Connect with us

Gadgets

The “Google Silicon” team gives us a tour of the Pixel 6’s Tensor SoC

Published

on

Enlarge / A promo image for the Google Tensor SoC.

Google

The Pixel 6 is official, with a wild new camera design, incredible pricing, and the new Android 12 OS. The headline component of the device has to be the Google Tensor “system on chip” (SoC), however. This is Google’s first main SoC in a smartphone, and the chip has a unique CPU core configuration and a strong focus on AI capabilities.

Since when is Google a chip manufacturer, though? What are the goals of Tensor SoC? Why was it designed in its unique way? To get some answers, we sat down with members of the “Google Silicon” team—a name I don’t think we’ve heard before.

Google Silicon is a group responsible for mobile chips from Google. That means the team designed previous Titan M security chips in the Pixel 3 and up, along with the Pixel Visual Core in the Pixel 2 and 3. The group has been working on main SoC development for three or four years, but it remains separate from the Cloud team’s silicon work on things like YouTube transcoding chips and Cloud TPUs.

Phil Carmack is the vice president and general manager of Google Silicon, and Monika Gupta is the senior director on the team. Both were nice enough to tell us a bit more about Google’s secretive chip.

Most mobile SoC vendors license their chip architecture from ARM, which also offers some (optional) guidelines on how to design a chip using its cores. And, apart from Apple, most of these custom designs stick pretty closely to these guidelines. This year, the most common design is a chip with one big ARM Cortex-X1 core, three medium A78 cores, and four slower, lower-power A55 cores for background processing.

Now wrap your mind around what Google is doing with the Google Tensor: the chip still has four A55s for the small cores, but it has two Arm Cortex-X1 CPUs at 2.8 GHz to handle foreground processing duties.

For “medium” cores, we get two 2.25 GHz A76 CPUs. (That’s A76, not the A78 everyone else is using—these A76s are the “big” CPU cores from last year.) When Arm introduced the A78 design, it said that the core—on a 5nm process—offered 20 percent more sustained performance in the same thermal envelope compared to the 7nm A76. Google is now using the A76 design but on a 5nm chip, so, going by ARM’s description, Google’s A76 should put out less heat than an A78 chip. Google is basically spending more thermal budget on having two big cores and less on the medium cores.

So the first question for the Google Silicon team is: what’s up with this core layout?

Carmack’s explanation is that the dual-X1 architecture is a play for efficiency at “medium” workloads. “We focused a lot of our design effort on how the workload is allocated, how the energy is distributed across the chip, and how the processors come into play at various points in time,” Carmack said. “When a heavy workload comes in, Android tends to hit it hard, and that’s how we get responsiveness.”

This is referring to the “rush to sleep” behavior most mobile chipsets exhibit, where something like loading a webpage has everything thrown at it so the task can be done quickly and the device can return to a lower-power state quickly.

“When it’s a steady-state problem where, say, the CPU has a lighter load but it’s still modestly significant, you’ll have the dual X1s running, and at that performance level, that will be the most efficient,” Carmack said.

He gave a camera view as an example of a “medium” workload, saying that you “open up your camera and you have a live view and a lot of really interesting things are happening all at once. You’ve got imaging calculations. You’ve got rendering calculations. You’ve got ML [machine learning] calculations, because maybe Lens is on detecting images or whatever. During situations like that, you have a lot of computation, but it’s heterogeneous.”

A quick aside: “heterogeneous” here means using more bits of the SoC for compute than just the CPU, so in the case of Lens, that means CPU, GPU, ISP (the camera co-processor), and Google’s ML co-processor.

Carmack continued, “You might use the two X1s dialed down in frequency so they’re ultra-efficient, but they’re still at a workload that’s pretty heavy. A workload that you normally would have done with dual A76s, maxed out, is now barely tapping the gas with dual X1s.”

The camera is a great case study, since previous Pixel phones have failed at exactly this kind of task. The Pixel 5 and 5a both regularly overheat after three minutes of 4K recording. I’m not allowed to talk too much about this right now, but I did record a 20 minute, 4K, 60 FPS video on a Pixel 6 with no overheating issues. (I got bored after 20 minutes.)

This is what the phone looks like, if you're wondering.
Enlarge / This is what the phone looks like, if you’re wondering.

Google

So, is Google pushing back on the idea that one big core is a good design? The idea of using one big core has only recently popped up in Arm chips, after all. We used to have four “big” cores and four “little” cores without any of this super-sized, single-core “prime” stuff.

“It all comes down to what you’re trying to accomplish,” Carmack said. “I’ll tell you where one big core versus two wins: when your goal is to win a single-threaded benchmark. You throw as many gates as possible at the one big core to win a single-threaded benchmark… If you want responsiveness, the quickest way to get that, and the most efficient way to get high-performance, is probably two big cores.”

Carmack warned that this “could evolve depending on how efficiency is mapped from one generation to the next,” but for the X1, Google claims that this design is better.

“The single-core performance is 80 percent faster than our previous generation; the GPU performance is 370 percent faster than our previous generation. I say that because people are going to ask that question, but to me, that’s not really the story,” Carmack explained. “I think the one thing you can take away from this part of the story is that although we’re a brand-new entry into the SoC space, we know how to make high-frequency, high-performance circuits that are dense, fast, and capable… Our implementation is rock solid in terms of frequencies, in terms of frequency per watt, all of that stuff. That’s not a reason to build an all-new Tensor SoC.”

Continue Reading

Gadgets

Apple reaches quiet truce over iPhone privacy changes

Published

on

Enlarge / A privacy notice appears on an iPhone 12 under the new iOS 14.5.1 operating system. Developers of an application have to ask for the user’s permission to allow cross-app tracking.

Picture Alliance | Getty Images

Apple has allowed app developers to collect data from its 1 billion iPhone users for targeted advertising, in an unacknowledged shift that lets companies follow a much looser interpretation of its controversial privacy policy.

In May Apple communicated its privacy changes to the wider public, launching an advert that featured a harassed man whose daily activities were closely monitored by an ever-growing group of strangers. When his iPhone prompted him to “Ask App Not to Track,” he clicked it and they vanished. Apple’s message to potential customers was clear—if you choose an iPhone, you are choosing privacy.

But seven months later, companies including Snap and Facebook have been allowed to keep sharing user-level signals from iPhones, as long as that data is anonymised and aggregated rather than tied to specific user profiles.

For instance Snap has told investors that it plans to share data from its 306 million users—including those who ask Snap “not to track”—so advertisers can gain “a more complete, real-time view” on how ad campaigns are working. Any personally identifiable data will first be obfuscated and aggregated.

Similarly, Facebook operations chief Sheryl Sandberg said the social media group was engaged in a “multiyear effort” to rebuild ad infrastructure “using more aggregate or anonymized data”.

These companies point out that Apple has told developers they “may not derive data from a device for the purpose of uniquely identifying it.” This means they can observe “signals” from an iPhone at a group level, enabling ads that can still be tailored to “cohorts” aligning with certain behavior but not associated with unique IDs.

This type of tracking is becoming the norm. Oren Kaniel, the chief executive of AppsFlyer, a mobile attribution platform that works with app developers, said that when his company introduced such a “privacy-centric” tool based on aggregated measurement in July 2020, “the level of pushback that we received from the entire ecosystem was huge.”

But now such aggregated solutions are the default for 95 percent of his clients. “The market changed their minds in a radical way,” he said.

It is not clear whether Apple has actually blessed these solutions. Apple declined to answer specific questions for this article but described privacy as its North Star, implying it was setting a general destination rather than defining a narrow pathway for developers.

Cory Munchbach, chief operating officer at customer data platform BlueConic, said Apple had to stand back from a strict reading of its rules because the disruption to the mobile ads ecosystem would be too great.

“Apple can’t put themselves in a situation where they are basically gutting their top-performing apps from a user-consumption perspective,” she said. “That would ultimately hurt iOS.”

For anyone interpreting Apple’s rules strictly, these solutions break the privacy rules set out to iOS users.

Lockdown Privacy, an app that blocks ad trackers, has called Apple’s policy “functionally useless in stopping third-party tracking.” It performed a variety of tests on top apps and observed that personal data and device information is still “being sent to trackers in almost all cases.”

But the companies aggregating user-level data said the reason apps continue to “leak” information such as a user’s IP address and location was simply because some require such information to function. Advertisers must know certain things such as the user’s language or the device screen size, otherwise the app experience would be awful.

The risk is that by allowing user-level data to be used by opaque third parties so long as they promise not to abuse it, Apple is in effect trusting the very same groups that chief executive Tim Cook has lambasted as “hucksters just looking to make a quick buck.”

Companies will pledge that they only look at user-level data once it has been anonymized, but without access to the data or algorithms working behind the scenes, users won’t really know if their data privacy has been preserved, said Munchbach.

“If historical precedent in adtech holds, those black boxes hide a lot of sins,” she said. “It’s not unreasonable to assume it leaves a lot to be desired.”

© 2021 The Financial Times Ltd. All rights reserved Not to be redistributed, copied, or modified in any way.

Continue Reading

Gadgets

Roku vs. Google drama winds down as companies forge multi-year YouTube deal

Published

on

Enlarge / Roku’s 4K Streaming Stick.

Roku

Roku and Google have arrived at a multi-year deal that will keep the YouTube and YouTube TV apps available on Roku’s devices, Roku announced on Twitter this morning. The agreement comes months after the YouTube TV app was pulled from the Roku Channel Store and just one day before the regular YouTube app would have been removed from the store.

Specific terms of the deal haven’t been announced, including how many years “multi-year” means and whether Roku will begin adding decoding support for the AV1 video codec to its hardware. We also don’t know whether the $65-per-month YouTube TV service will return to the Roku store as its own dedicated app or if it will continue to be rolled into the main YouTube app, as it has been since Google added it there to sidestep Roku’s restrictions in May.

Support for the AV1 codec has been one of the major sticking points between the two companies. The YouTube and YouTube TV apps use AV1 (which is backed by Google, among other companies) to deliver compressed 4K and 8K video streams. But because streaming devices tend to use slower, cheaper processors, they rely on dedicated video decoding hardware to be able to actually decompress and display those video files, and while most of these devices support the commonly used H.265/HEVC codec for high-resolution video streams, fewer support the royalty-free AV1 codec.

Roku has said that adding AV1 support to its devices would “increase consumer costs,” and requiring it for YouTube and YouTube TV support would effectively allow Google to dictate which chips Roku uses in its own products. Google has also accused Roku of using its position in the streaming-device market to secure more favorable terms (Roku’s devices account for a plurality of all streaming in North America, though its market share is lower in other regions). The YouTube and YouTube TV apps may not be able to stream high-resolution video on devices without AV1 support, though having those apps available in Roku’s store in any capacity is probably better for both companies than allowing them to be pulled entirely.

Continue Reading

Gadgets

Razer’s RGB smartphone cooler attaches to iPhones with MagSafe

Published

on

Enlarge / Razer Phone Cooler Chroma.

PC gamers know about heat. When you’re in the middle of an intense in-game battle, the last thing you want is for your computer to start acting up because your CPU or GPU got too hot. That’s why gamers and other extreme users rely on products like CPU coolers and liquid cooling systems. You probably haven’t been as concerned about your smartphone’s thermals while playing Candy Crush on your iPhone. Nevertheless, Razer released a new product, the Phone Cooler Chroma, on Tuesday to ensure your smartphone doesn’t overheat the next time you use it for gaming.

Of course, mobile gaming has grown beyond the likes of Candy Crush and Angry Birds. Razer (and some other vendors) have been trying to make mobile gaming a serious thing for a while. The company’s efforts are mostly focused on controllers, like the Razer Kishi, that attach to your smartphone. There’s also Razer’s finger sleeve for mobile gaming.

The Phone Cooler Chroma released Tuesday has a different purpose. Compatible with both iPhone and Android phones (it supports “most smartphones,” Razer’s product page claims), the product is meant to help keep your phone cool while it’s pushing those frames.

Interestingly, the fan takes advantage of Apple’s MagSafe, allowing you to attach the cooler magnetically. That’s convenient, but it also means the cooler won’t sit directly above the phone’s SoC.

If you don’t have a MagSafe-compatible phone, you can opt for the version with a universal clamp.

Clamp option.

We don’t know how adjustable the cooler is, but Razer says it works with phones that are 2.64-3.46 inches (67-88 mm) wide.

Staying cool?

1. RGB, 2. cover, 3. fan, 4. heatsink, 5. Peltier cooling tile, 6. cooling plate.
Enlarge / 1. RGB, 2. cover, 3. fan, 4. heatsink, 5. Peltier cooling tile, 6. cooling plate.

A cooling plate sits on the back cover and is topped by an electronic tile that uses Peltier cooling, also known as thermoelectric cooling, to transfer heat. The next layer is a heatsink under a seven-bladed fan spinning at up to 6,400 revolutions per minute, adjustable via Bluetooth. Razer says the cooler can stay at 30 dB.

On top of the fan lies a cover with air vents, and—of course—RGB lighting. Does the lighting help your phone stay cool? Absolutely not. But it almost wouldn’t be a Razer product without it. The gaming brand even put RGB on its N95 mask, so Chroma lighting here is no surprise.

RGB feels like a Razer requirement.
Enlarge / RGB feels like a Razer requirement.

There are 12 RGB LEDs in the cover, and each can be set to its own color and effect.

You’ll need a USB-C cable to power the Phone Cooler Chroma. The cooler comes with a 4.9-foot (1.5 m) USB-C to USB-C cable, but this seems like it could be burdensome when gaming on the go, as a mobile gamer is inclined to do.

Power over USB-C required.
Enlarge / Power over USB-C required.

Razer didn’t make any claims about how much cooler the product will keep your phone’s components. Unlike a CPU cooler, this cooler doesn’t come into direct contact with the processor, and it doesn’t have any exhaust vents to work with as some laptop fan coolers do. So the heat transfer from the actual SoC may be limited. Hardcore mobile gamers can find out for themselves for $60.

Ars Technica may earn compensation for sales from links on this post through affiliate programs.

Continue Reading

Trending