Last thread, we promised more on gas benchmarking. Through structured gas benchmarking, we’ve: — Surfaced opcode bottlenecks — Improved 7702 tx performance — Helped some clients cut worst-case execution by 25%+ — Unlocked 45M gas blocks Here’s how it happened 👇

2/ Each client behaves differently under load. So we created shared scenarios, ran gas-heavy transactions, and measured performance at scale, across time, versions, and configurations. It’s not about comparison. It’s about revealing the edges, together.

3/ Along the way, we surfaced issues that weren’t previously visible: — Bottlenecks in opcodes and precompiles — Execution slowdowns in complex tx types — Patterns that hinted at deeper optimizations Many of these led to upstream fixes, quietly, collaboratively.

4/ The impact so far: — Worst-case scenarios improved by ~25% — Notable gains in complex transaction handling — 45M gas block processing unlocked, with a clear path toward even higher limits — Infra teams already benefiting from more stable throughput

6/ Next up: PerfNet, and how it fits into this performance journey.