For Prover Nodes

Quick overview of what it takes to get onboarded as an Operator into Fermah AVS.

Recommended: Follow the quickstart guide Installation and go through the onboarding using the script.

For complete manual onboarding, you may follow these steps to get started:

1

Installation

Download one of our releases and get your environment setup.

2

Configuration

Customize your installation based on your hardware and provers.

3

Machine Secret

Setup proper authentication for your prover node and get whitelisted.

4

Registration

Register your Prover Node with Fermah Core.

5

Telemetry

Prepare your node for proving metrics collection.

6

Starting a Node

Launch the Prover Node 🚀

These steps will get a Prover Node fully onboarded into Fermah Testnet.

---

Prerequisites

Registered Prover Node Accounts: Ensure you have a fully registered Prover Node account, such as with EigenLayer. If you don't have one, follow the steps in the EigenLayer Operator Guide to create and fund your account.

You will also need the following:

• Ubuntu 22.04 LTS (GLIBC 2.34+)

• Docker v24+ (for running telemetry and future provers)

• CUDA drivers v12.2 (see NVIDIA docs)

• CUDA containers toolkit (see NVIDIA docs)

• Sepolia Ethereum Node RPC URL access with your API key (e.g. Alchemy, Infura or private)

• Grafana API key (provided by Fermah)

Hardware Requirements

Resources used by the Operator will vary greatly depending on the prover image that is running.

ZkSync

Minimum requirements for a ZkSync CPU witness generator.

Round	RAM	vCPUs
all	128 GB	32
basic_circuits	32 GB	16
leaf_aggregation	80 GB	16
node_aggregation	4 GB	8
recursion_tip	1 GB	4
scheduler	1 GB	4

Suggested Cloud Instance Types

AWS g6.8xlarge GCP g2-standard-32

Minimum requirements for a ZkSync circuit prover.

vCPUs	RAM	VRAM	Storage
32	128 GB	24 GB	375 GB SSD

Suggested Cloud Instance Types

AWS g6.8xlarge GCP g2-standard-32