HomeArchitecture

System Architecture

ComputeNet is built on a modular architecture where each component can evolve independently while preserving protocol interoperability.

High-Level Architecture

This diagram shows the intended architecture, not a fully active pipeline. Only bounded execution and signed evidence are publicly live; production native consensus remains inactive.

Compute Input
Execution
Verification
Consensus
Layer 1

Deterministic Execution

Layer 2

Receipt Generation

Layer 3

Native consensus — inactive

Core Components

Validator Nodes

Current processes execute and sign bounded jobs under one administration. Native consensus and genuinely independent operators remain unactivated.

  • Receipt validation
  • Output verification
  • Validity attestation
  • Consensus participation
  • FastAPI-based services

Deterministic Runner

Registered document hashing and exact integer aggregation are reproducible. Arbitrary workloads are not enabled by this service.

  • Stable execution
  • Predictable outputs
  • Reproducible hashing
  • Deterministic receipts
  • Workload isolation

Receipt Engine

Execution results are transformed into structured receipts containing job metadata and signed execution evidence.

  • Job identifiers
  • Manifest hashes
  • Execution hashes
  • Validator metadata
  • Policy-bound signatures

Proof Engine

Recomputes supported bounded results and checks signed evidence. ZK, zkVM and hardware-attestation integrations remain research, not live guarantees.

  • Local proofs
  • ZK placeholders
  • Aggregate structures
  • Future zkVM
  • Hardware attestation

Consensus Layer

The compound native candidate is isolated-tested and unactivated. It proposes accumulated resource-work ordering conditional on eligible useful computation; confirmations remain reversible.

  • Independent verification
  • Fresh resource work — proposed
  • Deterministic fork validation
  • Probabilistic confirmations — proposed
  • No production consensus activation

Telemetry Systems

Current public observations cover gateway availability and historical reads. Global uptime, peer latency and consensus metrics are not measured.

  • Uptime — not measured
  • Peer liveness — not measured
  • Receipt activity
  • Consensus — inactive
  • Runtime metrics

Peer Registry

The legacy API lists configured host metadata. Secure discovery, sync and recovery are isolated candidate capabilities, not authenticated live peer-health measurements.

  • Bootstrap peers
  • Gossip synchronization
  • Liveness scoring
  • No identity-based security weight
  • Discovery protocols

Public Observer

Public-facing infrastructure providing readable telemetry and bootstrap peer visibility.

  • Public telemetry
  • Peer visibility
  • Genesis candidates
  • Explorer access
  • Read-only mode
Verification Flow

Verification Pipeline

The protocol approaches verification through a layered pipeline where verification is designed to be cheaper than execution.

Job Manifest

Deterministic execution blueprint

1

Execution

Reproducible workload processing

2

Hashing

Immutable output fingerprint

3

Receipt

Portable evidence artifact

4

Validation

Independent re-execution

5

Consensus

Proposed; inactive

6

Data Flow Architecture

The protocol processes compute through a structured flow: manifests define workloads, runners execute deterministically, receipts capture evidence, and validators reach consensus.

1

Input Layer

Job manifests define deterministic workloads

2

Execution Layer

Reproducible compute with stable outputs

3

Evidence Layer

Receipts capture execution proof

4

Consensus Layer

Validators attest and finalize

>// Compute flow visualization
manifest.json
deterministic_runner
execution_hash
compute_receipt
validators[]
verify(receipt)attest()
consensus_ratio:1.00
Future Research

Long-Term Architecture Goals

The architecture is designed to evolve. Future research directions may include zkVM integrations, recursive proof systems, hardware attestation, and trusted execution environments.

View Research

zkVM Integration

Research

Recursive Proofs

Planned

Hardware Attestation

Exploration

Async BFT

Research