Measuring Gas
Profiling tools.
Measuring Gas is a free Web3 & DApp Development Fundamentals lesson on CoddyKit — lesson 4 of 4. You can read the complete lesson below for free — then practise it hands-on in the browser with a built-in code editor and a 24/7 AI tutor. It is part of the Web3 & DApp Development Fundamentals learning path, one of 4 lessons in the course, and your progress syncs across the web and the CoddyKit app.
Measure, Don't Guess
Gas optimization without measurement is guesswork. The compiler, opcode costs, and access patterns interact in non-obvious ways, so an intuitive change can sometimes make things worse.
Always profile before and after. A change that does not reduce measured gas is not an optimization, no matter how clever it looks.
Forge Gas Reports
Foundry's forge test --gas-report prints a per-function table of min, average, median, and max gas usage across all your tests. It is the fastest way to spot expensive functions.
$ forge test --gas-report
| Function | min | avg | median | max | # calls |
| transfer | 28012 | 34521 | 34521 | 51012 | 12 |
| approve | 24300 | 24300 | 24300 | 24300 | 5 |Gas Snapshots
forge snapshot writes the gas cost of every test to a .gas-snapshot file. Commit it to version control, then run forge snapshot --diff to see exactly how a change affected gas.
This turns gas regressions into reviewable diffs in pull requests.
$ forge snapshot
$ forge snapshot --diff
testTransfer() (gas: -1240 (-3.6%))
testMint() (gas: +85 (+0.2%))Measuring in Tests
For fine-grained measurement, read gasleft() before and after a call inside a test. The difference is the gas consumed by that block of code.
uint256 before = gasleft();
target.doWork();
uint256 used = before - gasleft();
emit log_named_uint("gas used", used);Verbose Traces
Running tests with -vvvv shows a full execution trace including the gas cost of each external call and opcode group. This reveals where gas is spent, not just the total.
$ forge test --match-test testTransfer -vvvv
[34521] Token::transfer(bob, 100)
emit Transfer(from: alice, to: bob, value: 100)
return: trueThe Optimizer Setting
The Solidity optimizer changes runtime and deployment gas. The runs parameter is a hint about how often functions are called.
- Low
runs(e.g. 1) optimizes for cheap deployment - High
runs(e.g. 10000) optimizes for cheap repeated calls
Set it in foundry.toml and always measure with the optimizer enabled, since it reflects production behavior.
# foundry.toml
[profile.default]
optimizer = true
optimizer_runs = 10000Comparing Implementations
To compare two approaches, write a test that exercises each under identical conditions and inspect the gas report side by side.
Keep inputs the same and isolate the change to one variable, otherwise you cannot attribute the gas difference to the optimization you made.
function testNaive() public { c.sumNaive(data); }
function testOptimized() public { c.sumOptimized(data); }
// compare both rows in --gas-reportBeware Warm vs Cold Skew
Gas reports can mislead because the first access in a test is cold and later ones are warm. A function called once in setup may show a higher cold cost than the same function later.
Look at min, median, and max columns together rather than a single average, and design tests that reflect real-world access patterns.
Etherscan and Real Receipts
On-chain transaction receipts report the actual gasUsed. Etherscan and block explorers display this for every transaction, letting you validate your local measurements against mainnet reality.
The eth_estimateGas RPC method gives a pre-flight estimate, though it adds a safety margin and may differ from the real execution cost.
CI Gas Regression Gates
Wire forge snapshot --check into continuous integration. It fails the build if any test's gas exceeds the committed snapshot beyond a tolerance.
This prevents accidental gas regressions from slipping into production and makes gas a first-class quality metric.
$ forge snapshot --check
# exits non-zero if gas increased vs committed snapshotA Measurement Workflow
A solid gas workflow:
- Write tests covering the hot paths
- Record a baseline with
forge snapshot - Make one optimization at a time
- Run
forge snapshot --diffto confirm the gain - Gate regressions in CI with
--check
Discipline beats cleverness.
Quick Check
Which Foundry command produces a per-function table of min, average, and max gas usage?
Recap
You learned to measure gas:
forge test --gas-reportfor per-function tablesforge snapshot --diffto track changesgasleft()for fine-grained measurement- Verbose traces reveal where gas goes
- Gate regressions in CI with
--check
This completes the Gas Optimization course. Next: the Foundry toolkit in depth.
Frequently asked questions
Is the “Measuring Gas” lesson free?
Yes — the full text of “Measuring Gas” is free to read here on the web, and the Web3 & DApp Development Fundamentals course includes 4 lessons in total. To practise it interactively (a built-in code editor and a 24/7 AI tutor) and unlock the rest of the Web3 & DApp Development Fundamentals course, upgrade to CoddyKit PRO.
What will I learn in “Measuring Gas”?
Profiling tools. You practise Web3 & DApp Development Fundamentals with hands-on code you run directly in the browser, and a 24/7 AI tutor answers your questions as you work through the lesson.
Do I need any experience to start Web3 & DApp Development Fundamentals?
No prior experience is required. Web3 & DApp Development Fundamentals on CoddyKit is structured for beginners through advanced learners; this is — lesson 4 of 4, so you can start here or from the beginning and move at your own pace.
How long does the “Measuring Gas” lesson take?
Most CoddyKit lessons take about 5–10 minutes. Each one is bite-sized and interactive, so you make steady progress and pick up exactly where you left off across the web and the app.
Can I write and run code in this Web3 & DApp Development Fundamentals lesson?
Yes. Every Web3 & DApp Development Fundamentals lesson includes a built-in code editor, so you write and run real code right in your browser and get instant AI feedback — no local setup required.