DeepBrain Chain Progress Report #131 04.01–04.15(2023)

6 min readApr 18


Dear DeepBrain Chain community members,

Please see below for the latest update on the project.

I. Product Development Progress

GPU Cloud Platform:

  • Cloud game platform wallet display page optimization and testing;
  • Cloud game platform wallet registration function testing;
  • Continue to improve the wallet login logic;
  • Added a communication interface between the client and the go service on the cloud game platform;
  • Modification of the rental logic of a single GPU on the GPU cloud platform;

Distributed Computing Network:

  • The API for dbc integrated diskless console has been launched;
  • Worked on the basic communication test optimization of the websocket module of the service program;
  • Cloud game function development, increased user login and device registration related interfaces;
  • Cloud game service program: the basic framework (IPC/WebSocket/Logging and other modules) has been set up,
  • Worked on the login process of docking the front end and the Go server;
  • Cloud game client shell: added support for caching;
  • The usage of the remote control SDK is being debugged;

DeepBrain Chain Mainnet:

  • The mainnet is upgraded to V2.5. It contains more than 10 function updates, including:

(1) Parameter changes on the chain:

  • Council election cycle changed to 120 days;
  • Computing power rental per unit changed to 60%;
  • The staking module penalty delay time has been changed from 28 days to 14 days;
  • The penalty will no longer be automatically added when the machine is offline for more than 10 days, but will be added when the machine actively goes online;

(2) New features:

  • Added a regular reward function for council members: the chief council member and the other top two council members receive rewards every month;
  • Council Chief: can be queried through council — prime; the second and third are the top two elected members of the Council other than the Chief after being ranked by backing stake. For specific prime election algorithm and backing stake calculation method, refer to:;
  • Reward distribution time: It is distributed every 30 days, and the first distribution is 15 days after the election of a new Council members. The specific calculation formula for the issuance time is: the block height is divided by 30*2880 and the remainder is equal to 15*2880;
  • Number of rewards: the Chief’s reward is min(5000USD, 600,000 DBC); the second or third ranked Council members get min(2000USD, 200,000 DBC). Among them, min means to calculate the value of two numbers and take the lower value. If the value of 600,000 DBC exceeds 5,000 USD, 5,000 USD’s equivalent of DBC will be rewarded; if the value of 5,000 USD exceeds 600,000 DBC, a maximum of 600,000 DBC will be rewarded;
  • Added an interruptible rental module. Machine online/offline/rental/reporting functions are separated from the original functions;
  • After allowing the machine to be online for one year, if it is idle for more than 10 days, it can be offline without penalty;
  • Bugs fixed:
    (1) It is now forbidden to create a rental order with the number of GPUs being 0;
    (2) Fixed the problem that redistribution does not remove the pledge of the audit committee when there is no consensus when the machine goes online, and the problem that may trigger punishment;
    (3) Fixed the problem of using the latest version of code to synchronize chain data panic;
    (4) Fixed the problem that the reward cannot be received after the machine exits;
    (5) Fixed the problem that the same person rents different GPUs of the same machine, and some statuses are incorrect when the lease ends;
  • Added asset module, added locking-when-transferring and unlocking-when-expired unlock function;

II. Marketing Progress

  • The DBC computing power mainnet has been officially launched, and as of April 15th, 2023 Singapore time, the GPUs in the computing power mainnet have reached 1,607. Other latest data on-chain are: there are 10 computing pools participating, total computing power value reached 492,516.74, these GPUs have pledged a total of 162,015,000, GPU rental rate reached 72.80%, and accumulated rental income reached 3,085,146,570 DBC, please refer to the official website for details:
  • Today Korea News reported that DeepBrain Chain’s third computing pool in South Korea, ZEST CLOUD, will launch cloud platform services to expand cloud Internet cafe business, link:
  • The 11th DBC Council meeting was successfully concluded. Council members and community members from more than 10 countries around the world discussed the implementation of DBC ecosystem applications, including DBC Global AIGC Competition, Singapore DeepLink cloud games, etc. Meeting minutes:
  • DeepLink has achieved great success in the workshops at VIT Chennai, IIITB and other universities in India. Students have witnessed how DeepLink and blockchain technology can change cloud games;
  • The HK NIGHT VC MEETUP event in Hong Kong was successfully concluded. Feng, the founder of DBC, introduced DBC and DeepLink to many investors and practitioners present at the meeting. Feng said that DBC will serve the AI and cloud gaming industry and provide GPU computing power resources for many enterprises; at the same time DeepLink will bring a large number of web2 users to web3 to realize the sharing of game resources between Internet cafes and individuals;

Liquidity:3.607 B(71.1%)

Token Pledged:1.033B(20.3%)

Confirmed block height:1,958,420

Signature transactions:1,932,136


Holding account:27,039
Total supply:5.072B
Inflation rate:5.45%

III. Ecosystem Building

  • Continuing to push for product application in the Korean, Singaporean, Vietnam and other markets outside of China; Korean DBC-IDC officially established and Cloud Internet Cafes are being set up;
  • In order to ensure the rental rate of your DBC computing nodes, please follow the recommended configuration and GPU model of the mainnet when launching online;

HaibaoGpu Cloud Ecosystem

  • Completed the personalized deployment of large language models. At present, Stanford’s Alpaca-LoRA based on LLaMA, 7 billion parameters, a single GPU A5000, and 3090 24G video memory runs smoothly. Released relevant tutorials to Zhihu and strategizing scientific research communities;
  • Commonly used AIGC application stable diffusion, LLaMA containerized image production and development;
  • Contacted users from Beijing Institute of Technology and Beijing University of Posts and Telecommunications to promote computing power;
  • Summarized the main issues of current customer complaints and organized the Q&A manual.

Contact information for joining the network:


— — — — — — — — — — — — — — — — — — — — — — —

Try our product as an AI user or a GPU power provider:

DBC Wiki:

Main social media platform:

Telegram (English) :


Reddit :

Official e-mail address:

About DeepBrain Chain

DeepBrain Chain is the world’s first AI computing network powered by blockchain technology.

Our Mission: To connect global computing power, build a distributed high-performance computing power network, and construct computing power infrastructure for the era of 5G +AI.

Our Vision: Become the world’s largest computing network.

DeepBrain Chain’s computing power network can help AI practitioners/enterprises/universities/research institutions, cloud gaming, rendering, blockchain and other users to save on computing power cost, improve computing efficiency and enhance product experience.

Yours sincerely,

The DeepBrain Chain Team




DeepBrain Chain-building the most important computing infrastructure in the era of AI + Metaverse

Recommended from Medium