This is the description of the current NuNet architecture for testing the ML use case. The business goal of implementing this use case is to allow users to onboard both latent GPU and CPU resources to be used by service providers to run compute jobs (in this case, ML jobs) on the NuNet platform and be compensated in NTX (NuNet’s Utility Token) for the job. For Public Alpha we are using the PreProd Cardano Network.
NuNet architecture for Public Alpha
These are the NuNet core elements for the Public Alpha:
- In NuNet, the term service provider refers to the individual or group that wants to run a job on NuNet’s decentralized community hardware.
- The term compute provider refers to an individual on the community who has on-boarded their devices onto the NuNet platform.
- The term DMS refers to Device Management Service, which is essentially the NuNet platform itself. It is the lightweight peer-to-peer connection between services and providers.
- Service Provider Dashboard is the application that service providers use to deploy jobs on NuNet and to lock funds on the smart contract.
- Compute Provider Dashboard is the application that compute providers use to claim their tokens for work carried out on their devices.
- NuNet Oracle and Haskell Server are components responsible for locking funds on the smart contract and validating results of compute jobs prior to the withdraw transaction. For Public Alpha we implemented a smart contract on the PreProd Cardano Network.
- Control Server is responsible to any control functionalities necessary on NuNet. On this use case, it servers the captcha functionality.
- NuNet Network Status displays real-time statistics about all computational processes executed on the network and their telemetry information.