Heterogeneous computing - Apsara 2022

Date: Oct 1, 2022

Metaverse, digital twin and virtual reality mean we are evolving from one-way interaction to full-element interaction. Apsara conference 2022 evolution is to achieve realism through immersive iterative real-time interaction, and realism brings about the improvement of human cognitive efficiency Apsara conference 2022.

Such a judgment stems from the famous pyramid model of human learning and cognition proposed by an educator in the last century Apsara conference 2022. The model believes that the most efficient learning method for humans is Feynman learning method, but it is not suitable for everyone. For ordinary people, it is more conducive to improve learning efficiency and cognitive efficiency through simulation practice. The most efficient way in simulation practice is real-time interaction of all elements. For example, pilots will undergo long and hard training in the flight simulation cabin before they actually fly into the sky. This approach is very close to the learning efficiency of real flying and is safer and more economical Apsara conference 2022.

Apsara conference 2022 improvement of cognitive style is essentially the improvement of productivity. As an important concept of human society, productivity makes it necessary for us to seriously understand the metaverse industry. What opportunities and challenges does it bring?

Apsara conference 2022 seven-layer Metaverse model is a relatively complete model for describing the metaverse industry. From the user experience layer to the creator economy, and finally to the infrastructure layer, it summarizes all related concepts, presentation layers and technologies, involving many industry.

Apsara conference 2022 mapping these concepts and technologies to the common areas of cloud computing will eventually have a very large impact on cloud computing in five aspects: computing, application, communication, human-computer interface and business ecology, which is also an unprecedented challenge for cloud computing. Even leapfrog challenges.

Apsara conference 2022 take smartphones, for example. Compared to PCs 20 years ago, smartphones today are a huge leap forward in every way. Thanks to the upgrade of the interactive mode of smartphones, an operating system that supports multi-touch has emerged, and finally a rich application ecology has grown on this platform, and the rich application ecology has in turn promoted computing performance. The continuous improvement of communication performance also promotes the continuous improvement of communication performance. Today, the performance and functionality of mobile phones even surpass laptops, and mobile phones provide a variety of sensors and interactions that laptops do not have Apsara conference 2022.

Apsara conference 2022 interaction methods faced by the Metaverse may be real-time 3D, gesture interaction, voice interaction, etc. The ultimate challenges brought by these interaction methods are more real-time computing and interactive communication. For example, there was only touch screen before, but now there are more recognition gestures, recognition Scenes and sensors that recognize space all require real-time computing, and the requirements for communication are more stringent, requiring larger bandwidth and lower latency.

At the same time, the business ecology faced by the Metaverse will be more complex than any previous IT system, so another problem that needs to be faced is how to operate this entire business system.

Apsara conference 2022, to sum up, the three main problems that cloud computing needs to face when adapting to the Metaverse business are computing, communication and ecology. It requires 10 times + computing performance and computing scale, 10 times + communication and low latency, 10x+ eco-efficiency and openness are required.

In terms of computing performance and scale, 10 times more graphics computing is required to render higher-fidelity 3D models, and ray tracing of more scenes is required in order to achieve a true description of the real world; in terms of AI, stronger To train and run AI models, more computing power is required for streaming media processing; in terms of bandwidth and delay, larger bandwidth and lower delay are required to be responsible for rendering the rendered high-fidelity images. Pushing to the client also requires low-latency bandwidth to solve the training and inference of large-scale AI models, and high-performance bandwidth to solve the communication between humans and virtual humans through subtle expressions Apsara conference 2022.

In addition, the bigger challenge comes from ecology. The technical ecology of the Metaverse is very complex, and the previous case-by-case industrial collaboration method could not expand the scale of the industry to 10 times or even 100 times. Therefore, we need new ecological collaboration methods to make collaboration more efficient and to make collaboration systems more open.

In response to the above three aspects, in the past ten years, Alibaba Cloud has done a lot of work on technical performance indicators and computing scale. For example, at the Apsara conference 2022 last year, the fourth-generation Shenlong architecture was released, and the large-scale networking of eRDMA was realized through hardware acceleration. The latency was as low as 5 microseconds, which was at the leading level in the industry. The ability to recognize subtle expressions.

Alibaba Cloud's GPU computing instances for AI and Metaverse have also made great progress. In the past few years, the performance of GPU instances has been improved by a hundred times. The 7th generation GPU instances were released last year, and there have also been significant improvements in image computing and AI computing. What's more worth mentioning is that Alibaba Cloud has achieved a very large-scale accumulation of computing. Through commercial operation, it has realized the largest GPU cluster in Asia, with a total computing power of more than 1700P, and is still maintaining rapid growth.

In addition to technical aspects, business ecological issues are also huge challenges we face. To prosper the industry, it is necessary to enhance the productivity of the industry and straighten out the production relations of the industry. Together with our partners and customers, we have built Alibaba Cloud's visual product system through ecological operation, and have achieved a lot of accumulation in products.

As we all know, heterogeneous computing is a very difficult computing system, and our production and research team has also made great efforts here. By means of software, the common problems of heterogeneous computing in various industries are solved, and finally more efficient, higher performance, more economical and easy-to-use heterogeneous computing products are realized.

More importantly, we have built visual computing software services with the most leading partners in all walks of life, and have already implemented many products on Alibaba Cloud.

We divide software services into three layers to solve the problem of industrial ecology.

General layer: Two of the more important products are the cloud XR platform and the cloud real-time 3D editor platform, which mainly solve the problems of 3D content and application hosting environment and the problem of 3D content and application R&D and production.

Industry layer: I hope to build a huge matrix of industry editors with editor partners from all walks of life to solve the problems of their own editors in various industries. The editor helps implement Metaverse enterprise applications more efficiently, engages everyone, and increases the productivity and creativity of the entire industry.

Ecological layer: The ecological layer mainly solves the problems of content production and content transaction. It is responsible for providing materials for the first two layers, so it also needs to focus on construction.

In the near future, these products will be launched on Alibaba Cloud, and they will work with industry editors and general editors to solve problems in all walks of life.

The most common application scenarios of the current Metaverse are shown in the figure above. After normalizing the requirements of the scenarios, they are finally concentrated in 5 levels, and the requirements of these 5 levels can be matched with the previous product system one-to-one.

For example, an important part of autonomous driving before the road test is to do real-time simulation of the 3D environment, which can generally be done through industry editors. With the industry editor, there is no need to use the general 3D editor to build houses, roads, etc., but to use the simulation platform to quickly complete the environment construction on a white map, greatly improving the efficiency, and even providing different weather, wind speed, Light conditions, rain and snow environment to simulate the real driving environment to detect the operation of the automatic driving program.

But not all industries have such editors, so we provide two more general products to solve similar problems in various industries.