NuNet Development Update: July 2022
Greetings Singularitarians,
The Nunet developers have made significant progress this month, particularly on the use cases for Fund 7 and 8 from Cardano’s Catalyst rounds, as we ramp up to complete these obligations that will pave the way for release of the NuNet Platform Public Alpha later this year. This article will detail each milestone met in these categories.
Decentralized SPO Computing (Fund7)
Development of a new program for managing devices onboarded to the NuNet platform is ongoing. This program is a replacement for the bash script used for onboarding devices to the NuNet platform and for the NuNet adapter that took care of networking, monitoring and deployment of services. Onboarding with the new device management service can handle the SPO use case.
Management of Firecracker virtual machines that house Cardano nodes is ongoing. This is necessary to allow the device management service to be able to deploy and manage firecracker virtual machines for the Decentralized SPO use case. Experiments and demo of Cardano node running inside Firecracker VM is done.
Research and development of pipeline for deployment of Cardano nodes is ongoing. This is the main component that is going to be responsible for the deployment of Cardano nodes on the NuNet platform, securing virtual machines, allowing remote access to the instances for SPOs and monitoring the state of instances.
Decentralized GPU ML Cloud (Fund8)
We’ve been successfully able to build a prototype of a GPU enabled machine learning platform, available on a web UI based command line and accessible from anywhere across the globe. The current prototype includes the machine learning frameworks: TensorFlow and PyTorch made readily available on newly onboarded systems. In a future update, we will be adding an offboard/onboard feature through the command line to be able to give more control to device providers. It would allow them to offboard their devices without having to uninstall the complete setup from their systems. When they would prefer to onboard again, they could simply turn it back on.
References:
- Run an application on NuNet that uses GPU resources.
- Device owners can choose to onboard/offboard a machine from the device management app.
- Compute provider should be able to turn on and off device’s availability to the platform.
While we work on that on the Device Management App, we’ve added a disconnect/reconnect feature based on reference 3 above. It allows compute providers to disconnect their devices without having to uninstall the complete setup from their systems. It is equivalent to pausing device resource usage on NuNet. It can be done through the following command:
sudo bash onboarding.sh pause
When they’d prefer to reconnect again, they could simply unpause it back online:
sudo bash onboarding.sh unpause
An additional benefit due to this feature is that disconnecting a machine (pause) from NuNet provides a very convenient way to plug & play with hardware for device providers. For example, after turning off the machine, we can remove a GPU from the system, assemble another one in its place, and power it back on. So, after booting again, we get some time to confirm whether the new GPU is locally seen by the machine without any issues (with a command like nvidia-smi). Once that is done, reconnecting the machine (unpause) to NuNet would show the new GPU as onboarded!
Reference: Device owners can choose to pause/unpause onboarding on a machine from the command line.
NuNet Is Hiring!
NuNet currently has a number of open positions for various roles within the team. If you have the skills and desire to join us in our journey, you can find more information and contact us through our jobs page.
About NuNet
NuNet lets anyone share and monetize their computing resources, turning cloud computing power from a centralized service into an open protocol powered by blockchain. Find out more via:
