Logical Flow and Communication

Components

  • Cerebrum Cloud (Web Application): User interface for the public.

  • Cerebrum Node (Application): User interface for GPU suppliers.

  • Cerebrum Validator (Application): User interface for validator nodes.

  • Cloud Infrastructure: Securely hosts Cerebrum Cloud, Cerebrum Node, and Validator platforms.

  • Validator Nodes: Geographically distributed nodes verifying transactions.

  • Blockchain: Public ledger for recording all transactions.

  • Load Balancer/Reverse Proxy: Distributes workload for fault tolerance and error checking.

  • GPU Supplier Network: Network of individual and supplier-owned GPUs.

  • Secure VPN Tunnels: Secure communication channels between components.

  • Docker Containers: Software packages containing GPU access and communication software.


Data Flow

Public Interaction

  • Any user can access Cerebrum Cloud via web application.

Resource Selection

  • Users can select the desired virtual cluster and specify processing power needs.

Job Scheduling

  • Cerebrum Cloud utilizes AI scheduling algorithms to identify the most suitable GPU node(s).

Factors include

  • Network latency.

  • Historical node performance.

  • Real-time GPU workload.

Secure Communication

  • Cerebrum Cloud establishes a secure next-generation VPN tunnel with the chosen node(s) using protocols like ChaCha20.

Task Distribution

  • Cerebrum Cloud transmits tasks amongst the collaborating nodes on the network.

Reverse Proxy/Load Balancing

  • Load balancer distributes the task(s) across available GPUs within the chosen node(s).

GPU Processing

  • Docker container on the supplier's GPU(s) endpoints to execute the task.

Results Transmission

  • Processing results are sent back through the secure tunnel to Cerebrum Cloud.

Result Delivery

  • Cerebrum Cloud delivers results to the user.

Transaction Recording

  • All transactions (resource allocation, rewards) are recorded on the blockchain for transparency.

Supplier Rewards

  • Our native tokens are distributed to suppliers and validators based on their GPU contribution.

Last updated