gas-optimize

Start with storage optimizations: pack variables into fewer slots, cache repeated storage reads in local variables, and use constant or immutable where possible. Then review data structures, external call caching, and unchecked arithmetic for loop counters.

Storage layout changes have the highest impact since SSTORE costs up to 20,000 gas and SLOAD costs 2,100 gas. Variable packing, struct packing, and replacing storage writes with events can save tens of thousands of gas per transaction.

Yes. On L2 rollups, calldata posted to L1 dominates costs while storage operations are relatively cheaper, so minimizing parameter sizes matters more. Standard optimizations still apply, but techniques like events-over-storage have reduced ROI on L2.

Only when a function is on an extremely hot path, savings exceed roughly 500 gas per call, and the logic is simple enough to audit. All assembly recommendations should be flagged for security review since they bypass compiler safety checks.

Low runs (around 200) optimize for cheaper deployment with slightly more expensive execution, while high runs (10,000+) reduce execution cost at the expense of deployment. Match the setting to expected call frequency of the contract.