Compute / processing

Aethir

A verified project description is currently not available.

Date of the profile text: . This date refers to the introduction, not a network observation.

About the project

A verified project description is currently not available.

The network

According to the documentation, resource owners, who own the GPU chips, provide compute resources to the network for use in Containers. The network comprises Containers, Checkers and Indexers, which house the compute resources, provide quality assurance and facilitate matching based on end-user requirements. Compute buyers and gamers consume the resources on demand, for AI training and inference or for gaming. A blockchain serves as the settlement layer and records the transactions.

Source: The network ↗

Taking part

According to the documentation, Cloud Hosts in Aethir Cloud must stake ATH tokens for the entire duration of their GPU’s operation. The stake requirement applies per GPU type, is updated each month and then remains fixed for the following month. If a GPU’s hardware configuration changes, it may be recognized as a new GPU; the old GPU must then be unstaked and the new one staked. Unstaking a machine requires a 180-day unlocking period before the staked tokens can be fully withdrawn.

Source: Taking part ↗

Equipment

According to the documentation, hosts contribute idle GPU resources to the network. They register their bare metal servers to become part of the global compute pool. Customers get access to GPU servers with chips such as NVIDIA A100 and H100.

Source: Equipment ↗

How the data is used

According to the documentation, the Container is where the actual use of the cloud takes place: a virtual endpoint that executes and renders the application, for example the game for the player. Containers are selected based on quality of service, latency and cost. Once set up, a Container waits for a Checker to check it, and once the Checker has approved it, it can be used in the network. A Container’s performance is regularly checked against the network’s standards.

Source: How the data is used ↗

Rewards and rules

According to the documentation, new node operators must stake ATH before they can contribute resources to a Container. Containers are rewarded for maintaining a state of high readiness and providing standby services, for example with the PoC reward. They receive additional rewards for compute resources actively used by the end user, for example PoD and Service Fee rewards. A node operator that deviates from quality control standards or disrupts the network risks having its stake slashed.

Source: Rewards and rules ↗

Related public introductions

Same network category; not a quality comparison or endorsement.

  • Acurast — Compute / processing
  • Phala — Compute / processing

Sign in to explore available evidence

Available evidence and CORE questions depend on your account access. An account does not guarantee an answer to every question.

Sign in Create a free account