Golem Network has taken a significant step toward supporting Ethereum’s decentralized infrastructure by launching a series of solo staking tests. These initiatives aim to strengthen the network while exploring scalable, secure, and cost-effective methods for staking ETH. After successfully completing multiple testing phases on both testnet and mainnet environments, we’re sharing a comprehensive overview of our findings, technical setups, performance benchmarks, and future plans.
Objectives of the Staking Tests
The primary goals behind Golem’s staking tests were multifaceted:
- Establish a secure and scalable method for depositing ETH into Ethereum validators
- Evaluate and optimize hardware and software configurations for maximum stability
- Proactively identify and mitigate risks associated with Ethereum staking
- Promote client diversity across the Ethereum ecosystem
By focusing on these pillars, Golem aims to contribute meaningfully to Ethereum’s long-term resilience and decentralization.
👉 Discover how decentralized networks are reshaping blockchain infrastructure
Securing Mainnet Staking Operations
Given the substantial amount of ETH involved in mainnet testing, operational security was paramount. To minimize exposure to spam transactions and external interference, we implemented several safeguards:
- Isolated staking pools with distinct deposit and withdrawal addresses
- A hybrid control model using multisig contracts and standard Ethereum addresses
- Transaction routing through centralized exchanges between Golem’s multisig wallet and deposit addresses
These measures ensured a controlled, auditable environment throughout the testing lifecycle.
Key Preparations
To facilitate large-scale ETH deposits, we adopted a modified version of the StakeFish BatchDeposit smart contract. This solution underwent internal auditing and was deployed with minor customizations to suit our workflow.
We also integrated the Gnosis Multisig Wallet to manage staking rewards from select withdrawal addresses—ensuring secure handling of yield.
Additionally, we conducted comparative evaluations of major Ethereum clients:
Execution Layer Clients:
- Geth
- Nethermind
- Erigon
Consensus Layer Clients:
- Lighthouse
- Nimbus
Client selection prioritized network diversity, aligning with Ethereum’s core principles of decentralization and fault tolerance.
Staking Test Phases (Chronological Overview)
Phase 1: Holesky Testnet
We began with the Holesky Testnet to validate our staking architecture in a low-risk environment. Using our customized BatchDeposit contract, we deployed a multisig wallet and initiated staking across multiple validators.
- Deposit Address:
0x899d2c70240a997394e55c95dd0a294d04db483d - Withdrawal Address:
0xac7571242d095abea8f463da7b7b4351271346b6
Validators successfully recognized deposits and actively participated in consensus. After several days of stable operation, we advanced to mainnet testing.
Phase 2: Mainnet “Sanity” Test (5 Validators)
Before scaling up, we executed a small-scale validation on Ethereum Mainnet using five validators:
- Deposit Address:
0x31fa755b6193da3Dfb53384a0C065A9bC268e0a3 - Withdrawal Address:
0x47d412E1b40cc3C73d05F4c070723d927971D6e3
This phase confirmed the integrity of our configuration, paving the way for larger deployments.
Phase 3: Scaling Validator Count on Mainnet
Encouraged by initial success, we expanded operations across five distinct staking pools:
Staking Pool A – Large-Scale Performance Testing
- Validators: 2,000
- Focus: Infrastructure scalability under high load
- Tested hardware/software scenarios including key migration and node expansion with minimal downtime
Staking Pool B & C – Software Configuration Analysis
- Each hosts 500 validators
- Dedicated to evaluating how different software setups affect performance and reliability
Staking Pool D – Golem Ecosystem Fund
- Validators: 1,250
- Publicly visible transactions reinforce transparency
- A significant portion of generated yield is allocated to the Golem Ecosystem Fund, supporting community-driven development
Staking Pool E – Raspberry Pi 5 Feasibility Study
- Validators: 200
- Deposit and withdrawal addresses will be disclosed post-testing
- Inspired by the Web3Pi project, this experiment explores low-cost, accessible staking solutions
👉 Explore how affordable hardware is enabling broader participation in blockchain networks
Raspberry Pi 5: A Viable Solo Staking Platform?
One of the most surprising outcomes was the exceptional performance of the Raspberry Pi 5 in a full Ethereum staking role.
Test Environment Setup
Our isolated network featured:
- Staking Node (Pi 5): Running Geth v1.14.7 and Nimbus v24.7.0
- Helper Node (Pi 5): Running Geth + Lighthouse for attestation gap monitoring
- Metrics Server (Pi 4): Grafana + InfluxDB for real-time system monitoring
- Router (Compute Module 4): OpenWRT-powered gateway with dual Gigabit Ethernet
- Switch: TP-Link TL-SG1005D (5-port, 1Gbps)
All devices operated within a single subnet connected via a symmetric 1Gbps internet link.
Hardware Specifications
Raspberry Pi 5 (Staking Node):
- Overclocked to 3.0 GHz
- PCIe Gen3 interface
- Samsung 980 PRO 2TB NVMe SSD (via Argon NEO 5 case)
- Boot from SSD
- OS: Ubuntu 23.10
- Clients: Geth + Nimbus
Router (CM4):
- 8GB RAM, WiFi, 32GB eMMC
- Operated well below memory limits; suggests even lower-spec models (1GB RAM, no WiFi) could suffice
Performance Benchmarks
Tests were conducted on Ethereum Mainnet with MEV-Boost enabled:
| Validator Count | Attestation Performance |
|---|---|
| 60 | 99.94% |
| 75 | 99.98% |
| 100 | 99.99% |
Encouraged by results, we scaled to 200 validators—running continuously for over 10 days without missed attestations or block production failures.
Downtime & Recovery Testing
We simulated a 5-hour internet outage while running 150 validators:
- Upon reconnection, Geth resumed syncing and regenerated snapshots within ~4 hours
- Attestations continued even during catch-up phases
- Despite degraded thermal pads causing CPU throttling (peaking at 300%) and temperatures reaching 80°C, no attestations were missed
With proper cooling:
- CPU temperature stayed under 70°C even under 400% load during sync
- SSD remained below 50°C in all conditions
Network & Resource Usage
Peak observed bandwidth:
- Inbound: 110 Mibit/s
- Outbound: 103 Mibit/s
- Active connections: ~1,183 UDP, ~634 TCP
System metrics confirmed stable CPU, memory, and storage utilization throughout extended operations.
Core Insights on Raspberry Pi Staking
Our findings echo Adrian Sutton’s observation: “The first 64 validators are the most resource-intensive—beyond that, scaling becomes increasingly efficient.” Once initial synchronization overhead is overcome, adding more validators incurs minimal incremental cost.
While we’ve paused further scaling at 200 validators, evidence suggests the practical limit is likely much higher.
Conclusion
Golem’s staking initiative highlights how accessible hardware like the Raspberry Pi 5 can play a meaningful role in securing Ethereum. The results demonstrate that affordable, energy-efficient devices can reliably participate in consensus, promoting greater decentralization.
We’ve identified gaps in accessible knowledge around Ethereum staking—particularly regarding risks, configurations, and hardware requirements. To address this, Golem plans to fund an external project to develop the “Whitebook of Staking”—a comprehensive guide covering best practices, setup tutorials, and risk assessments.
Contributors interested in shaping this resource are invited to apply through the Golem Ecosystem Fund.
👉 Learn how you can get started with blockchain staking today
We extend our gratitude to the Web3Pi team for their technical support and inspiration.
Stay tuned for further updates on our staking progress and the Whitebook initiative.
Frequently Asked Questions (FAQ)
Q: Why did Golem decide to stake ETH?
A: Staking supports Ethereum’s security and decentralization while generating yield that can be reinvested into the Golem ecosystem through the Ecosystem Fund.
Q: Can a Raspberry Pi really handle 200 validators?
A: Yes—our tests show that a well-configured Raspberry Pi 5 with NVMe storage and adequate cooling can maintain near-perfect attestation rates even under heavy load.
Q: What are the benefits of client diversity in Ethereum staking?
A: Diverse client usage reduces systemic risk—if one client has a bug or vulnerability, others can keep the network running smoothly.
Q: Is solo staking profitable on low-cost hardware?
A: Absolutely. While individual rewards depend on ETH price and network conditions, Raspberry Pi-based setups offer low power consumption and high efficiency—making them cost-effective long-term.
Q: What is MEV-Boost, and was it used in these tests?
A: MEV-Boost allows validators to earn additional revenue by including profitable transactions. It was enabled during our Raspberry Pi tests using builder networks.
Q: Will Golem continue expanding its staking operations?
A: Yes—ongoing infrastructure improvements and research into scalable, secure staking methods remain a priority.