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
Speedfriend - Thursday, November 4, 2021 - link
The average laptop costs $500 and most expensive laptops are bought by enterprises where Mac OS has a limited share. While the Macbookz are great devices, they are hobbled by poor monitor support at the Air end and cray prices at the MacBook Pro end. For most users the difference between the performance of a MacBook Pro and a $1000 laptop is unnoticeable except in their wallet!dukhawk - Tuesday, November 2, 2021 - link
The chip is very Exynos design related. Looking through the kernel source and there are a ton of Exynos named files.dukhawk - Tuesday, November 2, 2021 - link
https://android.googlesource.com/device/google/rav...defaultluser - Tuesday, November 2, 2021 - link
If anyone wants to know know why Nvidia is most interested in purchasing ARM, it's in order to put the inefficient Mali out of it's misery - and simultaneously replace it with their own license-able Geforce cores!Since ARM Corp started throwing in the GPU for free, they've had to cut GPU research (to pay for the increasingly complex CPU cores, all of which come out of the same revenue box!) But Nvidia has the massive Server Revenue to handle this architecture-design mismatch; they will keep the top 50% of the engineers, and cut the other cruft loose!
melgross - Tuesday, November 2, 2021 - link
That may be a side effect. But the reason for purchasing g it would be maki g money, and controlling the market. Yes, it’s true that Nvidia wa t to control all graphics and to turn the GPU into the main programming aim.TheinsanegamerN - Tuesday, November 2, 2021 - link
If nvidia wanted to do that they could simply license ARM and make their own superior chip. The fact they have fallen flat on their face every time they have tried speaks volumes.they want ARM for patents and $$$, nothing more.
defaultluser - Wednesday, November 3, 2021 - link
When a place like Rockchip can sell an Arm chip bundled with Mali for Peanuts, you can understand why superior GPU wasn't enough to win Phone customers!You also need integrated modem if you ever want to compete with Qualcomm (not something Nvidia was willing to do).
But that bundling system has been shorting ARM Mali development for years (Qualcomm, Apple, and soon Samsung (via AMD) are all bringing better high-end options into the field - you know your performance/watt must be pathetic when a company like Samsung is getting desperate-enough to pay the cost of porting AMD GPU over to ARM architecture.
Kvaern1 - Sunday, November 7, 2021 - link
"If nvidia wanted to do that they could simply license ARM and make their own superior chip."''simply'
No, no one can simply do that anymore and only two companies can. NVidia just bought one of them.
melgross - Tuesday, November 2, 2021 - link
I’m wondering about several things here.I don’t see the reason for using the A76 cores being one of time. This is a very new chip. The competitors on the Android side have been out for a while. They use A78 cores. Samsung uses A78 cores. So time doesn’t seem to be the issuer here, after all it does use the X1. So I wonder if it isn’t the size of the core on this already large, crowded chip that’s a reason, and possibly cost. If the newer cores take up more area they would cost slightly more. These chips are going to be bought in a fairly small number. Estimates have it that last year, Google sold between 4 and 7 million phones, and that they’re doubling this year’s order. Either would still be small, and give no advantage to Google in volume pricing compared to other chip makers.
The second is that you have to wonder if Google is following the Apple road here. Apple, of course, designs many chips, all for their own use. Will Google keep their chips for their own use, assuming they’re as successful in selling phones as Google hopes, or will they, after another generation, or two, when the chip is more of their own IP, offer them to other Android phone makers, and if so, how will Samsung feel about that, assuming their contract allows it?
SonOfKratos - Tuesday, November 2, 2021 - link
I think they went for the A76 cores because of cost, like you said Tensor is already huge and the A78 or A77 cores would be more power efficient but they are also much bigger than the A76 on 5nm process. Even if they were to clock an A78 lower it would just be a waste of money and space on the chip for them. They probably had a specific budget for the chip which meant a specific die size. This is not Apple who is willing to throw as much money as they can to get the best performance per watt.The display was rumored to be an E5 display from Samsung display which is in their latest display so I don't know why Google is not pushing for higher brightness but it could be because of heat dissipation as well...I highly doubt Samsung gave Google their garbage displays lol Also Google does not utilize the variable refresh rate very well and it's terrible for battery life. I have also seen a lot of janky scrolling with 120Hz in apps like Twitter..it has hiccups scrolling through the timeline compared to my Pixel 3.
The modem is very interesting probably more so than Tensor, this is the first competition for Qualcomm in the US at least. A lot of people have been saying that the modem is integrated in Tensor but why would Google integrate a modem that does not belong to them in "their" chip? That's like asking Apple to integrate Qualcomm modems in their chip. Also Samsung pays Qualcomm royalties for 5G so they probably have a special agreement surrounding the sale and implementation of the modem. It is definitely not as power efficient as Qualcomm's implementation but it's Good start. I got 400+ Mbps on T-Mobile 5GUC outdoors and 200 Mbps indoors (I don't know which band). It surprisingly supports n258 band like the iPhone.