Key Highlights
- Nethermind’s new benchmark runs identical workloads for all Ethereum clients, making performance results stable and repeatable.
- The tool supports two tests: real mainnet blocks and merged “super-blocks” to stress-test clients under heavy loads.
- Nethermind leads in both test types, with performance gaps widening under larger workloads.
Nethermind, a leading Ethereum client developer, has introduced a new benchmarking tool designed to measure Ethereum execution performance with clarity and accuracy. The benchmark, announced on Tuesday, runs the same workloads for every client, ensuring results are stable, reproducible, and easy to compare.
“We introduced a new benchmark that uses identical workloads, making results stable and easy to reproduce,” Nethermind said in a blog post. The tool is expected to help developers test clients like Geth, Reth, and Besu under the same conditions and identify real performance improvements or bottlenecks.
How the benchmark works
The benchmark works by replaying historical mainnet blocks, exactly as they happened on the network, and removes the variability found in live-network testing. By starting from the same snapshot and using the same hardware, Nethermind ensures every client receives identical input.
The test also eliminates the 12-second gaps between blocks, thereby creating continuous workloads that reveal how clients handle garbage collection, state writes, and heavy transaction loads.
Furthermore, there are two types of tests in the benchmark. The first type is real mainnet payloads: This replays historic blocks exactly as they appeared onchain, and provides a look at average performance under normal conditions. Gas usage in these tests averaged 18.3 million per block, with peaks up to 36 million.
The second type, merged mainnet payloads, combines multiple consecutive blocks into a single “super-block.” This mode simulates extreme scenarios with high gas and transaction volumes, and developers identify performance bottlenecks and edge cases. In these tests, gas usage averaged 1.1 billion per block, with a maximum of 2 billion, merging 100 blocks.
Why this matters for developers
Nethermind said the tool is useful not only for mainnet clients but also for Layer 2 teams. They can test network-specific changes without using the real mainnet. Previously, client teams compared two versions by syncing them on the live network, but this often gave inconsistent results due to differences in peers, blockheads, and network timing.
The firm added that the framework can also assist Layer 2 teams, allowing network-specific optimizations and tests without affecting the mainnet.
Also Read: BitMine Adds 102,259 ETH to Treasury, Now Holds 3.2% of Supply
