EMC Protocol
Last updated
Last updated
EMC Protocol unifies various compute resources and applications under a single, decentralized platform. It provides a compute marketplace where users can easily tap into the EMC network by utilizing âcompute sockets,â discovering the capacity they need for specific tasks or applications. Payments, settlements, and resource allocations are executed via smart contracts and crypto assets, enabling secure and immediate transactions without reliance on centralized intermediaries.
To support broader ecosystem functionality, EMC Protocol delivers a range of foundational services that simplify user engagement and product development. These include on-chain data storage, identity authentication, and a robust framework for integrating AI workloads. Innovations such as intelligent routing and data caching further optimize the utilization of computing power, enhancing network throughput and reducing latency. In parallel, EMC Protocol prioritizes security by incorporating advanced encryption and multi-layer protective mechanisms to preserve the confidentiality and integrity of user data and transaction information.
By consolidating compute power from different sources, EMC Protocol transforms it into a single, cohesive marketplace that fosters a thriving environment for AI and other computation-intensive services. This model ensures that participants from small-scale developers to large enterprises can access or monetize compute resources in a flexible, open, and highly efficient Web3 landscape.
Validator nodes play a critical role in maintaining the legitimacy of the compute consensus process and settling compute rewards within the EMC ecosystem. Node owners must stake a certain amount of EMC tokens in the on-chain Compute Consensus Contract to earn validator rights. Validator services then subscribeâvia an RPC nodeâto compute-invocation events or retrieve real-time invocation and billing records through the EMC Computing Network Dashboard (open API). Upon receiving these records, validators verify data integrity through cryptographic hash fingerprints and submit validation results back to the main chainâs Compute Consensus Contract. An IBFT mechanism finalizes and records these results, while the GPU Computing Reward contract disburses rewards to the Computing Node Owner based on validated compute tasks. After each compute consensus is complete, the validator node owner also receives compensation through the Computing Consensus Reward contract.
RPC nodes are designed for environments with robust network bandwidth, high-performance CPUs, and ample memory. By offering secure HTTPS-based API services, these nodes facilitate the initiation of compute requests, as well as the subscription to compute-invocation events. Acting as gateways, they bridge user queries and the broader EMC network, ensuring that requests and data flow reliably across the system.
Smart routing nodes are a key pillar of the EMC network. Requiring stable bandwidth and sufficient CPU and memory resources, each relay node must accommodate 500+ concurrent long connections from compute nodes. By serving as intermediaries between RPC nodes and compute nodes, they eliminate the need for each compute node to open a public TCP port or maintain a static public IP.These relay nodes perform several core functions within EMC:
Network Connectivity: They link all validator nodes and edge compute nodes, enabling consistent communication and coordination across the network.
State Synchronization: The EMC networkâs global state machine must be recorded and synchronized by all validator nodes. Smart routing nodes function as crucial intermediaries, ensuring state information is properly relayed and updated, thus maintaining consistent global state across the network.
Optimized Request Routing: By factoring in the shortest path and each compute nodeâs load capacity, smart routing nodes route requests to the most suitable compute node and return responses to the requester. This design boosts performance and stability within the EMC topology.
Real-Time Communications: These nodes also support real-time broadcasting and messaging across the entire network, enabling instantaneous communication and interaction among all nodes.
Smart routing nodes earn rewards for contributing both network bandwidth and any additional compute resources.
Computing nodes form another cornerstone of EMCâs network infrastructure. These nodes handle the actual compute tasks that power the ecosystem, providing the computational capacity needed by AI agents or other workloads. Any device capable of running EMCâs serviceâwhether a personal computer, a server, or even a smartphoneâcan serve as a compute node. Acting as endpoints within the network, they receive compute requests through smart routing nodes, execute the assigned tasks, and return results to improve EMCâs overall efficiency and throughput. Compute nodes earn two primary categories of rewards:
Compute Proof: Granted once a node is recognized and validated as an official part of the network.
Compute Tasks: Gained by successfully completing AI Agent workloads.
To handle large-scale computational demands, EMC supports the concept of a computing pool, composed of multiple compute nodes working in unison. By coordinating resource sharing and task scheduling, these pools enhance the overall efficiency and performance of intensive workloads. Often equipped with high-performance GPUs, a computing pool can tackle tasks such as AI model training or AI application inference at scale. These node clusters also commonly integrate various AI frameworks (e.g., TensorFlow, PyTorch), offering a flexible, efficient, and cost-effective environment. Compared to a single compute node, a computing pool delivers greater computational power, higher reliability, and meets the needs of large-scale workloads more effectively.
EMCâs network features 3F+1 validator nodes, which rely on IBFT (Istanbul Byzantine Fault Tolerance) to achieve consensus. Once two-thirds of these validators confirm a block, a consensus is established across validator nodes. This mechanism combines fairness, efficiency, low energy consumption, and minimal transaction fees, bestowing EMC with distinct advantages in rapid application interaction and low-cost transactions.
Node staking guarantees the networkâs reliability in both compute scheduling and reward distribution.
Validators must stake a designated amount of EMC as collateral. This requirement incentivizes honest behavior, as any malicious actions risk forfeiture of staked assets. Rewards for validators correlate with the quantity of EMC staked, encouraging those who stake more to participate more diligently in securing the network.
Smart Routing Nodes also require EMC staking. Since the number and distribution of these relay nodes directly influence the networkâs performance and stability, EMC staking ensures better network management and quality of service. Once selected and upon successful task completion, these nodes receive corresponding rewards.
Compute Nodes may stake EMC to validate their identity and commitment to the network. Their completed AI Agent tasks and corresponding rewards scale with their staked positions. Compute nodes may also opt to forgo staking and handle intermittent workloads, such as real-time communication or AIGC requestsâtasks that are less resource-intensive and can coexist with other compute demands. This flexibility benefits smaller or more specialized nodes, broadens opportunities for network participation, and furthers the growth of the EMC ecosystem.
A compute socket is a mechanism that tokenizes AI compute resources. Each AI Agent can issue tokenized assets by leveraging these sockets, giving end users the option to purchase or trade tokens representing specific compute resources. By rendering both compute and applications as tradable tokenized assets, EMC fosters liquidity and streamlines resource flow in the AI domain.
âEMC is open source and its design is public. No one owns or controls EMC, and everyone can participate.â (Adapted from the official Bitcoin website.)