LogoLogo
  • Blendr's Network
    • Executive Summary
  • Problem and Solution for Blendr - A Comprehensive Approach
  • Blendr's Decentralized Architecture
    • Blendr Compute Mesh (BCM)
    • BlendrChain
    • Blendr Token ($BLENDR)
    • BME | Native Token
    • Digital Asset Rights Ledger (DARL)
    • Node Engagement Protocol (NEP) and Creator Engagement Protocol (CEP)
    • Node Manager Selection
    • Task Scheduler Optimization active GPUs
    • Data Encryption and Integrity - Quantum-Resistant Encryption
    • Blockchain Smart Contract for Task Verification - Enhanced with Oracles
    • Reward Mechanism for GPU Contributors
    • Token Staking for Non-GPU Holders
    • Multi-Tier Pricing (MTP) Algorithm
    • Voting Power
    • Reputation Scoring System
    • Resource Allocation
    • Burn and Mint Equilibrium (BME)
    • Secure Transaction Protocol
    • Allocation Optimization
    • Dynamic Pricing Model
    • Incentive Mechanism (Burn and Mint Equilibrium - BME)
    • Token Staking
    • Reputation and Incentive Mechanism
    • GPU Selection and Task Allocation
    • CUDA + Blendr
    • Flow for User
  • In conclusion
Powered by GitBook
On this page
  1. Blendr's Decentralized Architecture

Allocation Optimization

PreviousSecure Transaction ProtocolNextDynamic Pricing Model

Last updated 1 year ago

The objective is to minimize the total processing time or cost while ensuring that the computational demand of tasks is met by the allocated nodes. Let xijx_{ij}xij​ be a binary decision variable that is 1 if task tjt_jtj​ is assigned to node nin_ini​ and 0 otherwise.

Minimize the total cost:

Minimize∑i=1m∑j=1kxij⋅Cij{Minimize} \sum_{i=1}^{m} \sum_{j=1}^{k} x_{ij} \cdot C_{ij}Minimizei=1∑m​j=1∑k​xij​⋅Cij​

Subject to the computational demand and capacity constraints:

∑i=1mxij⋅Dj≤Capacityi, ∀i\sum_{i=1}^{m} x_{ij} \cdot D_j \leq \text{Capacity}_i, \ \forall ii=1∑m​xij​⋅Dj​≤Capacityi​, ∀i
∑j=1kxij=1, ∀j\sum_{j=1}^{k} x_{ij} = 1, \ \forall jj=1∑k​xij​=1, ∀j

Where CijC_{ij}Cij​ is the cost of assigning task tjt_jtj​ to node nin_ini​, and Capacityi{Capacity}_iCapacityi​ is the computational capacity of node nin_ini​.