Use Idle Compute Cycles To Generate Income đź’˛

by sugardave

Published:

You need more OGPU

You can generate income with your idle CPU/GPU cycles

Tags:

crypto

how-to

OpenGPU

passive income

Render AI and Bittensor TAO are two of the more well-known projects promising the ability to use your idle compute cycles to generate an income stream. With this post, I’d like to talk about a lesser-known project: OpenGPU.

OpenGPU has developed an EVM-compatible L1 blockchain with blazing fast execution (~10K TPS) and low transaction fees specifically tailored for rapid task execution as block transactions. In other words, they are laying the groundwork for an infinitely scalable distributed task execution architecture.

The chain, of course, has the basics one would expect:

However, the most exciting bits (for me) are the Client dApp and Provider Suite. The Client dApp allows anyone with a containerized task workflow to define it as a model to be executed by anyone running the Provider Suite.

Clients

Clients are generally considered to be anyone that has a need for computational cycles, such as AI inferencing, machine learning, computer vision, etc.

Clients can configure their model(s) to control which providers will be considered, how much the “winning” provider will be paid per task, the length of time allowed to complete a given task, etc. For more detailed information, check out the Client dApp’s model submission page.

Providers

Providers can be anyone that has a supported GPU (NVIDIA, AMD, even Apple GPUS!) and has installed and configured the Provider Suite (Windows, Linux, and Apple!). For providers that aren’t necessarily deep on the technology front, the one-click nature of the installation procedure (Linux might require some futzing depending on your kernel/distribution) is a major plus.

Important Facts For Providers

To be a successful provider, there are a few things to note prior to starting out.

System Requirements

I consider the system requirements to be surprisingly low. At a minimum you are going to want 32 GB RAM, a GPU with 8 GB VRAM (though some providers claim they are successful with much lower VRAM GPUS - as low as 3 GB, anecdotally), and 10 GB free SSD/HDD capacity.

For reference, I am running on an MSI gaming laptop with 32 GB RAM, a 10th-generation Core i9 CPU, and an external RTX 3070 Ti hooked up via Thunderbolt 3 enclosure.

Windows providers will need to ensure that their system is capable of running Windows Subsystem for Linux (WSL). This typically requires certain features enabled in the system BIOS and Windows features that may not already be installed. For all you could ever want to know about WSL, see the Microsoft documentation.

Wallet Addresses

You will need two EVM-compatible wallet addresses (i.e. MetaMask); the “master wallet” (which I will refer to as the “management wallet”) and a “provider wallet”.

Provider Wallet

During the installation you will be asked to create a new or import an existing provider wallet. This wallet address will be used to submit responses to tasks for which you have registered in the management console (more on that later).

Management Wallet

As part of the installation procedure, the provider wallet you specified/created will be linked to a management wallet.

The management wallet will be used to fund the provider wallet with at least one OGPU (the native currency of the OpenGPU chain) and is also the source of funds for depositing/locking OGPU in the vault (more on that later) and withdrawing rewards from the vault.

If you have already been using a wallet to interact with the OpenGPU ecosystem (staking, for example), you would typically use that same wallet as the management wallet. However, this is not a requirement.

Provider Announcement

In order to participate in the system (which you will want to do if you came this far, right?!), you need to “announce” your provider address. This can be done using the management dashboard.

Connect your management wallet to the dashboard and you should see an “Add Provider” button near the bottom of the sidebar menu. Click that and enter your provider wallet address. As part of this process, you will need to send at least 1 OGPU from the management wallet to the provider wallet. The reason for this is that, in order to submit task responses, the provider wallet will need to pay the transaction fees.

It may take a few minutes for the announcement to be completed.

Register For Sources

Once the provider announcement is complete, you can head to the Sources page to see what is available. Each has a “View” button to show more detailed information about that source, including what amount (if any) of OGPU you need locked in the Vault to be able to register for it.

Once you’ve identified which source(s) you want to provide for, you can click the “Register” button. There is a small fee for registration that will be paid from the management wallet.

The Good Part

Once you’re up and running with a provider and registered for sources you are ready to earn OGPU for completing tasks.

Who Gets Paid

The first provider to respond with a valid task completion submission will receive the reward for that task. Admittedly, this is biased to those with superior setups. However, there are a couple of mitigating factors to consider if you’re concerned that “smaller fish” might not get rewarded.

Node Validation

There is a source provided by OpenGPU named OpenGPU Activity Rewarder that runs very often and will distribute rewards to providers who are online. The amount of payment depends on the amount of OGPU you have locked in the vault. The more locked in the vault, the larger the reward.

I compare this source to traditional node validation where all participants can receive a share of rewards just by validating blocks. It’s not exactly the same, but close enough for a high-level description.

Number Of Clients

At the time of writing this post, there are approximately zero (0, zip, zilch, nada) clients using the network. As clients are onboarded, the number of sources will increase. Some providers will not register for all sources (due to system constraints, personal preferences, etc.) so this can represent an opportunity for smaller setups to earn rewards.

AI Agents

A promised feature is the ability to use/create AI agents that will match your provider to optimum task requests.

Wrapping Up

Do you like earning passive income? Do you have spare computing cycles? If so, then the OpenGPU Provider Suite might be just what you’re looking for!