Skip to main content

Canbench

Intermediate
Tutorial

Overview

Canisters deployed on ICP use resources such as memory and compute. Resources are finite, and there are limits configured that restrict how many resources a canister may consume. The canbench Rust crate provides a way to benchmark a canister's resources so developers can gain an insight into how their canister uses memory, instructions, and other resources.

canbench can be used to provide information such as:

  • Insight into how a canister consumes stable memory, heap memory, and instructions.

  • Detecting possible performance regressions, both locally or on the mainnet.

  • Analyzing performance bottlenecks.

Features of canbench include:

  • Supports benchmarks that are up to 10 trillion instructions.

  • Reports the number of instructions consumed alongside changes to stable and heap memories.

  • Persistence of benchmarking results in your canister's repo.

  • Language agnostic, though initially only support for Rust exists.

Installation

You can install canbench using cargo.

cargo install canbench

Add an optional dependency for canbench in your Cargo.toml file:

canbench_rs = { version = "0.1.1", optional = true }

Then, edit the canbench.yml file to point to your canister's Wasm file:

build_cmd:
cargo build --release --target wasm32-unknown-unknown --features canbench-rs

wasm_path:
./target/wasm32-unknown-unknown/release/<YOUR_CANISTER>.wasm

Example benchmark test

Below is an example canister that uses a query function to compute and return the fibonacci sequence of a given number:

#[ic_cdk::query]
fn fibonacci(n: u32) -> u32 {
if n == 0 {
return 0;
} else if n == 1 {
return 1;
}

let mut a = 0;
let mut b = 1;
let mut result = 0;

for _ in 2..=n {
result = a + b;
a = b;
b = result;
}

result
}

An example benchmark test using canbench for this canister could look like this:

#[cfg(feature = "canbench-rs")]
mod benches {
use super::*;
use canbench_rs::bench;


#[bench]
fn fibonacci_20() {
// NOTE: the result is printed to prevent the compiler from optimizing the call away.
println!("{:?}", fibonacci(20));
}
}

The benchmark output of this test would resemble the following:

canbench

---------------------------------------------------

Benchmark: fibonacci_20 (new)
total:
instructions: 2301 (new)
heap_increase: 0 pages (new)
stable_memory_increase: 0 pages (new)

---------------------------------------------------

Executed 1 of 1 benchmarks.

For more examples and information, view the crate's documentation or the GitHub repo for this crate.