC bill of materials is just pure speculation, but it is a possibility in my mind.
In general, I do think Google has achieved its goals with the Tensor SoC. The one thing it promises to do, it does indeed do quite well, and while the other aspects of the chip aren’t fantastic, they’re not outright deal-breakers either. I still think energy efficiency and battery life are goals of highest priority in a design, and there we just absolutely need to see better improvements in the next generation Tensor. We don’t know what path Google is taking for future designs, but it’ll be interesting to see.
We’ll be following up with a more in-depth review of the actual Pixel 6 phones, starting with a camera-focused article – stay tuned.
Facebook
Twitter
RSS
108 Comments
View All Comments
anonym - Sunday, November 7, 2021 - link
I don't have any data but A76 is more efficient than A78 while relatively lower performance region. According to following DVFS carves, A77 is out of the question.https://images.anandtech.com/doci/15813/A78-X1-cro...
boozed - Tuesday, November 2, 2021 - link
So do we call this design "semi-custom" or "very-slightly-custom"?watzupken - Wednesday, November 3, 2021 - link
I think we have come to a point that pushing performance for mobile devices is starting to slow down big time, or in some cases like Exynos where we see regressions. The SOC gets refreshed each year, pushing for higher performance. The fabs however is slower to catch up, and despite the marketing of 7nm, 5nm, 3nm, etc, they may not be anywhere near what is being marketed. In this case, squeezing a fat GPU sounds great on paper, but in real life, the sustained performance is not going to make a huge difference because of the power and heat. In any case, I feel the push for an annual SOC upgrade should slow down because I certainly don't see significant difference in real life performance. We generally only know that last years SOCs are slower only when running benchmarks. Even in games, last gen high end SOCs can still handle challenging titles. Instead, they should focus on making the SOCs more power efficient.damianrobertjones - Wednesday, November 3, 2021 - link
All I want is for all phones to be able to record the front and rear camera at the same time. VLog fun. Such a simple thing... .Whiteknight2020 - Wednesday, November 3, 2021 - link
Not India, China, UK, Russia, most of the EU, Africa. Which is the vast majority of the world's population and the vast majority of the world's phones, a great many of which are still feature phones.Whiteknight2020 - Wednesday, November 3, 2021 - link
Not India, China, UK, Russia, most of the EU, Africa. Which is the vast majority of the world's population and the vast majority of the world's phones, a great many of which are still feature phones.eastcoast_pete - Wednesday, November 3, 2021 - link
To me, one of the most interesting points about this "meh" first Google custom SoC is that it was created with lots of Lego blocks from Samsung; I guess Google working with Qualcomm was either out of the question or not something either was willing to do. Maybe this was about Google wanting to show QC that they can develop a Pixel smartphone without them, maybe the two compete too closely on ML/AI, or maybe they just don't like each other much right now - who knows? Still, an SD 888-derived SoC with Google TPU would have likely been better on performance and efficiency. This one here is an odd duck. As for the Pixel 6, especially the Pro: camera is supposed to be spectacular, but with the battery life as it is and, of course (Google, after all), no expandable storage and no 3.5 mm headphone connectors, it missed the mark for me. But, the Pixels are sold out, so why would Google change?Whiteknight2020 - Wednesday, November 3, 2021 - link
If you want a "really excellent camera", sorry to disappoint you but you'll need to be buying an actual camera. The only thing a multipurpose portable computing device can ever be excellent at is being a multipurpose portable computing device.FunBunny2 - Wednesday, November 3, 2021 - link
"a multipurpose portable computing device."isn't that pretty much verbatim what Stevie said when he showed the original iPhone? nothing has really changed since. it was, kinda, a big deal when Stevie intoned that the thingee incorporated 3, count em 3!, devices that you had to carry that day!!! internet, phone, and number 3 (whatever that was). is a 2021 smartphone really anything more?? I mean, beyond the capacity of more transistors. thank ASML (and some really smart physicists and engineers) for that not Apple or Samsung or Google or ... last time I checked Apple's 'our own ARM' SoC is just bigger and wider ARM ISA, due to the, so far, increasing transistor budget available at the foundries.
that all begs the fundamental question: if Apple and The Seven Dwarfs have access to the same physical capital (ASML, et al) why the difference? if everybody spends time and money tweaking a function (that they all need, one way or another), in some time (short, I'll assert) The One Best Way emerges. the task, in the final analysis, is just maths. of course, Best is not a point estimate, as many comments make clear; there're trade offs all along the line.
it would be fun to use one of the Damn Gummint's supercomputers (weather or nucular bomb design) to spec a SoC. wonder how different the result would be?
NaturalViolence - Wednesday, November 3, 2021 - link
The math for the memory bandwidth doesn't check out. From the article:"4x 16-bit CH
@ 3200MHz LPDDR5 / 51.2GB/s"
But 3200MHz x 64 bit is 25.6GB/s, not 51.2GB/s. So which is it?