Without computing power, there’s nothing to talk about.
Introduction
The metaverse has become hot trending.
On October 28, Facebook officially changed its name and logo to Meta, and soon, many internet companies say they are going to explore the metaverse, including Tencent and Microsoft. Internet giants are decisive in metaverse deployment, and the metaverse has become a new opportunity in this era in the IT sector.
Though metaverse is a huge and obscure concept, it represents a development direction of mankind in the future. This is why it is a hot topic and much chased.
In addition to games, social media and other visually sensible application products, what are the things more noteworthy on the way to the metaverse?
1. Core resources in metaverse era
Neal Stephenson, an American science fiction writer, coined the term “metaverse” in his 1992 novel Snow Crash. Roblox, a hot game company going public in 2021, included the term of metaverse in its prospectus for the first time ever. This is just a start. The fire of metaverse is burning throughout the world, and from tech companies to media, metaverse is the hottest buzzword.
Then, what is the metaverse?
No precise definition is available for the term at present. You may simply think of it as a virtual reproduction of the real world, laying more stress on perception. Most elements in the real world can present in the metaverse, and you may create, play, display, have social interactions, trade or conduct other activities in the metaverse.
Ready Player One directed by Steven Spielberg has shown us a corner of the metaverse in some aspects, offering us a more intuitive understanding of the metaverse. But the real metaverse is huger and more complicated.
To talk about what are the core resources in building the metaverse, we may as well take a look at the technological infrastructures of the metaverse.
From the technological point of view, metaverse building is quite complicated, requiring a variety of hardware, software and technological infrastructures.
Specifically speaking, the following types of resources shall be included:
1. Network and computing technologies, low-cost GPU server, edge computing, location algorithm, etc.
2. Computer game technologies, extremely sophisticated game engines, 3D modeling, real-time rendering, etc.
3. Display technologies, VR, AR, MR, XR, etc.
4. Blockchain technologies featured in decentralization, point to point, open and transparency, value assurance, trustfree efficiency, etc.
5. Supports of several tech clusters, such as IoT, AI.
Ultimately, the development of the metaverse largely depends on the above technology sectors, with the core actually pointing to computing power, i.e. data processing capacity.
We know that large online games require the support of a powerful computing power for better experiences. Though the metaverse is not an online game, it is a virtual world staging activities, like games, and it is a visual, aural and sociable world with economic system…
Building the graphical display of such a virtual world requires the support of computing powers; the creation of virtual contents, including modeling and interaction, is based on the computing power; blockchain technologies guaranteeing decentralized value networks relies on the computing power; artificial intelligence, which may be the “brain” technology in the metaverse, also requires a large amount of computing powers.
Without computing power, there’s nothing to talk about.
The industrial revolution and the modern society driven by the industrial revolution rely on petroleum and other energies. Nevertheless, the computing power is an important foundation and the core resource in the digital age and the metaverse age.
The role of computing power in the metaverse is just as petroleum to engine.
2. Where does the high-performance cost effective GPU computing power come from?
When realizing the core resource for building and exploring the metaverse is the computing power, let’s have a look at the computing market.
At present, the mainstream cloud computing service providers generally concentrate their computing power at multiple data centers consisting of a few hundred thousands of servers in a relatively closed manner, to ceaselessly provide computing services to global networks. With the rapid growth of market demands, these cloud computing service providers will further develop hardware, but the overall price of the computing power is still very high.
For instance, the AI sector requires the support from massive computing, including the provision of a large amount of computing powers. Even GPU computing hardware costs a few hundred thousand yuan to 1 million yuan, and the cost to run AI programs, e.g. Alphago defeated Lee Se-dol, a Go master, is as high as several hundred thousand US dollars each time. The high computing cost is also one of the factors impeding the development of the AI sector.
Due to trust factors such as data security, a conventional centralized computing platform may only be able to provide services to users in some regions, and it is difficult to expand its business to the whole world. Similarly, those large centralized computing service providers build their data centers in remote regions with few natural disasters, and that means they can hardly satisfy the local computing requirements in different regions. In particular, they can hardly satisfy the requirement of some application scenarios, such as autonomous driving.
From the above two factors, the current centralized supply system of computing powers cannot support the development of the metaverse ecology. Instead, it becomes the main factor impeding its development and is hard for landing.
A distributed, high-performance and cost effective computing resource will be an urgent need in developing the metaverse ecology.
In respect of hardware generating the computing power, CPUs are now the main hardware foundation of the mainstream cloud computing service providers to generate the computing power. Though CPUs and GPUs both generate the computing power, CPUs are mainly used for sophisticated logic computing, while GPUs can, with hundreds or thousands of special core processors, conduct parallel computing at a large scale, and is more suitable for visual rendering and deep learning algorithm.
The computing mode provided by GPUs is faster and more cost effective than that provided by CPUs, as the cost of GPU computing is usually one tenth of that of CPU.
The GPU computing power has been deeply built in high-end scenarios requiring massive computing, such as AI, cloud game, autonomous driving, weather forecasting, space observation, etc. Especially, with concentrated growth spurt of these high-end sectors, the market demand for GPU computing powers will be much higher than the demand for CPU computing powers in the future.
From the perspective of metaverse building, many video card manufacturers think the actual effect of the metaverse virtual world in front of end consumers is largely determined by GPUs. With advanced data and tensor / pipeline parallelism technologies, AI will also be able to have the training of large language models be effectively distributed on thousands of GPUs, to obtain the support of massive high-performance computing powers.
Of course, the supply model of computing powers should be further remodeled, and through distributed coordination, the high-performance GPU computing power can be reasonably supplied and distributed. Blockchain is an important technology for solving the trust and efficiency problems. In a blockchain built distributed ecology, all members in the network can autonomously and deeply play a part in running the ecology and be benefited.
At present, the blockchain technology is mainly used in the financial sector. A distributed supply ecology of computing powers built with blockchain technology to accommodate users throughout the world is expected to substitute the current centralized supply system, and after further cost reduction of the computing power and improvement of the supply efficiency, it can also be used for industrial entities.
With the advent of the metaverse age, the market has an urgent requirement for the supply of distributed, cost effective and high-performance computing powers, to which, blockchain may be an ice breaking solution, and a business as well.
3. Unblocking bottlenecks of blockchain computing power
From the point of view of resources regarding blockchain, Web3.0 oriented distributed ecologies have occurred in the storage and CDN sectors, such as IPFS based Filecoin, Arweave and Storj in the storage section, and Theta in the CDN sector, while DFINITY and DeepBrain Chain are the main ones in the computing power sector.
If blockchain is the key for unblocking bottlenecks of computing power, DFINITY and DeepBrain Chain may be noteworthy.
DFINITY, an organization focusing on the decentralized computing power market, has attracted much attention but mainly works on the CPU computing power, while DeepBrain Chain focuses on the GPU distributed high-performance computing power. This is the main difference between them.
DFINITY mainly works on realizing the blockchainization of public network application requirements, such as decentralization of information websites and chat software, etc. The GPU computing power supplied by DeepBrain Chain is more suitable for serving high-performance computing requirements, such as artificial intelligence, cloud game, deep learning, etc.
Established in 2017, DeepBrain Chain is a distributed GPU computing network developed based on blockchain, and contains three important parts: high-performance computing network, blockchain mainnet and GPU computing mainnet.
What features does it have? As it allows GPU computing servers distributed all over the world to be its nodes, it is infinitely scalable in theory, and any computing power providers satisfying the conditions can become a supply node of computing powers and gain benefits. Computing power supply nodes throughout the world can be automatically converted to metro area nodes and edge nodes to satisfy the local computing requirements, and the supply of GPU computing powers would not be affected even if a single node fails, as the error tolerant rate of the whole system is high due to its distributed nature.
Special GPU computing power providers can deploy these GPU servers in IDC machine rooms higher than T3 level to ensure the stability, and on that basis, install DBC software in the severs to access a DBC computing network.
For some idling computing powers, the GPU availability can be improved by accessing the pool of DeepBrain Chain to gain benefits in idle. Therefore, a large number of distributed computing powers will be gathered in DeepBrain Chain, its computing power generation cost will be much lower than that of a centralized computing platform, and the cost to acquire the GPU computing power will also be greatly reduced.
With the launch of the computing mainnet of DeepBrain Chain, anyone in the world can freely join their GPU resources meeting the DeepBrain Chain network requirements into the net and rent GPU resources in the net to support their business development.
DeepBrain Chain has unique advantages in supplying high-performance and cost effective computing powers.
To sum up, the metaverse will be the theme of technological development in the next decade. Computing powers are the power and a core and scarce resource on the way to the metaverse age. Those who can provide high-performance and cost effective computing resources will be able to better grasp and utilize the resources and really seize the opportunity of the metaverse age.