Everything you need to know about Snapdragon 8 Gen 2

0
Everything you need to know about Snapdragon 8 Gen 2
Everything you need to know about Snapdragon 8 Gen 2

The next-generation smartphone’s foundation.

This week, Qualcomm released its latest mobile platform – the Qualcomm Snapdragon 8 Gen 2 – at an event in Hawaii on November 15th. Snapdragon 8 Gen 2 isn’t simply an improvement over Snapdragon 8 Gen 1, but the company’s latest entry into its flagship Snapdragon series packs a number of improvements as well as new features that will be beneficial to high-end smartphones in 2023 and beyond.

Here is our essential summary of the main points we want to cover, but there is a lot more to discover if you want to explore a lot more deeply. There are a lot of firsts for Qualcomm with the Snapdragon 8 Gen 2 because it features a completely new CPU cluster arrangement, a GPU that is capable of ray tracing, high-end audio features and connectivity, as well as a deeper integration of imaging and machine learning.

Must Read: Snapdragon AR2 Gen 1 platform revealed

Qualcomm clarifies 32-bit support for Arm Cortex A510 CPU cores November 16, 2022.

Snapdragon 8 Gen 2 versus Snapdragon 8 Gen 1

CPU Config1x 3.19GHz (Cortex-X3)
2x 2.8GHz (Cortex-A715)
2x 2.8GHz (Cortex-A710)
3x 2.0GHz (Cortex-A510)
1x 3.0GHz (Cortex-X2)
3x 2.5GHz (Cortex-A710)
4x 1.8GHz (Cortex-A510)
1x 2.84GHz (Cortex-X1)
3x 2.4GHz (Cortex-A78)
4x 1.8GHz (Cortex-A55)
GPUAdreno 740
(ray tracing support)
Adreno 730Adreno 660
DSPHexagon
(fused scalar, tensor, and vector)
Mixed precision INT8/INT16
INT4 support
Hexagon
(fused scalar, tensor, and vector)
Mixed precision INT8/INT16
Hexagon 780
(fused scalar, tensor, and vector)
RAM supportLPDDR5XLPDDR5LPDDR5
Camera support• 200MP single shot
• 108MP single with zero shutter lag
• 64MP+36MP with zero shutter lag
• Triple 36MP with zero shutter lag
• Hybrid AF
• 10-bit HEIF image capture
• HDR video
• Multi-frame noise reduction
• Real-time object classification, segmentation, and replacement
• Video super resolution
• 200MP single shot
• 108MP single with zero shutter lag
• 64MP+36MP with zero shutter lag
• Triple 36MP with zero shutter lag
• Hybrid AF
• 10-bit HEIF image capture
• HDR video
• Multi-frame noise reduction
• Real-time object classification, segmentation, and replacement
• Video super resolution
• 200MP single shot
• 84MP single with zero shutter lag
• 64MP+25MP with zero shutter lag
• Triple 24MP with zero shutter lag
• Hybrid AF
• 10-bit HEIF image capture
• HDR video
• Multi-frame noise reduction
• Real-time object classification, segmentation, and replacement
Video capture8K @ 30fps (HDR)
4K UHD @ 120fps
720p @ 960fps
8K @ 30fps (HDR)
4K UHD @ 120fps
720p @ 960fps
8K @ 30fps
4K UHD @ 120fps
720p @ 960fps
Video playback8K up to 60fps
4K HDR up to 120fps
AV1, H.265 and VP9 video decoder
360 degree
8K
4K HDR up to 120fps
H.265 and VP9 video decoder
360 degree
8K
4K HDR up to 120fps
H.265 and VP9 video decoder
360 degree
ChargingQuick Charge 5Quick Charge 5Quick Charge 5
4G/5G ModemX70 LTE/5G (integrated)
10,000Mbps down
3,500Mbps up
X65 LTE/5G (integrated)
10,000Mbps down
(3,000Mbps up assumed)
X60 LTE/5G (integrated)
7,500Mbps down
3,000Mbps up
Other networkingBluetooth 5.3
Wi-Fi 7, Wi-Fi 6/6E (802.11ax), Wi-Fi 5 (802.11ac), 802.11a/b/g/n
Bluetooth 5.2
Wi-Fi 6E, Wi-Fi 6 (802.11ax), Wi-Fi 5 (802.11ac), 802.11a/b/g/n
Bluetooth 5.2
Wi-Fi 6E, Wi-Fi 6 (802.11ax), Wi-Fi 5 (802.11ac), 802.11a/b/g/n
ProcessTSMC 4nm (N4?)Samsung 4nm 4LPESamsung 5nm LPE

Snapdragon 8 Gen 2 CPU arrangement explained

Everything you need to know about Snapdragon 8 Gen 2

Among the most notable changes in the Snapdragon 8 Gen 2 is the move from the well-established 1+3+4 CPU cluster arrangement to a more interesting 1+4+3 arrangement, which is one of the features that is immediately noticeable. Furthermore, Qualcomm has also decided to use two different CPU cores in the middle/performance cluster, consisting of two Cortex-A715s based on the latest Arm architecture and two Cortex-A710s based on the last-gen architecture. The design choice of doing so is sure to boost multi-core benchmarking scores, but it is also clearly a very specific one in terms of how the chip is designed.

As Qualcomm explained, the reason for the decision was to continue supporting legacy applications for as long as possible. There are no longer any cores developed by Arm, such as the Cortex-A710, that support 32-bit applications (AArch32) – all subsequent and future cores, such as the Cortex-A710 do not support 32-bit applications (AAArch32), at least in theory. It is also worth pointing out that the Snapdragon 8 Gen 2 also makes use of Arm’s refreshed Cortex-A510 small cores, which, in addition to a 5% reduction in power consumption, are also designed for 32-bit operation starting in 2022.

Must Read: There has been confirmation from Qualcomm that Samsung Galaxy S23 will only use Snapdragon processors

The 8 Gen 2 CPU layout ensures legacy 32-bit support.

There is no doubt that Qualcomm has supported 32-bit applications in the revised A510s, resulting in a total of five cores capable of supporting legacy applications. It should be noted that for 32-bit applications, this combination of two A710 performance cores should provide a level of performance that is better than the four A510 cores offered by the Mediatek Dimensity 9200 when those two are combined.

It will be interesting to see how more demanding legacy applications perform on this chip because it can utilize all of the chip’s cores, so it will be interesting to see how they perform against the 64-bit apps that can use all the cores of the chip. However, 32-bit support could effectively be redundant for many Snapdragon users and this tradeoff might even be a poor one for battery life if you consider the fact that one little efficiency core has been sacrificed as a result. According to Qualcomm, they claim that in order to mitigate the issue, they have further optimized the performance cores.

The reason for this mandatory 64-bit support is that Google has mandated it since 2019. There is now a 64-bit version of any app that has been updated on the Play Store in the past few years. Even so, with the Snapdragon 8 Gen 2 including the A710 and revised A510 cores, it ensures that it will be able to run older applications and those that are outside of Google’s Android ecosystem in future releases. As an example, it is China and third-party app stores that are still lagging behind in mandating 64-bit compatibility in their apps.

Is there a benefit for low-power use cases with an extra middle core?

With the extra middle core and the Arm Cortex-X3, the CPU clusters claim 35% performance improvement. As far as efficiency is concerned, Qualcomm claims that it has been able to improve it by up to 40%. It’s worth mentioning that the majority of this increase can be attributed to the move to TSMC’s 4nm process (Qualcomm hasn’t confirmed whether it’s using TSMC’s N4 or its newer N4P process, so we’re presuming the former), but it’s still a pretty impressive number given that only one efficiency core has been lost. The Qualcomm Snapdragon 8 Plus Gen 1 was successfully manufactured by TSMC after Qualcomm switched from Samsung to TSMC.


Performance cluster
Middle ClusterEfficiency Cluster
CPU cores1x Arm Cortex-X32x Arm Cortex-A715
2x Arm Cortex-A710
3x Arm Cortex-A510
Clock Speed3.19GHz2.8GHz2.0GHz
L1 cache(unknown)(unknown)(unknown)
L2 cache1MB(unknown)(unknown)
L3 cache8MB (shared)8MB (shared)8MB (shared)
64/32-bit support64-bit only2x A715: 64-bit only
2x A710: 64 and 32-bit
64 and 32-bit

As you can see in the table above, Qualcomm has provided us with an overview of the CPU setup, at least for as far as we are aware. Currently, we do not have the full cache information, which can have implications for the performance of the middle and efficiency cores. It is still noteworthy that Qualcomm has provided an upgraded shared L3 cache that will play a key role in optimizing performance in heavy multi-threaded workloads with the additional middle core, which will be available at 8MB instead of 6MB.

Graphs for mobile ray tracing

Everything you need to know about Snapdragon 8 Gen 2

There is one thing that is arguably the most mouth-watering feature in this release – mobile ray tracing graphics hardware has gone mainstream. It is not the first time Qualcomm has announced hardware-accelerated features for ray tracing in mobile devices; Qualcomm is joining AMD’s Xclipse GPU in Samsung’s Exynos 2200 and Arm’s Immortalis-G715 in Mediatek’s Dimensity 9200, which supports hardware-accelerated ray tracing. There is no doubt that Qualcomm’s shipment volume will make this announcement the one that will enable mobile ray-tracing to become a viable possibility for developers.

Must Read: There has been confirmation from Qualcomm that Samsung Galaxy S23 will only use Snapdragon processors

A frustrating fact about Qualcomm is that its Adreno GPU technology is closely guarded family secret. However, we know that the Snapdragon 8 Gen 2 enables the acceleration of intersections between ray-boxes and ray-triangles. Moreover, with the BVH acceleration (or the retrieval and decompression acceleration structure nodes, as Qualcomm calls it), the GPU is significantly more capable than before of testing ray collisions with optimal performance, which is a significant improvement.

Game developers can add reflections, shadows, and global illumination using ray tracing.

According to these details, Qualcomm’s implementation of ray tracing appears to be more robust than other options for ray tracing on mobile phones. It is worth noting that Arm’s hardware for ray tracing does not support BVH. Despite this, Qualcomm has not divulged any information as to how powerful the Snapdragon 8 Gen 2’s accelerator actually is or how well its ray tracing hardware will scale as the chip ages. It is therefore prudent to be skeptical of any claims you might hear about the graphical fidelity or the resolution of the PC or console ray tracing capabilities that you might hear. Despite that, Qualcomm’s ray tracing hardware runs on less than five watts of power while performing ray tracing.

The 8 Gen 2 will be available on Oppo’s next flagship Find X smartphone, which will be based on Qualcomm’s partner. In comparison to the same effects which are rendered using software, Oppo’s open-source PhysRay engine can result in a fivefold increase in ray-tracing rendering efficiency and a 90% reduction in CPU workload. According to the company, the 8 Gen2 is able to run its ray tracing engine at 60 frames per second for a duration of 30 minutes, with 720p resolution.

Everything you need to know about Snapdragon 8 Gen 2

According to Qualcomm, GPUs are now able to function as ray tracing accelerators for ray-tracing-capable Vulkan Android games in ways that can’t be achieved through software, such as rendering soft shadows, reflections, ambient occlusion, and global illumination. Hence, gaming in the future should look a little bit better as a result of these advancements. In fact, Qualcomm believes ray tracing hardware acceleration for AAA games will be available in the first half of 2023, which is just a few years away.

The first mobile game with ray tracing support is expected in H1 2023.

Besides being able to support ray tracing, the latest, unnamed Adreno GPU (let’s assume Adreno 740 internally) promises a 25% increase in performance over the previous generation, as well as a 45% reduction in power consumption, depending on the use case. As a result of Qualcomm’s optimization of the drivers, the Vulkan 1.3 API is supported, and its performance has been improved by 30% in some Vulkan-powered scenarios when using Qualcomm’s drivers. During Unreal Engine 5’s development phase, Qualcomm was the first to claim support for Unreal Engine 5’s Metahumans framework, while Qualcomm’s Adreno Display engine boasts support for Adaptive HDR, HDR Vivid, HDR10+, Dolby Vision, and OLED Aging Compensation. It sounds as if Snapdragon gamers are in for a big year this year, with all this news.

Extra AI for imaging and more

Everything you need to know about Snapdragon 8 Gen 2

Although Qualcomm hasn’t offered large numbers for us to ogle at this time, it has made some significant improvements in this area in previous years. In order to better understand the capabilities of the Snapdragon 8 Gen 2, let’s explore what is new with Qualcomm’s latest Hexagon DSP, the engine that drives Snapdragon 8 Gen 2’s system-wide artificial intelligence.

Many seemingly small improvements add up to a considerable amount. First of all, there is now a dedicated power delivery system. This means that the Hexagon DSP will no longer have to compete with other components like the GPU for the same amount of power. Efficiency is enhanced by a unique power domain. Consequently, Qualcomm claims a 60% improvement in power efficiency over the previous generation.

Also Read: Redmi Note 11 Pro (2023) with Snapdragon 712 was spotted on Google Play Console. Google Play Console

It supports low-resolution INT4 ML and Qualcomm’s Hexagon processor doubles Tensor crunching capabilities.

A new optimization for language processing has been made to the Tensor accelerator inside the DSP in order to achieve twice the performance and twice the size. The company is also launching what it refers to as micro tile inference support, which is essentially dividing up imaging and other problems into smaller tiles in order to reduce memory usage at the cost of some accuracy. With INT4, developers can now implement machine learning problems requiring high bandwidth, although some accuracy may be lost when compressing a larger model. In order to support INT4, Qualcomm is providing partners with tools; therefore, it may be necessary to retool existing applications in order to make them compatible.

The Snapdragon 8 Gen 2 Hexagon DSP offers, depending on the machine learning model, a performance increase of 4.35x over its predecessor. The Qualcomm Natural Language Processing is being compared with mobileBERT in this situation. Although this may sound impressive, I believe the most significant change is the introduction of Hexagon Direct Link, which allows the ISP to be more closely connected to the AI Engine. “Cognitive ISP” is what the company calls this service.

Everything you need to know about Snapdragon 8 Gen 2

It has been enhanced by Qualcomm to double the physical link between the image signal processor (ISP), the Hexagon DSP, and the Adreno GPU, resulting in higher bandwidth and lower latency. Consequently, the Snapdragon 8 Gen 2 is able to process much more powerful machine-learning tasks on image data immediately after it is captured by the camera sensor. In imaging workloads, for example, raw data can be passed directly to the DSP/AI Engine, or Qualcomm can use the link to upscale low-resolution gaming scenarios so as to help balance GPU loads.

With Hexagon Direct Link, imaging and other data is passed directly to the AI Engine without being slowed down by sluggish DDR memory.

segmentation and processing. As a result, elements within a scene are identified, such as faces, plants, sky, etc., and layers are created in real-time, followed by custom processing.

Various machine learning-type features, including face detection and video bokeh segmentation, have been moved closer to ISPs by Qualcomm in previous years. Last generation of the product certainly claimed segmentation capabilities. It should be noted, however, that the slower link previously required imaging data to be pulled into main memory first, a high-latency and costly procedure that usually required segmentation to be applied after capture. It is anticipated that Qualcomm will be able to eliminate this bottleneck in the near future, enabling complex workloads, such as image processing, to be processed in real time by its AI Engine. Qualcomm’s product partners have yet to demonstrate an ability to take advantage of these capabilities.

More wireless connectivity options

We will begin by examining the more exciting, user-defining new connectivity features. Snapdragon Sound’s new Dynamic Spatial Audio capability is now included in the chip’s updated audio suite. The dynamic head tracking technology on compatible headphones enables you to move your head within the space and hear the content move around you rather than following your head statically. There are many existing multi-channel spatial audio formats and decoders that work with this technology. These include Dolby Atmos and Sony 360 Reality Audio.

Everything you need to know about Snapdragon 8 Gen 2

Continuing the discussion of audio, Qualcomm’s aptX Lossless codec is now supported in both Bluetooth Classic and LE Audio use cases, thus combining the benefits of low energy and lossless audio playback in future products. With a compatible headset, wireless latency can be reduced to just 48ms – 47% lower than its predecessor.

Music lovers benefit from Dynamic Spatial Audio and lossless Bluetooth streaming.

There was a time when you might have thought that 5G was over, but that was not the case. Snapdragon 8 Gen 2 is a new version of Qualcomm’s flagship chipset that shakes up a few things. The device features a Snapdragon X70 modem in the form of four carriers aggregation, which is capable of delivering speeds of 10Gbps down and 3.5Gbps up. Additionally, there are AI smarts on board the device.

Also Read: The vivo X Fold+ will feature fast charging at 80W, Snapdragon 8+ Gen 1 confirmed

With its artificial intelligence capabilities, Qualcomm claims that the modem is able to improve both the throughput and the robustness of the connections in the sub-6GHz and mmWave bands, especially at the cell edge. There is however the possibility of dual 5G SIMs being supported at the same time, which might be more practical. The advantage of this is that you will be able to receive messages and data on a secondary 5G SIM while making a call on the primary SIM at the same time.

Adding early support for Wi-Fi 7 to Qualcomm’s latest Snapdragon Connect suite, the company also offers support for Wi-Fi 6 and 6E along with Wi-Fi 7. It is important to note that although there is still no finalized specification, Qualcomm is leveraging its inside track to support the standard early on. Using High Band Simultaneous Multi-Link technology, it is claimed that the data rate can reach up to 5.8 Gbps using a 320MHz channel in the 6GHz band. With only a latency of 2ms, Qualcomm claims it will be invaluable when it comes to supporting cloud gaming, virtual reality, and many other applications that require a low latency. For this to work, you will, of course, have to use a Wi-Fi 7 router, which is currently only available for purchase in China. It is a good idea to put this on the list of future proof items, then.

Other Snapdragon 8 Gen 2 features

The following are a few of the Snapdragon 8 Gen 2 features which deserve special attention based on the launch presentations and press materials:

  • The Qualcomm Snapdragon 835 is the first Qualcomm processor capable of playing back AV1 at up to 8K 60 frames per second. AV1 decoding is now supported by all major SoCs set to be used in future Android smartphones.
  • It is predicted that dual Bluetooth radios will double the range of connectivity and speed up pairing between devices.
  • The Snapdragon 8 Gen 2 is designed to support new image sensors when it is available in the future, for example the 200MP Samsung ISOCELL HP3 which can be remerged in real-time, as well as Sony’s quad digital overlap HDR technology which is available in the IMX800 and IMX989.
  • Qualcomm has not made any changes to the specs of its ISP capabilities since the introduction of the 8 Gen1. This year when it comes to the camera, there are the same features as last year, such as a 200MP single-shot camera, a 36MP triple-camera capture, and 4K HDR simultaneous capture.
  • There has been an addition of a second AI processor to Qualcomm’s 4th generation sensor hub. This is combined with 50% more memory, so there is now twice the performance on offer here to fully take advantage of technologies like Qualcomm’s always-sensing camera to add privacy features to the screen.

Which smartphones include Snapdragon 8 Gen 2?

Everything you need to know about Snapdragon 8 Gen 2

There is a good chance we won’t see the first smartphones in most consumers’ hands until 2023. However, Qualcomm notes that we may see a commercial release before the end of 2022, especially since the SoC announcement was just released.

Asus, Honor, Iqoo, Motorola, Meizu, Nubia, Oneplus, Oppo, Redmagic, Redmi, Sharp, Sony, Vivo, Xiaomi, and ZTE are some of the partners Qualcomm has confirmed are working on devices powered by Snapdragon 8 Gen 2: Asus, Honor, Iqoo, Motorola, Meizu, Nubia, Oneplus, Xiaomi, Vivo, 3C, ZTE, and Xiaomi. Qualcomm’s list does not include Samsung, as usual, but Snapdragonen 2 will also power Samsung’s Galaxy S23 series throughout the year.

Source: https://www.androidauthority.com/

LEAVE A REPLY

Please enter your comment!
Please enter your name here