#benchmark

nightly bin+lib pew

A benchmarking library for Rust based on google/benchmark

7 releases

Uses old Rust 2015

0.2.3 Apr 29, 2018
0.2.2 Apr 25, 2018
0.1.2 Apr 17, 2018

#94 in Profiling

Apache-2.0

27KB
365 lines

Pew

Pew is a benchmarking library written in Rust based of Google's C++ Benchmarking library. It is currently in very alpha stages (I'd consider it an MVP). It was built to be able to do the following (which you cannot do in the rust benchmarking library):

  1. Pause and unpause the benchmark timer
  2. Run multiple benchmarks by specifying a range of arguments
  3. Creating some initial state that gets passed to all runs of the benchmark

The benchmark will run for at least 1 second (or the user specified --min_duration) and at least 8 runs (or the user specified --min_runs). The average of these runs is output as the Time (ns) column.

More comprehensive docs can be found here.

TODO: Output other state, not just the mean.

Installation

This is available on crates.io. You can link it as a library, or cargo install to also get the transpose script (see output).

Add pew as a dependency in your Cargo.toml. Then, add this to your crate root:

#[macro_use]
extern crate pew;

I usually create benchmarks in the bin/ directory and run them with cargo run --bin <benchmark-name> --release.

Usage

View the docs here.

Example

This can be used as follows:

fn main() {
    Benchmark::with_name("range_bench")
        .with_range(1 << 10, 1 << 20, 4)
        .with_generator(generator)
        .with_bench(pew_bench!(bm_vector1))
        .with_bench(pew_bench!(bm_vector2))
        .with_bench(pew_bench!(bm_vector3))
        .run();
}

There are more complete examples in the examples/ directory of how to use this.

Output

The output is a comma separated list of benchmark results (this is what cargo cargo run --example example1 will output):

Name,Time (ns)
range_bench/bm_vector_range/1024,104715
range_bench/bm_vector_range/4096,554838
range_bench/bm_vector_range/16384,2068971
range_bench/bm_vector_range/65536,7739376
range_bench/bm_vector_range/262144,31389948
range_bench/bm_vector_range/1048576,114633815
gen_bench/bm_vector_gen/1024,123643
gen_bench/bm_vector_gen/4096,545581
gen_bench/bm_vector_gen/16384,2590869
gen_bench/bm_vector_gen/65536,7799209
gen_bench/bm_vector_gen/262144,29498657
gen_bench/bm_vector_gen/1048576,113458415

You can also pass a --filter flag to the benchmark which would only run benchmarks who's name contains the filter string. For example, running cargo cargo run --example example1 -- --filter gen will output:

gen_bench/bm_vector_gen/1024,123643
gen_bench/bm_vector_gen/4096,545581
gen_bench/bm_vector_gen/16384,2590869
gen_bench/bm_vector_gen/65536,7799209
gen_bench/bm_vector_gen/262144,29498657
gen_bench/bm_vector_gen/1048576,113458415

while running cargo cargo run --example example1 -- --filter 1024 will output:

range_bench/bm_vector_range/1024,104715
gen_bench/bm_vector_gen/1024,123643

Oftentimes I run multiple benchmarks on the same range and plot the results. I find it is easier to plot the results by transposing the output. There is a binary pew_transpose that does just that. It can be run by piping the output of the benchmark code:

cargo run --example example1 | pew_transpose

The transposed output of the above would be:

Size,bm_vector_range,bm_vector_gen
1024,105974,106845
4096,418835,409143
16384,1646391,1655855
65536,6566094,6668369
262144,27025772,26948818
1048576,108144239,107596626

TODO: Add more output types (e.g. pretty printed, JSON, etc).

Cli

The Cli offers convenience flags and options while running benchmarks:

Akshay Nanavati <akshay.nanavati1@gmail.com>
A benchmarking library for Rust based on google/benchmark

USAGE:
    example1 [OPTIONS]

FLAGS:
    -h, --help       Prints help information
    -V, --version    Prints version information

OPTIONS:
    -f, --filter <FILTER>             Only run benchmarks that contain this string
    -d, --min_duration <RUN_UNTIL>    Run benchmarks till this time (in s) and then output average [default: 1]
    -r, --min_runs <MIN_RUNS>         Run benchmarks for at least this many runs [default: 8]

These can be passed to the main binary that is running your benchmark.

License

This code is licensed under Apache 2.0.

Contributing

This library is in very early stages and I'd love any contributions I can get. This is my first time writing a becnhmarking library (and a library in Rust) and would love input from those who are more experienced.

A good starting point would be the issues (which are also linked here as TODOs).

Dependencies

~3.5MB
~81K SLoC