LightBlog

mercredi 2 décembre 2020

Samsung Internet 13.0 introduces beefed up privacy and security features

Samsung has begun rolling out Samsung Internet 13.0, which the company said introduces features that will provide users with a safer browsing experience.

One of the big new features is known as “Secret mode.” This new mode allows users to automatically clear their browsing history as soon as all Secret mode tabs are closed. Users will see an icon to indicate when Secret mode is enabled.

The update will also warn users when they visit a potentially malicious website. “Samsung Internet 13.0 offers a new permission request UI that will display a warning message if a website seems malicious and is attempting to trick users into allowing notifications.” Samsung Internet

Permission Request UI – Before (left) and Samsung Internet 13.0 (right)

In addition to Samsung Internet 13.0’s improved privacy and security features, the update also provides users with a better user experience:

  • You can now use High contrast mode with Dark mode on, to make fonts and other components stand out even more.
  • Samsung Internet 13.0 also introduces an Expandable App Bar for menus such as Bookmarks, Saved pages, History and Downloads.
  • Get more screen space by hiding the status bar, to immerse yourself in the content you’re browsing.
  • When watching a video in full screen with Video assistant, pause it by double tapping the middle of the screen.
  • Easily edit the title of your bookmarks so that they’re easier to recognize and search.

Finally, Samsung said new Application Programming Interface (API) modules are now available for Samsung Internet 13.0, which will give developers the opportunity to use these APIs to build extensions. Some of the new modules include WebRequest, Proxy, Cookies, Types, History, Alarms, Privacy, and more. Interested developers can learn more about developing a third-party extension by visiting this link.

Samsung Internet 13.0 was first detailed when Samsung released a beta It’s been a few months, but now the features are available for everyone.

Samsung Internet Browser (Free, Google Play) →

The post Samsung Internet 13.0 introduces beefed up privacy and security features appeared first on xda-developers.



from xda-developers https://ift.tt/3mytBQi
via IFTTT

Sentons CameraBar uses ultrasound to turn your phone’s frame into a zoom slider

Smartphones are incredibly versatile portable PCs, but with only a handful of physical buttons available on most devices, you have to rely on touchscreen controls for most things. When it comes to playing games or using the camera app, you have to juggle between many different onscreen buttons and sliders, resulting in a subpar, limiting experience due to limited screen real estate and awkward hand ergonomics. A company called Sentons wants to change this reality by introducing what they’re calling “Software-Defined Surfaces” (SDS) in place of physical buttons. Today, they’re introducing CameraBar, a new SDS that uses ultrasound to detect taps and slides on the frame of a phone to mimic the physical shutter and zoom buttons on a traditional camera.

With CameraBar, users can avail of virtual shutter and zoom controls without their fingers obstructing the view by touching the screen. The default configuration of CameraBar is to listen for a light press on the right side to set the focus, a hard press on the right side to snap a picture, and a slide-to-zoom on the left side for optical zoom. The video embedded below demonstrates CameraBar in action on a retail ASUS ROG Phone 3 unit as well as on custom development hardware.

The ROG Phone 3 shown in the video above is presumably running custom firmware to allow for Sentons’ custom camera app to react to inputs from the sensors, as the AirTriggers feature on the ASUS ROG Phone cannot currently be mapped to any actions in the stock ASUS Camera app. For this feature to make its way to the ROG Phone, ASUS will have to add support for it through a software update.

While the ROG Phone 3 can technically add support for the gestures shown in this demonstration, Sentons CTO Sam Sheng told XDA that the ideal device to feature CameraBar will have a larger area for sliding to allow for more fine-grained control of the zoom level. No such device currently exists on the market, though Sentons is in talks with several undisclosed partners who are taking this technology to production shortly. The company is providing OEMs with recommended sensor topology, guidance on how to design the module, and reference software on how to implement this as part of the stock camera app. OEMs can customize the gesture activation region, and if they choose to do so, they can also extend the same customization options to the consumer.

Eventually, it’s believed that OEMs making new smartphones with all-screen designs and “waterfall” displays will be the first to adopt Sentons’ new CameraBar technology, though as previously mentioned, smartphones that have implemented Sentons’ existing GamingBar technology (which includes the ROG Phone 3 and Lenovo Legion Phone Duel) can inherit functions of CameraBar.

Replacing Buttons with Ultrasound

Buttons are a common point of failure in smartphones and a hindrance to achieving a truly all-screen design, so it makes sense for smartphone manufacturers to attempt to get rid of them. The only problem is finding a worthwhile alternative to a physical button, and we’ve seen a few lackluster attempts at replacing them in the past. Huawei’s Mate 30 Pro used “invisible” touch buttons for the volume rocker which some users struggled to trigger. HTC’s U12+ featured faux buttons that were similarly frustrating for some reviewers. While Huawei tried to implement its volume keys capacitively, HTC used Sentons’ ultrasonic sensors, though I’m told HTC used a simple strain-gauge sensor. In contrast, the ROG Phone models from ASUS can sense much lighter touches, under 5 grams-force. Although I haven’t had the opportunity to test the HTC U12+ myself, my experience with the ROG Phone 3 and its customizable AirTriggers gestures has been mostly positive, so I’m looking forward to seeing how technology from Sentons can not only replace the buttons on phones but also augment their functionality.

So how exactly do OEMs actually replace a button with Sentons’ tech? Replicating a physical button on a smartphone using ultrasound involves combining piezoelectric and strain-gauge sensors. Sentons likens its technology to sonar, which uses ultrasonic waves for echolocation. The time-of-flight of the vibration field created by the piezoelectric sensors is used to uniquely determine the position of the user’s finger, and the coupling of the finger and substrate that’s vibrating is used to determine the force from the vibrating sound wave. In other words, ultrasonic waves help identify the location, while a strain-gauge sensor determines the level of force applied.

Source: Sentons

Thus, the principles behind the technology aren’t new, but what Sentons is selling to OEMs is its line of SDSwave force-and-touch processors, its machine learning algorithms to weed out false touches from taps and gestures, and its ultrasonic strain-gauge sensor. The piezoelectric sensors, though, can be off-the-shelf, making them very inexpensive to incorporate into the smartphone design. So long as the material used in the smartphone body is stiff enough, and thus allows for ultrasonic waves to propagate, it can be turned into a virtual touch sensor.

Sentons says its ultrasonic sensors can recognize finger taps through glass, plastic, and even millimeters of aluminum, meaning the sensing elements can be mounted on the phone’s mid plate rather than right behind where the finger is expected to be placed. The caveat, though, is that this can only be done when the smartphone maker wishes to replace “lower performance” buttons like volume or power buttons — replicating gestures that need more precision, such as a slider, will generally require the sensing element to be mounted on the sidewall behind the contact point. These sensing elements are said to be very, very tiny and can easily be slotted in between antenna elements (such as mmWave antennas placed around the body of a 5G smartphone), and since there are no wires involved, there won’t be any degradation of the antenna performance.

The small size of the sensing sensors even makes it possible for them to be used in devices as small as smartwatches and hearables (like true wireless earbuds). For smartwatches, ultrasonic gestures could be used to replace a physically rotating crown or a touch-sensitive capacitive bezel. For true wireless earbuds, ultrasound could bring us better tap and gesture detection for music controls. Sentons is currently experimenting with implementing their technology in more form factors, with even automotive uses on the table, but there haven’t been any commercial products (outside of smartphones) to use their technology just yet. But Sentons is far from the only company using machine learning to analyze ultrasound for use in virtual smart sensors—there’s also Elliptic Labs which has partnered with multiple smartphone makers for its ultrasonic proximity detection tech—so there’s a good chance ultrasound will stick around and become even more widely adopted.

The post Sentons CameraBar uses ultrasound to turn your phone’s frame into a zoom slider appeared first on xda-developers.



from xda-developers https://ift.tt/39Dn3Mu
via IFTTT

Samsung Galaxy S20 FE and Galaxy S9 receive updates with December 2020 security patches

For long, Samsung has had to face criticism for its sluggish approach in rolling out software updates, especially for older devices. That has changed to a fair bit in 2020, as nowadays the company frequently outperforms Google when it comes to delivering monthly security patch updates. For example, the Galaxy S20 lineup was the first to receive the December 2020 Android security patches nearly three weeks ago as part of the One UI 3.0 public beta program. Now, Samsung has brought the December 2020 patchset to the stable channel of One UI, starting with the Galaxy S9 and the Galaxy S20 FE.

Samsung Galaxy S9 XDA Forums || Samsung Galaxy S9+ XDA Forums

Samsung Galaxy S20 FE XDA Forums

Tagged with the version number G96xFXXSCFTK2, the new firmware for the Galaxy S9 series is intended for the Exynos 9810-powered global variants of the regular S9 (model number SM-G960F) and the S9+ (model number SM-G965F). The FOTA is available in the DBT region, which is the Korean OEM’s internal code for Germany. For the Galaxy S20 FE, it’s the 5G variant (model number SM-G781B) that’s getting the update in the form of software version G781BXXS1ATL1 in the United Kingdom, Spain, and Germany.

We have yet to locate the official changelog for these builds, but the software version numbers are sufficient to conclude that there is no new feature apart from the obvious bump in the Android security patch level. The bootloader version also remains unchanged in both cases, which means software-based downgrading is possible in theory.

You can always skip the waiting queue and download the updated firmware package directly from the Samsung update server right now using tools like Samloader. Unless you’re rocking the Snapdragon Galaxy S9 or the 4G variant of the Galaxy S20 FE, you can take the path of installation using Odin, but only do so if you’re comfortable with manual flashing.

At the time of writing this article, Samsung’s Mobile Security portal has not been updated with the details of the December 2020 security bulletin.

The post Samsung Galaxy S20 FE and Galaxy S9 receive updates with December 2020 security patches appeared first on xda-developers.



from xda-developers https://ift.tt/2Jy1wtI
via IFTTT

The Qualcomm Snapdragon 888 will power flagship 5G phones in 2021 – Here’s what you need to know

During Day 2 of its annual Tech Summit, Qualcomm unveiled the chip that will power the majority of 2021 Android flagships. A successor to the Snapdragon 865, the Snapdragon 888, as expected, brings major improvements in the CPU, GPU, DSP, ISP, modem, and a lot more. It features the new Kryo 680 CPU, Adreno 660 GPU, 6th generation AI Engine with the Hexagon 780 DSP, Spectra 580 ISP, Quick Charge 5, and the Snapdragon X60 modem-RF system.

The Snapdragon 865 enjoyed a successful 2020 as it was featured in the majority of this year’s flagship phones, and the Snapdragon 888 will build on its success. Qualcomm has already confirmed that 14 device makers will build smartphones with it. Let’s take a look at its new features one-by-one, as there is a lot to unpack here.

Qualcomm Snapdragon 888 block diagram

Source: Qualcomm


Table of Contents

  1. CPU
  2. GPU
  3. Modem & Connectivity
  4. Camera
  5. AI Engine & DSP
    1. Qualcomm Sensing Hub
    2. AI software
  6. Gaming
    1. Qualcomm Game Quick Touch
    2. Variable Rate Shading
  7. Security
  8. Comparison with Snapdragon 865 and 855
  9. Full Specifications & Features List
  10. Conclusion

Snapdragon 888 CPU: Kryo 680

Qualcomm has been reminding the industry for the last few years that its SoCs are more than just a CPU with a GPU. However, the CPU and GPU remain the most important areas of an SoC. To that end, the Snapdragon 888 brings the new Kryo 680 CPU, which brings 25% performance improvements over the Snapdragon 865’s Kryo 585, according to the company. The 25% performance improvements are brought by IPC improvements in the CPU core architectures as well as the benefits of being manufactured on a more efficient 5nm process node (which is expected but not confirmed to be Samsung Foundry’s 5nm LPE process).

The Snapdragon 888 features an octa-core CPU, with 1x Kryo 680 Prime, 3x Kryo 680 Performance, and 4x Kryo 680 Efficiency cores. The DynamIQ System Unit (DSU) has 3MB system cache as well as 4MB L3 cache.

Kryo 680 CPU on the Qualcomm Snapdragon 888

Source: Qualcomm

The Kryo 680 Prime core features the ARM Cortex-X1, which was announced by ARM in May 2020 as the first CPU core under the Cortex-X Custom program (CXC). The Cortex-X1 specifically aims to break away from the Cortex-A series in terms of PPA, as it’s intended to be a larger, more performant, and more power-hungry core. It has the ambitious goal of taking on Apple’s custom high-performance cores in the A-series. With a 5-wide decode width and a more complex back-end, the Cortex-X1 represents ARM’s most ambitious big CPU core yet, and Qualcomm is the first to adopt it in a mobile SoC with the Snapdragon 888.

The Prime core is clocked at 2.84GHz, which is a bit disappointing as it means ARM’s 3GHz clock speed projection for the Cortex-X1 won’t come true yet again, at least initially. It has 1MB L2 cache. Despite the 5nm process, the Cortex-X1 Prime core has the same clock speed as the last-generation Cortex-A77 Prime core. Qualcomm increased the Prime core’s clock speed to 3.1GHz in the mid-cycle Snapdragon 865 Plus refresh, so it may be that the same is on the cards with this new generation. Apple’s Firestorm core is clocked at 2.89GHz-3GHz (depending on per-core clock speed), for reference. With its IPC advantage, the Apple A14 will still have a single-threaded performance advantage (higher clock speed + higher IPC). ARM has narrowed the gap as the Snapdragon 888 should be competitive with the Apple A13 unlike previous generations where ARM was essentially two years behind, but the gap will still exist.

The three Kryo 680 Performance cores use ARM’s Cortex-A78 design. The Cortex-A78 is a more traditional ARM big core with a 4-wide decode width that focuses on the company’s traditional strength of PPA. It features a 7% IPC improvement over the Cortex-A78, with 13% additional performance gains being achieved thanks to the 5nm process fabrication. The Cortex-A78 cores are clocked at 2.4GHz and have individual 512KB L2 caches. The A78’s design goal is well-targeted for the function of middle cores in a flagship chip.

Finally, the three Kryo 680 Efficiency cores are still based on the aging three-year-old ARM Cortex-A55 design, as ARM hasn’t announced a successor to its little core yet. The cores are clocked at 1.8GHz and feature individual 128KB L2 caches. This is another area where Apple is much ahead, as the A14’s Ice Storm little cores are much faster (4x) as well as more energy-efficient (3x) than the Cortex-A55 cores featured in all Android flagships.

In terms of memory bandwidth, the Snapdragon 888 supports LPDDR5 memory at up to 3200MHz, and LPDDR4 memory up to 2133MHz, with a maximum of 16GB RAM.

Overall, the Snapdragon 888’s CPU represents a solid, but incremental step forward for Qualcomm. The company hasn’t made any custom CPU core since the original Kryo core in 2016, so it’s dependent on ARM to make steps forward. The 1x Cortex-X1 + 3x Cortex-A78 combination seems a good fit to balance performance and power draw, although the single-threaded mobile CPU performance crown will still remain out of reach for Qualcomm. The clock speed of the Prime core is a bit conservative, but that should mean reduced power levels. This is more a reflection of Apple’s outstanding CPU cores rather than an indictment of ARM’s CPU cores, which still remain great in a vacuum. The Snapdragon 888 should be about 25% slower than the A14 in single-threaded CPU performance. If it does achieve parity with the A13’s single-threaded performance, it means it can potentially stand head-to-head or even outperform Intel’s Tiger Lake CPU core as well as AMD’s Zen 2 core in terms of IPC.


Snapdragon 888 GPU: Adreno 660

In the Android SoC market, Qualcomm has long been the leader when it comes to GPU performance with its custom Adreno GPUs. There was a time when it was also competitive with the GPUs featured in Apple’s A-series, but since 2017’s Apple A11 and the start of Apple’s custom GPUs, it hasn’t been able to keep up either in terms of peak or sustained performance. With respect to its competitors in the Android SoC market, Qualcomm’s Adreno GPUs are still best-in-class compared to ARM’s Mali GPUs, which have worse peak performance, sustained performance, as well as power efficiency.

So on the one hand, Qualcomm can afford to take things easy and build on its lead in the Android market. However, Apple’s GPUs have been consistently getting faster and more efficient, and they have been getting faster and more efficient at a significantly quicker rate than the Adreno GPUs, to the point where the Apple A14’s custom GPU is essentially two generations ahead of the Snapdragon 865’s Adreno 650 GPU. It is here where Qualcomm needed to make big improvements with the Snapdragon 888’s GPU, but unfortunately, the company hasn’t quite delivered.

The Snapdragon 888 features the new Adreno 660 GPU, which features 35% faster graphics rendering than the previous generation. It’s also said to be 20% more power efficient. Qualcomm’s Adreno GPUs largely remain a black box, as the company doesn’t reveal many details. The GPU nomenclature means the Adreno 660 still isn’t the fastest Adreno GPU Qualcomm has ever made. Instead, that honor still belongs to the Adreno 680 GPU, which was featured in 2019’s Snapdragon 8cx SoC for Always On, Always Connected PCs. It’s not an apples-to-apples comparison as the Snapdragon 8cx isn’t intended for smartphones, but it still shows that Qualcomm could have aimed higher this generation in order to take on Apple.

As it is, the numbers mean that the Adreno 660 in the Snapdragon 888 will still fall below the Apple A14’s four-core GPU in terms of both peak and sustained performance, as well as power efficiency. It may even fail to match the A13 GPU’s peak performance, which means that Qualcomm will still remain two generations behind. Relative to the Mali-G78 GPU expected to be featured in the upcoming Exynos 2100 SoC as well as the next MediaTek Dimensity flagship SoC, the Snapdragon 888 will still enjoy a sizable advantage. Therefore, the competitive GPU landscape will still be the same in 2021: Apple will be on top with quite a lot of room to spare, Qualcomm will enjoy the top spot in the Android SoC market, while flagship SoCs featuring Mali GPUs will occupy the bottom spot. The Adreno 660 represents a respectable 35% performance improvement in a vacuum, but it won’t be enough to match Apple’s GPU efforts.

In terms of display enhancements, the Adreno 680 brings enhancements for OLED display uniformity, picture quality improvements, as well as de-mura and subpixel rendering.


Snapdragon 888 connectivity: Integrated Snapdragon X60 modem-RF system and FastConnect 6900

The Snapdragon 888 brings an integrated 5G modem, which is big news in itself. The Snapdragon 865 was an outlier last year as it didn’t have an integrated 4G or 5G modem, as device makers were forced to buy the Snapdragon X55 5G modem-RF system alongside the SoC to provide connectivity. This meant that flagship and affordable flagship phones got much more expensive in 2020, as the combined price of the SoC and the X55 modem-RF system was higher than the Snapdragon 855. It also resulted in the fact that the majority of 2020 flagship Snapdragon 865 phones featured 5G support, with the exception of outliers such as the iQOO 3 4G and the Sony Xperia 1 II’s U.S. variant.

With the Snapdragon 888, on the other hand, Qualcomm has gone back to an integrated modem. The Snapdragon X60 modem-RF system was announced in February 2020 as Qualcomm’s third-generation 5G modem, and it succeeds the X55. The integrated 5G modem should lead to theoretical power savings as well as lower cost for device makers, but it remains to be seen if this plays out in practice.

Qualcomm Snapdragon X60 modem

Source: Qualcomm

We did a deep dive on the Snapdragon X60 back in February, so readers should check that out. In brief, the Snapdragon X60 modem-RF system brings 5G carrier aggregation across FDD and TDD, which is a 1st for 5G modems. The peak downlink speeds are increased to 7.5Gbps for mmWave and 5Gbps for sub-6GHz, while the peak uplink speeds are 3Gbps. The X60 features Global 5G multi-SIM, which is a unique feature according to Qualcomm.

The Snapdragon 888 also features the Qualcomm FastConnect 6900 system for Wi-Fi and Bluetooth. This was first featured in the Snapdragon 865 Plus. It features Wi-Fi 6E, Bluetooth 5.2, 4K QAM, 160MHz channels, and 4-stream DBS. It’s the first mobile connectivity system to support these features.


Camera Features with the Snapdragon 888’s Spectra 580 ISP

Qualcomm has been achieving a number of milestones with its Spectra ISPs over the last few years, which have been dual-ISP since their beginning five years ago. The Spectra 280 ISP brought support for 10-bit color depth HDR video capture, then the Spectra 380 ISP in the Snapdragon 855 was the world’s first CV-ISP, and in 2019, the Spectra 480 ISP boasted of an impressive 2 gigapixels/second processing speed. Now, the Spectra 580 ISP brings quite a few major leaps forward with a new triple ISP architecture, 35% speed increase, support for staggered HDR sensors, and more. This is potentially the most exciting new IP of the SoC, even more so than the CPU.

Spectra 580 ISP on the Qualcomm Snapdragon 888

Source: Qualcomm

Recommended Reading: How Qualcomm is Improving the Camera Experiences on Android Phones with its Spectra ISPs

The Spectra 580 is the first Spectra with a triple ISP, which Qualcomm says will take professional image quality to the “next level”. It delivers triple camera concurrency and triple parallel processing. Qualcomm explains that most flagship phones these days come with at least three rear cameras with three different lenses: ultra-wide, wide, and telephoto. Triple concurrency enables users to record video from three different cameras at the same time in 4K HDR quality. It’s also applicable for photos, where the triple ISP can capture three photos at the same time at 28MP each.

Triple concurrency will provide smoother transitioning when zooming between cameras. As of now, when users start shooting with their wide-angle (standard) camera on a dual ISP, Qualcomm had to guess if they were going to zoom in to their telephoto or zoom out to their ultra-wide. The company no longer needs to do that with triple concurrency, as it can now run all three cameras in the background and instantly switch to the camera users choose.

The Spectra 580 is 35% faster than the Spectra 480, which means it can now capture 2.7 Gigapixels/second. Qualcomm uses that speed for faster burst photography. In one second, the ISP can now capture 120 photos at 12MP each.

The Spectra 580’s architecture is designed for new staggered HDR image sensors. Qualcomm says they will debut in smartphones soon, and they have the potential to “dramatically enhance HDR video quality”. It explains that staggered HDR image sensors output separate long, medium, and short exposures. Current image sensors capture one image in the same time that staggered HDR can capture three images, all with detail in different bright or dark parts of the scene. Then the Spectra 580’s triple concurrency can merge all of these images together to bring the user one final image with “incredible” dynamic range. This technique has been available for photo capture with previous SoCs, but for the first time with the Snapdragon 888, users will be capture 4K HDR video with computational HDR.

Improvements are there for photo capture as well. The Spectra 580 can now capture photos in 10-bit color depth in the HEIF format. Users will be able to capture photos in 1.08 billion shades of color, up from the 16.7 million colors that 8-bit color depth has. Qualcomm is four years late in this aspect as Apple has been able to take 10-bit HEIF photos since the iPhone 7 back in 2016. However, it’s good to see that this feature will finally arrive on flagship Android phones in 2021. Qualcomm notes that the Snapdragon 865 added video capture in the Dolby Vision format, but as of now, no Android phone supports Dolby Vision capture or playback, with the features being restricted to the Apple iPhone 12 series. A few Android phones can capture 4K HDR video in HDR10 or HDR10+ formats, though.

Snapdragon 888 devices will be able to capture 4K at 120fps just like the Snapdragon 865. Now, they will also be able to play such videos at 120fps for smooth video playback.

Qualcomm notes the basics of a professional-quality photo start with 3A: autofocus, auto-exposure, and auto white balance. For sharpness, dynamic range, and color accuracy, these aspects must be correct. The company notes that it puts “massive amounts of time and resources” into refining its 3A. The Spectra 580 features its 10th generation 3A algorithms. It’s also the first time that 3A will be powered by AI.

The company states that its new Saliency Auto Focus and Auto Exposure Engines are “incredible”, as they were built using virtual reality headsets equipped with eye-tracking. It trained the Saliency Auto Focus and Auto Exposure neural nets by showing people images in VR and tracking their eyes to see which part of the image they focused on. The new 3A promises to make image accuracy better.

The Spectra 580 ISP also brings a new low light architecture. Users will now be able to take photos in 0.1 lux, which is near darkness. This could mean less reliance on multi-frame image stacking in the form of camera night modes, and a renewed emphasis on zero shutter lag.

The Snapdragon 888’s camera experience also benefits from its 6th generation AI Engine (more on this below). Arcsoft, a third-party vendor, has shown how the AI Engine can improve the camera experience. Qualcomm notes that in the past, point-and-shoot wasn’t point-and-shoot in the literal sense, as users had to select what they wanted to focus on, then zoom in and out to frame their photo and video. The Triple ISP is now always capturing video, and Arcsoft will use the ISP and the AI Engine to track and zoom in and out automatically, which will deliver on the true promise of the point-and-shoot paradigm.

Ultimately, Qualcomm claims that Snapdragon 888 smartphones will become professional-quality cameras thanks to the Spectra 580 ISP. If these claims do play out, we could be looking at significantly improved Android smartphone cameras in 2021.

Camera highlights of the Qualcomm Snapdragon 888

Source: Qualcomm


AI & Machine Learning: 6th Generation AI Engine and Hexagon 780 DSP

Unlike other vendors, Qualcomm doesn’t use the term “Neural Processing Unit”, “AI Processing Unit”, or “Neural Engine”. Instead, since the Snapdragon 855, it has used the “AI Engine” term, which encompasses the CPU, GPU, and DSP. The company has been steadily improving its AI and ML capabilities with the introduction of a Tensor Accelerator in the Snapdragon 855 and real-time translation with all AI processed on the device in the Snapdragon 865’s 5th generation AI Engine. Now, with the Snapdragon 888, the 6th generation AI Engine delivers 26 TOPS (trillion operations per second) of performance. In comparison, the previous generation Snapdragon 865 delivered 15 TOPS, while the Apple A14 delivers 11 TOPS, so it’s a great achievement.

The Snapdragon 888’s 6th generation AI Engine is more powerful and sophisticated. At the core of it is the Hexagon DSP. This year, Qualcomm is launching the Hexagon 780 DSP, which is completely redesigned and which features the company’s “biggest leap” in architecture and performance in years. The company calls it the fused AI accelerator architecture. In previous generations, it used scalar, vector, and tensor accelerators. For the Snapdragon 888, the company has removed the physical distances between the accelerators and has fused them together, so it’s all now on one big AI accelerator. It has also added a dedicated large shared memory across the three different accelerators for sharing and moving data efficiently. The shared memory is 16x larger than its predecessor, which means the hard-off time between the accelerators is in the nanosecond range – it’s up to 1000x faster in certain use cases.

Fused AI Accelerator on the Hexagon 780 DSP

Source: Qualcomm

Qualcomm has also made improvements on the accelerators themselves. The scalar accelerator is 50% more powerful, while the tensor accelerator is 2x faster than that in the Snapdragon 865. The Hexagon Vector eXtensions (HVX) now supports additional data types.

Other parts of the AI Engine have also received upgrades, as the Adreno 660 GPU now offers a 43% AI performance boost and includes new instruction sets like 4-input mixed-precision dot product and wave matrix multiply for faster floating-point calculation.

Qualcomm notes that the 26 TOPS is the highest TOPS performance on mobile. The power consumption is also ultra-low, as the Hexagon 780 DSP is now 3x faster in terms of performance-per-watt than the previous generation.

This year, the company is demonstrating a brand new AI use case that fully utilizes the 6th gen Qualcomm AI Engine: Tetris.AI’s super movie app. For example, users will be able to erase a character and put yourself inside a movie scene or a video that they recorded and interact with the other characters inside. They can see this in real-time in preview mode even before they start acting and recording. The Qualcomm AI Engine is running and accelerating Tetris.AI’s video instance segmentation and fusion algorithms at 30 fps, up to 4K resolution.

2nd generation Qualcomm Sensing Hub

The Snapdragon 888 introduces the company’s 2nd generation Qualcomm Sensing Hub. Qualcomm has added a dedicated always-on, low-power AI processor, and it claims to have seen a 5x AI performance improvement because of it. The extra AI processing power on the Sensing Hub allows it to offload up to 80% of the workload that usually goes to the Hexagon DSP so that power can be saved. All of the processing on the Sensing Hub is at less than 1mA of power consumption. The company is also working with Google and its TensorFlow Micro Framework to give developers easier access to the Sensing Hub, so that it can be optimized and accelerated on both the Hexagon DSP and the AI processor in the Sensing Hub.

The Sensing Hub also has a new feature where it has the ability to collect and decipher data from all different cores and create contextual awareness use cases. For the first time, Qualcomm is able to collect connectivity data like 5G, Wi-Fi, Bluetooth, and location streams. New always-on and contextually aware use cases will be enabled because of the Sensing Hub. Qualcomm gives an example of its work with Audio Analytic, which will allow the user’s phone to recognize the acoustics around them, which enables capabilities such as matching the ring volume to their environment.

AI software

Qualcomm has completely ramped up its AI software, where it has been operating from a position of strength. It was the first to commercialize on-device AI SDK in the form of the Qualcomm Neural Processing SDK, which now powers AI experiences in over 500 million Android phones globally. This year, improvements in the SDK include support for additional models and expanded support for Windows 10 AI use cases on laptops powered by the Snapdragon 888.

The company notes that it introduced Hexagon NN Direct on the Snapdragon 865 to give developers direct access to Hexagon from their applications. The 6th generation AI Engine features a significant upgrade here, as it brings direct APIs across the whole mobile platform. Qualcomm is introducing AI Engine Direct with its new AI Engine, where it extends and enhances the capabilities of its AI software solutions to provide developers with access directly to the hardware for not only the Hexagon DSP but also for the GPU and the CPU.

The AI Engine Direct has been built up from the ground to bring a unified AI API across the whole Snapdragon platform. It’s backward compatible with the 5th generation AI Engine. Qualcomm is also focused on modularity and extensibility as it expands on its user-defined operator concept to bring new capabilities for developers to create AI solutions.

The Snapdragon 888 sees the beginning of Qualcomm’s collaboration with Hugging Face, which is claimed to be a leader in “innovative” national language processing NLP solutions. Qualcomm is using the AI Engine to enable and accelerate the robust NLP library, Hugging Face transformers, for precision and responsiveness, with examples of use cases being autocomplete suggestions in the email app, improvements in AI voice assistants, and faster and more accurate language translation apps.

Qualcomm explains that in 2019, as a part of its 5th gen Qualcomm AI Engine, it introduced the concept of user-defined operators. This enabled developers to write custom operators in OpenCL or use the Qualcomm Hexagon SDK and then plug them into the Qualcomm Neural Processing SDK. However, even for developers that are already experienced with Hexagon, developers often needed to write complex and long routines in low-level languages to create operators. To rectify this, Qualcomm has extended TVM, an open-source compiler for AI accelerators with support for Hexagon. Custom operators can now be written in a few short lines of Python, then compiled for Hexagon, and plugged directly into the Qualcomm AI Engine direct framework.

Finally, the company has added additional support to the AI Model Efficiency Toolkit (AIMET) for better quantization of networks, with little or no loss in accuracy, using post-training techniques such as Adaround, and quantization aware training with range learning. It has also included support for RNN and LSTM networks. With the addition of support for mixed-precision networks, developers will be able to maximize power/performance tradeoffs while maintaining accuracy. Just as it did with TVM, it has open sourced the AIMET on Github, and it invites collaboration with its researchers.

Qualcomm is continuing to work with Snapchat to enable AIMET on its popular lenses. Snapchat is using AIMET to quantize an array of its AI lens models to improve accuracy and performance for face detection.


Snapdragon Elite Gaming Features on the Snapdragon 888

Qualcomm notes that there are an estimated 2.6 billion mobile gamers around the world, and gamers are estimated to play 25% more games than a year ago. It notes its own mobile gaming achievements that include bringing top AAA games to mobile, delivering smooth gaming with high frame rates up to 144fps, true 10-bit HDR in mobile gaming, and being the first to bring desktop-level features like per-game updatable GPU drivers to mobile platforms. The company first introduced Snapdragon Elite Gaming software features with the Snapdragon 855.

The company notes that the Adreno 660 GPU is at the heart of its gaming experience. It has focused on sustained performance over long periods of time while achieving its biggest leap in graphics rendering speeds (35%). The two new features announced are Qualcomm Game Quick Touch and Variable Rate Shading (VRS).

Qualcomm Game Quick Touch

Acknowledging the importance of touch response times, the Snapdragon 888 introduces Qualcomm Game Quick Touch. This is a new feature that greatly reduces touch latency. Qualcomm notes that touch latency depends on many factors like the timing of a game’s display v-sync and its frame submission. A game may miss the v-sync deadline due to heavy game workloads, which results in a delayed frame, which, in turn, then impacts the latency of the touch event. Game Quick Touch is optimized at the millisecond level to avoid these delays, enabling games to experience faster response times.

Qualcomm says that its lab testing has shown that Game Quick Touch can decrease the touch latency by up to 20%. A game running at even 120fps will see an improvement in touch response times, and the technology will be automatically enabled to work with any game, which will provide a pro-gamer-level experience and improvements to all games.

Visual demonstration of touch latency reduction via Qualcomm Game Quick Touch. Source: Qualcomm

Variable Rate Shading (VRS)

Qualcomm has announced that Snapdragon Elite Gaming is bringing Variable Rate Shading (VRS) to mobile devices for the first time. VRS has only been available on PCs and next-generation consoles (PS5, Xbox Series X, and Series S) until now. VRS is powered by the Adreno 660 GPU, and it helps to reduce GPU workloads while providing “significant enhancements” to games. The next generation of mobile games will run faster and at higher resolutions while still maintaining high visual fidelity.

What does VRS mean? Qualcomm explains that when rendering a frame, the GPU executes a shader program on each pixel to compute its color. In AAA games today, there are 3.6 million pixels being shaded on the display as an example. VRS allows developers to specify that the shader program only runs once in groups of two or four pixels, and it then reuses those color results for the surrounding pixels. It means that a developer can shade the entire frame using only 1.4 million pixels, which results in 40% more efficiency, while greatly diminishing the workload for the GPU, which, in turn, provides greater power savings.

The GPU workload is reduced via VRS, but that doesn’t mean the graphics fidelity will be lowered – it will stay constant. Games will see a 30% increase in gameplay performance from previous Snapdragon SoCs (Qualcomm didn’t specifically state which SoC) while running soother and longer with lower power. The end game? Developers will have more headroom to use the hardware, and they can create larger experiences for next-generation mobile games. Qualcomm notes that ultimately, its mission for Snapdragon Elite Gaming is to transform mobile devices into powerful gaming machines.


Security

In terms of security features, the Snapdragon 888 features a new Type-1 Hypervisor, which provides a new way to secure and isolate data between apps and multiple operating systems on the same device. It instantly switches between isolated operating systems and has an isolated operating system for each app as well with no performance degradation.

The Snapdragon 888’s security measures include the Qualcomm Secure Processing Unit, Qualcomm Trusted Execution Environment (TEE), and support for Qualcomm Wireless Edge Services, which is a cloud service that the chip can interact with for apps and services to measure the security of the devices and its wireless connections in real-time. The Snapdragon 888 provides sandboxing across all VMs, with the isolation being provided below OS level at the EL2 level.

The Snapdragon 888 is the world’s first CAI compliant smartphone camera. In collaboration with Truepic, the chip can capture cryptographically-sealed photos that are compliant with the open Content Authenticity Initiative standard.

Verifiable metadata of images captured using Truepic’s technology. Source: Truepic


Comparison: Snapdragon 888 vs Snapdragon 865 vs Snapdragon 855

Qualcomm Snapdragon 855 Qualcomm Snapdragon 865 Qualcomm Snapdragon 888
Announcement Date December 5, 2018 December 4, 2019 December 2, 2020
CPU
  • 1x Kryo 485 (ARM Cortex A76-based) Prime core @ 2.84GHz, 1x 512KB L2 cache
  • 3x Kryo 485 (ARM Cortex A76-based) Performance cores @ 2.42GHz, 3x 256KB L2 cache
  • 4x Kryo 385 (ARM Cortex A55-based) Efficiency cores @ 1.8GHz, 4x 128KB L2 cache
  • 2MB L3 cache
  • 1x Kryo 585 (ARM Cortex A77-based) Prime core @ 2.84GHz, 1x 512KB L2 cache
  • 3x Kryo 585 (ARM Cortex A77-based) Performance cores @ 2.4GHz, 3x 256KB L2 cache
  • 4x Kryo 385 (ARM Cortex A55-based) Efficiency cores @ 1.8GHz, 4x 128KB L2 cache
  • 4MB L3 cache
  • 25% faster performance YoY
  • 1x Kryo 680 (ARM Cortex X1-based) Prime core @ 2.84GHz, 1x 1MB L2 cache
  • 3x Kryo 680 (ARM Cortex A78-based) Performance cores @ 2.4GHz, 3x 512KB L2 cache
  • 4x Kryo 680 (ARM Cortex A55-based) Efficiency cores @ 1.8GHz, 4x 128KB L2 cache
  • 4MB L3 cache
  • 25% faster performance YoY
GPU
  • Adreno 640 @ 600MHz
  • Vulkan 1.1
  • Snapdragon Elite Gaming
  • Adreno 650
  • Vulkan 1.1
  • Snapdragon Elite Gaming with new Desktop Forward Rendering, Game Color Plus, updatable GPU drivers
  • 20% faster graphics rendering YoY
  • 35% more power efficient YoY
  • Adreno 660
  • Vulkan 1.1
  • Snapdragon Elite Gaming with new Qualcomm Game Quick Touch and Variable Rate Shading features
  • 35% faster graphics rendering YoY
  • 20% more power efficient YoY
  • 43% AI performance boost YoY
Display
  • Maximum On-Device Display Support: UHD
  • Maximum External Display Support: UHD
  • HDR support
  • DisplayPort over USB Type-C support
  • Maximum On-Device Display Support: UHD @ 60Hz, QHD+ @ 144Hz
  • Maximum External Display Support: UHD @ 60Hz
  • HDR support
  • DisplayPort over USB Type-C support
  • Maximum On-Device Display Support: UHD @ 60Hz, QHD+ @ 144Hz
  • Maximum External Display Support: UHD @ 60Hz
  • HDR support
  • DisplayPort over USB Type-C support
  • Demura and subpixel rendering for OLED uniformity
AI
  • Hexagon 690 with Hexagon Vector eXtensions and Hexagon Tensor Accelerator
  • 4th generation AI Engine
  • 7 TOPS
  • Hexagon 698 with Hexagon Vector eXtensions and new Hexagon Tensor Accelerator
  • 5th generation AI Engine
  • Qualcomm Sensing Hub
  • 15 TOPS
  • Hexagon 780 with Fused AI Accelerator architecture
  • 6th generation AI Engine
  • Qualcomm Sensing Hub (2nd generation)
    • New dedicated AI processor
    • 80% task reduction offload from Hexagon DSP
    • 5X more processing power YoY
  • 16X larger shared memory
  • 50% faster scalar accelerator, 2x faster tensor accelerator YoY
  • 26 TOPS
Memory
  • 4 x 16-bit LPDDR4 @ 2133MHz, 16GB
  • 3MB system level cache
  • 4 x 16-bit LPDDR4 @ 2133MHz, 16GB
  • LPDDR5 @ 2750MHz
  • 3MB system level cache
  • 4 x 16-bit LPDDR4 @ 2133MHz, 16GB
  • LPDDR5 @ 3200MHz
  • 3MB system level cache
ISP
  • Dual 14-bit Spectra 380 ISP
  • Single camera: Up to 48MP with ZSL
  • Dual camera: Up to 22MP with ZSL
  • Video capture: 4K HDR @ 60 fps; Slow motion up to 720p@480 fps; HDR10, HDR10+, HLG
  • Dual 14-bit Spectra 480 ISP
  • Single camera: Up to 64MP with ZSL
  • Dual camera: Up to 25MP with ZSL
  • Video capture: 4K HDR @ 60 fps + 64MP burst images; 4K @ 120 fps; 8K @ 30 fps; Slow motion up to 720p@960 fps (unlimited); HDR10, HDR10+, HLG, Dolby Vision
  • Triple 14-bit Spectra 580 ISP
  • Single camera: Up to 84MP with ZSL
  • Dual camera: Up to 64+25MP with ZSL
  • Video capture: 4K HDR @ 60 fps + 64MP burst images; 4K @ 120 fps; 8K @ 30 fps; Slow motion up to 720p@960 fps (unlimited); HDR10, HDR10+, HLG, Dolby Vision
  • Designed for staggered HDR image sensors
  • Support for 10-bit color depth photo capture in HEIF
  • New low-light architecture (capture photos in 0.1 lux)
  • 2.7 Gigapixels per second throughput (35% speed increase YoY)
Modem
  • Snapdragon X24 4G LTE integrated modem
    • Downlink: 2.0Gbps
    • Uplink: 316Mbps
  • Snapdragon X50 5G external modem
    • Downlink: 5.0Gbps
    • Modes: NSA, TDD
    • mmWave: 800MHz bandwidth, 8 carriers, 2×2 MIMO
    • sub-6 GHz: 100MHz bandwidth, 4×4 MIMO
  • Snapdragon X55 4G LTE and 5G multimode external modem
    • Downlink: 7.5Gbps (5G), 2.5Gbps (4G LTE)
    • Uplink: 3Gbps, 316Mbps (4G LTE)
    • Modes: NSA, SA, TDD, FDD
    • mmWave: 800MHz bandwidth, 8 carriers, 2×2 MIMO
    • sub-6 GHz: 200MHz bandwidth, 4×4 MIMO
  • Snapdragon X60 4G LTE and 5G multimode integrated modem
    • Downlink: 7.5Gbps (5G)
    • Uplink: 3Gbps
    • Modes: NSA, SA, TDD, FDD
    • 5G CA across FDD and TDD
    • mmWave: 800MHz bandwidth, 8 carriers, 2×2 MIMO
    • sub-6 GHz: 200MHz bandwidth, 4×4 MIMO
Charging Qualcomm Quick Charge 4+ (27W)
  • Qualcomm Quick Charge 4+ (27W)
  • Qualcomm Quick Charge AI
Qualcomm Quick Charge 5 (100W+)
Connectivity
  • Location: Beidou, Galileo, GLONASS, GPS, QZSS, SBAS, Dual Frequency support
  • Qualcomm FastConnect 6200
    • Wi-Fi: Wi-Fi 6 ready; 2.4/5GHz Bands; 20/40/80 MHz Channels; DBS, TWT, WPA3, 8×8 MU-MIMO
    • Bluetooth: Version 5.0, aptX TWS and aptX Adaptive
  • Location: Beidou, Galileo, GLONASS, GPS, QZSS, SBAS, NavIC capable, Dual Frequency support
  • Qualcomm FastConnect 6800
    • Wi-Fi: Wi-Fi 6 certified; 2.4/5GHz Bands; 20/40/80 MHz Channels; DBS, TWT, WPA3, 8×8 MU-MIMO, OFDMA, 1024QAM
    • Bluetooth: Version 5.1, aptX TWS, aptX Adaptive, and aptX Voice
  • Location: Beidou, Galileo, GLONASS, GPS, QZSS, SBAS, NavIC capable, Dual Frequency support
  • Qualcomm FastConnect 6900
    • Wi-Fi: Wi-Fi 6 & 6E certified; 2.4/5GHz/6GHz Bands; 20/40/80/160 MHz Channels; 4-stream DBS, TWT, WPA3, 8×8 MU-MIMO, OFDMA, 4KQAM
    • Bluetooth: Version 5.2, LE Audio Features (one-to-many broadcast), Qualcomm TrueWireless Mirroring, aptX TWS, aptX Adaptive, and aptX Voice
Manufacturing Process 7nm (TSMC’s N7) 7nm (TSMC’s N7P) 5nm

Qualcomm Snapdragon 888 Full Specifications & Features

Complete Feature List. Click to expand.

Artificial Intelligence

  • Adreno 660 GPU
  • Kryo 680 CPU
  • Hexagon 780 Processor
    • Fused AI Accelerator
      • Hexagon Tensor Accelerator
      • Hexagon Vector eXtensions
      • Hexagon Scalar Accelerator
  • Qualcomm Sensing Hub (2nd Generation)

5G Modem-RF System

  • Snapdragon X60 5G Modem-RF System
    • 5G mmWave and sub-6 GHz, standalone (SA) and non-standalone (NSA) modes, FDD, TDD
    • Dynamic Spectrum Sharing
    • mmWave: 800 MHz bandwidth, 8 carriers, 2×2 MIMO
    • Sub-6 GHz: 200 MHz bandwidth, 4×4 MIMO
    • Qualcomm 5G PowerSave
    • Qualcomm Smart TransmitTM technology
    • Qualcomm Wideband Envelope Tracking
    • Qualcomm Signal Boost adaptive antenna tuning
    • Global 5G multi-SIM
  • Downlink: Up to 7.5 Gbps
  • Uplink: Up to 3 Gbps
  • Multimode support: 5G NR, LTE including CBRS, WCDMA, HSPA, TD-SCDMA, CDMA 1x, EV-DO, GSM/EDGE

Wi-Fi & Bluetooth

  • FastConnect 6900 System
    • Wi-Fi Standards: Wi-Fi 6E, Wi-Fi 6 (802.11ax), Wi-Fi 5 (802.11ac), 802.11a/b/g/n
    • Wi-Fi Spectral Bands: 2.4 GHz, 5 GHz, 6 GHz
    • Peak speed: 3.6 Gbps
    • Channel Utilization: 20/40/80/160 MHz
    • 8-stream sounding (for 8×8 MU-MIMO)
    • MIMO Configuration: 2×2 (2-stream)
    • MU-MIMO (Uplink & Downlink)
    • 4K QAM
    • OFDMA (Uplink & Downlink)
    • Dual-band simultaneous (2×2 + 2×2)
    • Wi-Fi Security: WPA3-Enterprise, WPA3-Enhanced Open, WPA3 Easy Connect, WPA3-Personal
  • Integrated Bluetooth
    • Bluetooth Version: Bluetooth 5.2
    • Bluetooth features: LE Audio Features (one-to-many broadcast), Dual Bluetooth antennas
    • Bluetooth audio: Qualcomm aptX Voice audio for crystal-clear voice calls, aptX Adaptive audio for robust, low latency, high quality audio, Qualcomm TrueWirelessTM Mirroring

Camera

  • Qualcomm Spectra 580 Image Signal Processor
    • Triple 14-bit ISPs
    • Up to 2.7 Gigapixels per Second computer vision ISP (CV-ISP)
    • Up to 200 Megapixel Photo Capture
    • Up to 28 MP triple camera @ 30 FPS with Zero Shutter Lag
    • Up to 64+25 MP dual camera @ 30 FPS with Zero Shutter Lag
    • Up to 84 MP single camera @ 30 FPS with Zero Shutter Lag
  • Rec. 2020 color gamut photo and video capture
  • Up to 10-bit color depth photo and video capture
  • 10-bit HDR HEIF photo capture
  • 4K Video Capture + 64 MP Photo
  • 8K Video Capture @ 30 FPS
  • Slow-mo video capture at 720p @ 960 FPS
  • HEIF: HEIC photo capture, HEVC video capture
  • Video Capture Formats: HDR10+, HDR10, HLG, Dolby Vision
  • 4K Video Capture @ 120 FPS
  • 4K HDR Video Capture with Portrait Mode (Bokeh)
  • Multi-frame Noise Reduction (MFNR)
  • Real-time object classification, segmentation, and replacement
  • Locally compensated Multi-Frame Noise Reduction
  • Multi-Frame and Staggered HDR sensor support
  • Low light photography architecture
  • Video super resolution
  • AI-based auto-focus and auto-exposure
  • Advanced HW-based face detection with deep learning filter

Audio

  • Hexagon Voice Assistant Accelerator for hardware accelerated voice signal processing
  • Qualcomm AqsticTM audio codec (Up to WCD9385)
  • Total Harmonic Distortion + Noise (THD+N), Playback: -108dB
  • Native DSD support, PCM up to 384 kHz/32-bit
  • Customizable “Golden Ears” filter
  • New Qualcomm Aqstic smart speaker amplifier (up to WSA8835)

Display

  • On-Device Display Support:
    • 4K @ 60 Hz
    • QHD+ @ 144 Hz
  • Maximum External Display Support: up to 4K @ 60 Hz
    • 10-bit color depth, Rec. 2020 color gamut
    • HDR10 and HDR10+
  • Demura and subpixel rendering for OLED Uniformity

CPU

  • Kryo 680 CPU
    • Up to 2.84GHz, with Arm Cortex-X1 technology
    • 64-bit Architecture

Visual Subsystem

  • Adreno 660 GPU
    • Vulkan 1.1 API support
    • HDR gaming (10-bit color depth, Rec. 2020 color gamut)
    • Physically Based Rendering
    • API Support: OpenGL ES 3.2, OpenCLTM 2.0 FP, Vulkan 1.1
    • Hardware-accelerated H.265 and VP9 decoder
    • HDR Playback Codec support for HDR10+, HDR10, HLG and Dolby Vision

Security

  • Platform Security Foundations, Trusted Execution Environment & Services, Secure Processing Unit (SPU)
  • Qualcomm wireless edge services (WES) and premium security features
  • Qualcomm 3D Sonic Sensor and Qualcomm 3D Sonic Max (fingerprint sensor)
  • Qualcomm Type-1 Hypervisor

Charging

  • Qualcomm Quick Charge 5 Technology

Location

  • GPS, Glonass, BeiDou, Galileo, QZSS, NavIC capable, and SBAS
  • Dual Frequency Support
  • Low Power Geofencing and Tracking, Sensor assisted Navigation
  • Near Field Communications (NFC): Supported

Memory

  • Support for LP-DDR5 memory up to 3200 MHz
  • Support for LP-DDR4x memory up to 2133 MHz
  • Memory Density: up to 16 GB

General Specifications

  • Full Suite of Snapdragon Elite Gaming features
  • 5 nm Process Technology
  • USB Version 3.1; USB Type-C Support
  • Part Number: SM8350

Initial Conclusions

Qualcomm says devices featuring the Snapdragon 888 are expected to be commercially available in the first quarter of 2021. We can expect the very first flagship phone to feature it to be the Xiaomi Mi 11 next month, while the Snapdragon variants of the Galaxy S21 series won’t be too far behind. Phones like the Realme Race, OPPO Find X3 series, and the OnePlus 9 series are expected to launch sometime in February and March 2021 respectively.

List of OEMs building phones with the Snapdragon 888

The Snapdragon 888 is a respectable step forward for Qualcomm. Yes, it’s overshadowed and outgunned in both CPU performance as well as GPU performance by the new behemoth in the chip industry – Apple. However, as Qualcomm keeps reminding us, there’s more to a great chip than a CPU and a GPU. Qualcomm’s resources this generation have been spent on the AI Engine and the Spectra ISP, and the improvements made in both fields seem quite promising. If we constrain ourselves to the Android SoC market, it’s hard to see a 2021 where the Snapdragon 888 isn’t the best Android flagship SoC. The Exynos 2100 is expected to make a big leap in CPU performance, but the two chips will be roughly tied even in the best case, depending on clock speeds. Qualcomm still enjoys a comfortable lead in GPU performance over both Samsung and MediaTek, as Samsung won’t switch to AMD’s RDNA GPU architecture until 2022. Also, Qualcomm still seems to be in the lead when it comes to the AI software stack.

Overall, with support for taking photos in 0.1 lux, 144Hz displays, Snapdragon Elite Gaming, and meaningful new software features, it’s hard to make the argument that Qualcomm is chasing only numbers. Instead, the company continues to show an admirable focus on real-world performance.

The post The Qualcomm Snapdragon 888 will power flagship 5G phones in 2021 – Here’s what you need to know appeared first on xda-developers.



from xda-developers https://ift.tt/37rvdFd
via IFTTT

Jetpack Compose for Desktop, a cross-platform UI development framework, adds a Swing interoperability layer and Apple Silicon support

If you do any sort of development work, you’ve probably heard of JetBrains. They’re the company behind the huge suite of IDEs that includes IntelliJ (the base for Android Studio), CLion, PhpStorm, and many others. They’re also the ones who made Kotlin, the hugely-popular cross-platform programming language.

And JetBrains is back at it again. A little over a year ago, Google introduced a new layout engine for Android called Jetpack Compose. It’s gone through quite a few major changes since then, but it’s turned into a competent (albeit still somewhat unstable) alternative to Android’s classic XML layouts.

What does Jetpack Compose have to do with JetBrains? Well, for one, it’s written in Kotlin. But also, JetBrains has been silently working on porting Compose to the desktop space. It’s been public as an early developer preview, and now JetBrains is ready to officially announce that it exists.

What is Jetpack Compose?

I talked about it a bit in the introduction, but I think it deserves some more explanation. If you’ve developed on Android before, you’re probably used to how layouts work. First, you design your layout in an XML file, and then, you interact with that layout from Java or Kotlin. While it’s functional, it’s a little outdated, and having your layouts split across languages can be hard to manage.

As an alternative to this, Google started developing Jetpack Compose. Compose is a layout engine for Android built on top of Kotlin. All of your layout and logic code is in one place, which makes interactivity a lot easier. It’s also declarative, instead of the imperative style of XML layouts.

Moving to a declarative layout engine from an imperative one can take some getting used to, but Compose is definitely an improvement over XML, even though it’s still in its early stages.

Jetpack Compose for Desktop

So, Compose is a nice layout alternative for native Android apps. But it’s not (normally) cross-platform. That’s where JetBrains comes in. The company has ported Jetpack Compose to the desktop space, with support for Windows, Linux, and macOS (both Intel and ARM).

While there are other layout engines that exist for desktop, such as Electron, JavaFX, and UWP, it’s a bit of a mess right now. Some aren’t cross-platform, so they only work on one specific operating system. Others are technically cross-platform, but require a lot of work to distribute. And still, others are just a pain to work with, like JavaFX.

Jetpack Compose for Desktop, on the other hand, is relatively easy to use, supports display scaling, has built-in styling, and is (almost) fully cross-platform. While you can’t yet compile for all distributions from one operating system, the code itself is completely portable. Just like with Android, a packaged app is based on Java, although users don’t need a JDK installed to use it.

Cross-Compatibility

Since Compose for Desktop is still new, it is missing some features. To help developers work around these limitations, Compose for Desktop is interoperable with two of the more popular Java layout engines, JavaFX and Swing. If you have a desktop app developed in one (or both) of these engines, you can start transitioning to Compose while maintaining your current codebase.

Android + Desktop

What if you want to bring your Compose Android app to desktop? Well, you can also (sort of) do that. The UI elements of your app can easily be shared between Android and desktop. You won’t be able to share everything, but it certainly makes development easier.

Apple Silicon Support

I mentioned this above, but I’m going to mention it again. The new Apple Silicon Macs are pretty amazing, but the processors are based on the ARM CPU architecture, which means a lot of macOS apps are currently running through the Rosetta 2 x86 translator. While Apple did a pretty good job with that translation, native ARM programs will still run better.

Well, Jetpack Compose for Desktop supports Apple Silicon natively. That means you won’t have to worry about performance drops from translation or about future support when Apple discontinues it.

Functionality

So it’s all well-and-good that JetBrains made a new layout engine for desktop, but who’s to say it’s any better than what we have now?

Me.

I used the latest available build as of writing this (0.2.0-build132) to make a simple permissions granter app for SystemUI Tuner. Since I’m not that familiar with Compose and declarative design, it was a little more difficult for me than an XML-layout app, but I did it. And it’s a lot easier to use than JavaFX.

While theme support (i.e. dark mode) is still a little iffy, it’s pretty easy to build an interactive layout, and aside from a few flickers when resizing the window, everything works fine.

If you want to check the app out, the source code is available on GitHub. Just import it with IntelliJ or Android Studio and it should be ready to run.


Personally, I’m pretty excited about Jetpack Compose for Desktop. I didn’t even know it existed before, but it’s already my desktop layout engine of choice. You can learn more on the Jetpack Compose for Desktop website or on GitHub.

What do you think about it? Is it worth checking out and maybe even using instead of other available frameworks? Let me know!

Featured image credits: JetBrains on GitHub

The post Jetpack Compose for Desktop, a cross-platform UI development framework, adds a Swing interoperability layer and Apple Silicon support appeared first on xda-developers.



from xda-developers https://ift.tt/3lvBXXq
via IFTTT

Samsung Galaxy Buds Pro leak points to new in-ear design and improved ANC

Samsung has been working on the next generation of true-wireless earbuds from the Galaxy Buds range. The Samsung Galaxy Buds Pro aren’t due for release until early in 2021, but a post from legendary leaker Evan Blass has given us a good idea what to expect. As we ventured in our review of this year’s Galaxy Buds Live, one of our takeaways was that the ANC (active noise-cancelling) was somewhat underwhelming, with most of the work actually being done with passive noise-isolation. The Buds Pro will be the first in-ear design from Samsung with ANC onboard. We’re hoping, therefore that means they’ll have taken on board that kind of feedback and give us something that actually blocks out the outside world properly.

Samsung Galaxy Buds Pro Leak Box

Certainly, the design has changed, moving back to something more akin to the OG Buds or Buds+, and away from “beans”. The latest leak suggests that the Galaxy Buds Pro will launch alongside the Samsung Galaxy S21 range, early in 2021, though take that with a pinch of salt. Blass also quotes SamMobile, which notes that the FCC filing for the product points to a 500mAh battery. You’ll note that these renders are rather… erm… violet. We’re hoping there’ll be other color variants that won’t make you look like you have a sprig of heather in your ear.

Blass suggests that the Samsung Galaxy Buds Pro (an assumed moniker, though don’t rule out ‘Galaxy Buds Beyond’ which has also been rumored as a name) will be an additional, rather than replacement line, and earlier models may be reduced in price to accommodate their new sibling. He also warns that whilst it is assumed that they’ll launch with the S21 range, that sort of tie-in has been rumored before and come to nothing, so don’t be disappointed if that delivery date slips a bit.

The post Samsung Galaxy Buds Pro leak points to new in-ear design and improved ANC appeared first on xda-developers.



from xda-developers https://ift.tt/3qeOaDs
via IFTTT

Android Auto could soon be set to start automatically, even if your phone is locked

Android Auto is Google’s answer to the question “how do I connect my phone to my car”. The app essentially lets you access a limited set of optimized functions while you drive, with the primary objectives of balancing driver focus, connectivity, and entertainment. But to actually connect to Android Auto, you need to connect your smartphone to your car’s infotainment unit, either through a wire or wirelessly. Google is apparently making it slightly more convenient to connect to Android Auto, letting users get connected without requiring to unlock their smartphones.

An APK teardown can often predict features that may arrive in a future update of an application, but it is possible that any of the features we mention here may not make it in a future release. This is because these features are currently unimplemented in the live build and may be pulled at any time by the developers in a future build.

We spotted the following new strings in Android Auto 5.9:

<string name="settings_connection_allowed_while_locked_description">Even when your phone is locked</string>
<string name="settings_connection_allowed_while_locked_title">Start Android Auto automatically</string>

Currently, the Android Auto app automatically launches itself if you connect your phone to a car and unlock your phone (presuming you have the app already installed on your phone and that you have a head unit that supports Android Auto). We know some units have trouble launching Android Auto automatically upon connection with an unlocked phone. While the hints within the app do not mention this, this new setting could possibly be to remedy these instances of connection failure. If this feature completes implementation, the app should launch automatically when you connect your smartphone, without requiring you to unlock your phone, when you toggle the setting.

We don’t know if this feature is for wired connections or for wireless connections, or both. We will have to wait for Google to roll out the change to find out more.

Android Auto - Google Maps, Media & Messaging (Free, Google Play) →

The post Android Auto could soon be set to start automatically, even if your phone is locked appeared first on xda-developers.



from xda-developers https://ift.tt/3fWsW8O
via IFTTT