Methodology

Workload Composition

Benchmarks run two workload types at varying target TPS levels (10K, 20K, 40K):

TIP-20100% TIP-20 token transfers (ERC-20 equivalent).
Mixed80% TIP-20 transfers, 20% MPP (multi-party payment) channel operations (approve → open → close).

Hardware

Benchmarks run on dedicated bare-metal servers to ensure consistent results:

CPUAMD EPYC 4585PX (16C/32T, 4.3/5.7 GHz)
Memory128 GB 3600 MHz
Storage2×960 GB NVMe + 2×1.92 TB NVMe via schelk

Node Configuration

Benchmarks run two local validators in consensus on a dedicated bare-metal server with 100 GiB pre-bloated state.

Metrics

ThroughputAverage and peak gas consumed per second across all blocks in the run.
TPSTransactions per second (derived from gas / avg tx gas).
P50/P99 LatencyBlock execution time percentiles.
Block TimeAverage wall-clock time between blocks.

Data Pipeline

Benchmark results are produced by txgen and uploaded to ClickHouse for visualization.