tfgrid/docs_depin
10
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
b87c107b0c
docs_depin/docs/depin_specs/compute_networks/io_net.md
mik-tf b87c107b0c update ionet page
2025-02-02 11:17:53 -05:00

3.9 KiB
Raw Blame History

title description sidebar_position
io.net Detailed information about io.net 7

io.net

Overview

io.net is a decentralized AI computing network built on Solana, designed to aggregate underutilized GPU resources globally to power AI and machine learning applications. By leveraging a Decentralized Physical Infrastructure Network (DePIN), io.net connects independent data centers, crypto miners, and projects like Filecoin and Render to provide scalable, cost-effective computing power. Founded by Ahmad Shadid (later succeeded by Tory Green in 2024), the platform aims to democratize access to GPU resources, reducing costs by up to 90% compared to traditional cloud services.

Key Features

  • Currency: $IO (native token) and $IOSD (USD-pegged stablecoin).
  • Buying Currency: Available on exchanges like Binance, where it was listed via a Launchpool campaign in June 2024.
  • Utility:
    • Payments: $IO is used to pay for GPU/CPU rentals and transaction fees; $IOSD offers stable pricing.
    • Staking: Stake $IO to secure the network, earn block rewards, and participate in governance.
    • Governance: $IO holders vote on protocol upgrades and ecosystem decisions.
  • Automation:
    • AI-driven dynamic resource allocation optimizes task distribution based on hardware specs, location, and cost.
    • Smart contracts automate payments, slashing penalties, and reward distribution.
  • API: Yes. Integrates with Ray and Kubernetes for scalable AI/ML workflows.
  • User Perspective:
    • Developers: Deploy AI models, access 1M+ GPUs, and use tools like IO Cloud for cluster management.
    • Providers: Monetize idle GPUs/CPUs through hourly rewards and staking.

How It Works

  1. Resource Aggregation:
    • GPU/CPU providers register devices on io.net, which are verified and pooled into a global network.
  2. Task Allocation:
    • Users submit compute jobs (e.g., AI training), and the platform dynamically assigns tasks to optimal hardware using smart scheduling algorithms.
  3. Execution & Payment:
    • Jobs run on distributed nodes, with results validated via test cases. Payments in $IO/$IOSD are automated via blockchain.

Staking

  • Staking Requirements:
    • Minimum stake per GPU: 200 $IO. High-end GPUs (e.g., NVIDIA H100) require 10x multiplier (16,000 $IO for 8 H100s).
    • 14-day cooldown period for unstaking.
  • Rewards:
    • Block Rewards: Distributed hourly—80% to GPU providers, 20% to CPU providers.
    • Slashing: Malicious actors lose staked $IO; slashed tokens are burned after a 1-month appeal window.

Documentation

For technical guides, tokenomics, and API integration:

Additional Resources

  • Official Links:
  • Community:
  • Partnerships:
    • Collaborations with Aptos Labs, Aethir, and Mind Network enhance AI security and cross-chain interoperability.

Ecosystem Components

Product Functionality
IO Cloud Manages decentralized GPU clusters for AI/ML deployments.
IO Worker Dashboard for providers to monitor hardware performance and earnings.
IO Explorer Analytics tool tracking network activity, rewards, and resource utilization.

Tokenomics

  • Initial Supply: 500M $IO (20M allocated to Binance Launchpool).
  • Max Supply: 800M $IO over 20 years, with deflationary rewards decreasing 1.02% monthly.
  • Distribution:
    • 50% to community, 20% to investors, 20% to core contributors, 10% ecosystem fund.