Today, the demand and threshold of VR chips are relatively clear, while AR chips are still relatively early.
Since the second half of last year, the concept of the Metaverse has become one of the biggest hot spots in the technology industry, and major technology companies have announced that they will vigorously deploy the Metaverse. At present, it is generally believed that AR/VR will become an important entrance to the Metaverse, and therefore it has become an area where major companies invest heavily. In the field of AR/VR hardware, Facebook’s Quest series has become the most popular VR device in the world, and Chinese giants such as ByteDance and Tencent have also made major investments recently, of which ByteDance has acquired the highest market share among Chinese companies. The VR equipment manufacturer Pico, and Tencent also recently acquired Black Shark, will focus on VR hardware equipment.
In the field of AR hardware, in the second half of last year, Ray-Ban and Facebook jointly launched the first smart glasses Ray-ban Stories for the public. Companies such as Huawei in China have also launched concept prototypes. Released in 2022 or 2023, it is expected that AR hardware will also embark on a path of rapid growth in the future.
The performance of ARVR hardware is extremely critical to the user experience that it can support, and the chip in it has become the core component that enables its hardware performance. At present, in the ARVR market, the most important supplier is undoubtedly Qualcomm, and its Snapdragon XR series has become the choice of almost all VR devices. At the Qualcomm Investment Day at the end of last year, Qualcomm CEO Cristiano Amon also emphasized The ARVR (XR) market will become one of Qualcomm’s most critical development directions in the future, and it is expected that related chips will be shipped in the order of tens of millions in 2021. Earlier this year, Qualcomm and Microsoft officially released their AR cooperation plan at CES 2022, in which Microsoft and Qualcomm will jointly develop customized Snapdragon chips for AR glasses, which will further include support for Microsoft Mesh Platform and Qualcomm Snapdragon spaces are supported by two major AR development platforms.
Judging from the dynamics from last year to this year, the ARVR chip market is in a stage of rapid growth, and Qualcomm currently has an almost monopoly in this market. However, with the further growth of the market size, more chip manufacturers will definitely enter the market.
The technical threshold of AR VR chips
The technical threshold of ARVR chips determines how many chip companies can enter the market to compete. We might as well start with the analysis of VR chip technology.
First of all, in terms of architecture, the architecture of the VR chip is basically similar to the SoC of other smart devices. A processor (AP) runs the operating system and controls the operation of the overall hardware. Therefore, VR chips first need to have a strong processor, but this is not unattainable at present – the use and integration of ARM’s high-performance processor IP has become the standard operation of most SoC manufacturers, so we think The processor in a VR chip is not an insurmountable threshold.
Among the many IPs of VR SoCs, the most important IP that determines the user experience is the GPU responsible for rendering and display. At present, the main application of VR equipment is games, and almost all VR games are 3D images (because 3D images can provide users with the best immersive experience), so the performance of the GPU determines the number of polygons and the number of polygons in these 3D games. rendering effect. The 3D image quality of the current VR games is roughly equivalent to the PC and console games 10 years ago. Therefore, with the expansion of the VR game market and the formation of the ecology, the game image quality will become the core indicator of competition between consoles, and this will also This will drive demand for GPU rendering capabilities in VR chips. Different from the desktop GPU, the GPU in the VR chip must consider the energy efficiency ratio, because the current mainstream VR devices are powered by batteries, so the energy efficiency ratio will determine the battery life of the VR device, and it is also necessary to ensure that the VR device does not dissipate heat. problem.
Aside from game image quality, a big requirement for user experience in VR devices is resolution. At present, an important user experience problem in VR devices is the “screen window effect” caused by insufficient screen resolution (that is, the image that the user sees seems to be separated by a layer of screen window), so VR devices require 4K or even higher resolutions more than a cell phone. Therefore, we expect that in the next few years, the screen resolution of VR equipment will also be the target that major manufacturers are chasing, which also requires the GPU of the VR chip to have sufficient performance to support high-resolution screens.
In addition to GPUs, the other most important IP in VR chips is artificial intelligence accelerators. Unlike traditional mobile phones, many interactions in VR devices require the intervention of artificial intelligence. For example, an important requirement of VR equipment is the positioning and tracking of equipment and users, and to integrate this information into the virtual environment of the game, which requires the use of SLAM technology in artificial intelligence to achieve high quality. In addition, in the current VR devices, the interaction is increasingly using advanced technologies such as hand tracking and eye tracking to achieve natural interaction, and these tracking require the use of artificial intelligence models, and as the tracking accuracy increases, the model needs to be computing power is also rising.
Finally, artificial intelligence is also an important technology to solve the bottleneck of GPU rendering capacity. The increase in screen resolution and the user’s pursuit of image quality put a lot of pressure on the GPU, but in fact users don’t notice all the pixels on the screen at the same time, but only focus on the pixels near where their eyes are focused.Therefore, a technical solution is foveated rendering, which uses eye tracking technology combined with artificial intelligence models to determine where the user’s eyes are focused and render high-quality rendering in the corresponding places, while the user does not pay attention. You can reduce the rendering quality, thereby reducing the pressure on the GPU. At present, this technology has been used in Sony’s PSVR2, and it is expected to be used in more VR devices in the future.
From the above analysis, we can see that artificial intelligence in VR devices is even more important than in mobile phones, because these artificial intelligence models are first and foremost the key to enabling important features of user experience, and they must also run in real-time in VR devices. , and it is difficult to rely on cloud technology. The complexity of these artificial intelligence models determines that there must be a powerful enough artificial intelligence accelerator on the VR chip, with strong computing power and good energy efficiency ratio, to support these applications.
In addition to GPU and artificial intelligence, wireless connectivity is also an important part of VR chips. For some high-quality games, the current mainstream approach is to run the game on a desktop computer and stream the images to the VR device, which requires a stable and high-speed WiFi connection.
Today, the demand and threshold of VR chips are relatively clear, while AR chips are still relatively early. Compared with VR, AR eyes do not need to run a lot of image rendering continuously, but need to be able to wear all day, and have very strict requirements for weight (for example, the weight is comparable to ordinary sunglasses), which further limits the battery capacity on the eye , so for the chip, it will be necessary to further optimize the low power consumption. Whether from Magic Leap or Microsoft’s Hololens, AR image rendering is also an extremely important user experience in AR glasses, so the GPU of AR chips will be a challenge, that is, how to balance rendering capabilities and power consumption. Artificial intelligence is similar. AR glasses also need to do SLAM to users, and also need to support interactions such as hand tracking. Therefore, we believe that AR chips can be considered as lower-power version VR chips.
AR VR chips and China
At present, Qualcomm is the absolute leader in ARVR chips, but with the further maturity of the ARVR market, we believe that Chinese chip companies have considerable opportunities.
We believe that the basic architecture of ARVR chips is similar to that of smartphone chips, except that the performance requirements and importance of the IP in them are slightly different from smartphone chips. For example, cellular network modem is the absolute core of smartphone chips, and it is also an area that China’s chip industry has taken a long time to catch up with. However, cellular wireless network is not a necessary option in ARVR; Chinese chip companies already have considerable experience in IP, GPU, AI acceleration, etc. Therefore, we believe that from a design perspective, as long as Chinese chip companies can cooperate with relevant ARVR equipment companies to deeply understand the specific needs of chips , we are optimistic that Chinese chip companies can enter the ARVR field in the next few years.
From another point of view, since ARVR chips have strong requirements for the power consumption and energy efficiency ratio of chips, advanced semiconductor processes will be of great help for such high energy efficiency chips. It is expected that the latest technology will be used in the next few years. Semiconductor process (7nm and below) to manufacture.From this point on, China’s semiconductor manufacturing needs to catch up further to benefit from the development of ARVR chips, which not only include semiconductor chip manufacturing processes, but also core technologies such as advanced packaging that can improve the overall system energy efficiency from another dimension. .
*Disclaimer: This article was originally written by the author. The content of the article is the author’s personal opinion. The reprint of Semiconductor Industry Observation is only to convey a different point of view. It does not mean that Semiconductor Industry Observation agrees or supports the view. If you have any objection, please contact Semiconductor Industry Observation.
Posted by:CoinYuppie，Reprinted with attribution to:https://coinyuppie.com/the-chip-opportunity-behind-the-metaverse/ Coinyuppie is an open information publishing platform, all information provided is not related to the views and positions of coinyuppie, and does not constitute any investment and financial advice. Users are expected to carefully screen and prevent risks.