Glossary

Key terms and concepts used throughout the Dispersed Network documentation.

Network Concepts

Dispersed Network

The Dispersed Network is a distributed network of GPU-equipped computers (Nodes) that perform computational work (Jobs) for users. Node Operators earn RENDER tokens for contributing compute power.

Node

A Node is a computer registered on the Dispersed Network that executes Jobs. Nodes must have NVIDIA GPUs and run the Dispersed client software.

Dispersed Client

The Dispersed Client (or simply dispersed) is the software that Node Operators run to connect their machines to the network. It handles registration, job assignment, and container management.

---

Job Concepts

Job

A Job is a work request submitted by a Service Consumer. Jobs specify hardware requirements (CPU, GPU, RAM, storage) and container configuration (Docker image, parameters).

Job Run

A Job Run is a single execution of a Job on a Node. One Job can have multiple Job Runs (e.g., if retried or distributed).

Job Recipe

A Job Recipe is a pre-configured template for creating Jobs. Recipes define hardware requirements and container settings, allowing users to launch common workloads by filling in minimal inputs (like an SSH key).

BATCH Job

A BATCH Job runs until completion or timeout. It requires a max_timeout_run_ms value and is billed for actual execution time. Use BATCH for rendering, data processing, or any finite task.

PERSISTENT Job

A PERSISTENT Job runs indefinitely until explicitly stopped. It must have max_timeout_run_ms set to null and is billed hourly. Use PERSISTENT for SSH access, development environments, or long-running services.

---

Rewards and Billing

Compute Epoch

A Compute Epoch is a weekly reward cycle running from Sunday 00:00 UTC to the following Sunday 00:00 UTC. Node Operators earn rewards based on their uptime and work completed during each epoch.

Availability Reward

The Availability Reward is RENDER earned for keeping a Node online and ready to accept jobs. Maximum of 6 RENDER per week at 100% uptime.

Job Reward

The Job Reward (or Execution Reward) is RENDER earned for completing computational work. Amount depends on hours worked and the hardware’s Compound Multiplier.

Compound Multiplier

The Compound Multiplier is a factor that adjusts Job Rewards based on hardware performance. An RTX 4090 has a multiplier of 1.0 (baseline). Better hardware earns proportionally more per hour worked.

---

Authentication

API Key

An API Key is a credential for authenticating API requests, consisting of:

• public_key (pk_…) — included in the X-API-Key header
• secret_key (sk_…) — used to generate HMAC signatures

API Keys require HMAC signature verification on each request. See Authenticate Requests for implementation details.

---

Roles

Node Operator

A Node Operator contributes GPU compute power to the network by running the Dispersed client on their machines. They earn RENDER rewards for availability and completed work.

Service Consumer

A Service Consumer submits Jobs to the network to run computational workloads. They pay for compute time based on job type (BATCH or PERSISTENT).

---

Technical Terms

Docker Container

Jobs execute inside Docker Containers — isolated environments that package an application with its dependencies. This ensures consistent execution across different Nodes.

GPU (Graphics Processing Unit)

GPUs are specialized processors optimized for parallel computation. The Dispersed Network currently requires NVIDIA GPUs with CUDA support.

VRAM (Video RAM)

VRAM is dedicated memory on a GPU. Jobs can specify minimum VRAM requirements to ensure adequate GPU memory for their workloads.