Module rayon

Expand description

Data parallelism library for Rust.

rayon is a data-parallelism library for Rust that makes it easy to convert sequential computations into parallel ones with minimal changes to existing code.

The crate provides parallel versions of common iterator methods through the ParallelIterator trait, allowing you to simply change .iter() to .par_iter() to parallelize operations. Rayon uses work-stealing to efficiently distribute computations across CPU cores.

Key features include parallel iteration with par_iter, parallel collection operations like map, filter, and reduce, parallel sorting with par_sort, and parallel searching. The library also provides join for fork-join parallelism and scope for structured parallelism with lifetimes.

Rayon automatically manages thread pools and work distribution, making parallelism accessible without manual thread management. It’s particularly effective for CPU-bound tasks that can be decomposed into independent units.

Rayon’s global ThreadPool is the recommended threadpool implementation in rustmax for general-purpose parallel computation. It provides excellent work-stealing performance and integrates seamlessly with rayon’s parallel iterators.

§Examples

Basic parallel iteration:

use rayon::prelude::*;

let data = vec![1, 2, 3, 4, 5, 6, 7, 8];
let sum: i32 = data.par_iter().map(|&x| x * x).sum();

assert_eq!(sum, 204); // 1 + 4 + 9 + 16 + 25 + 36 + 49 + 64

Parallel filtering and collection:

use rayon::prelude::*;

let numbers = (1..100).collect::<Vec<_>>();
let evens: Vec<_> = numbers
    .par_iter()
    .filter(|&&n| n % 2 == 0)
    .map(|&n| n * 2)
    .collect();

assert_eq!(evens[0], 4);  // 2 * 2
assert_eq!(evens[1], 8);  // 4 * 2

Parallel reduction with custom operation:

use rayon::prelude::*;

let data = vec![1, 5, 3, 9, 2, 8, 4];
let max = data.par_iter().cloned().reduce(|| 0, |a, b| a.max(b));

assert_eq!(max, 9);

Using rayon’s threadpool for custom parallel work:

use rayon::ThreadPoolBuilder;

let pool = ThreadPoolBuilder::new().num_threads(4).build().unwrap();

let result = pool.install(|| {
    // This closure runs on the custom threadpool
    (0..1000).into_iter().map(|i| i * i).sum::<i32>()
});

assert_eq!(result, 332833500);

Modules§

array: Parallel iterator types for arrays ([T; N])
collections: Parallel iterator types for standard collections
iter: Traits for writing parallel programs using an iterator-style interface
option: Parallel iterator types for options
prelude: The rayon prelude imports the various ParallelIterator traits. The intention is that one can include use rayon::prelude::* and have easy access to the various traits and methods you will need.
range: Parallel iterator types for ranges, the type for values created by a..b expressions
range_inclusive: Parallel iterator types for inclusive ranges, the type for values created by a..=b expressions
result: Parallel iterator types for results
slice: Parallel iterator types for slices
str: Parallel iterator types for strings
string: This module contains the parallel iterator types for owned strings (String). You will rarely need to interact with it directly unless you have need to name one of the iterator types.
vec: Parallel iterator types for vectors (Vec<T>)

Structs§

BroadcastContext: Provides context to a closure called by broadcast.
FnContext: Provides the calling context to a closure called by join_context.
Scope: Represents a fork-join scope which can be used to spawn any number of tasks. See scope() for more information.
ScopeFifo: Represents a fork-join scope which can be used to spawn any number of tasks. Those spawned from the same thread are prioritized in relative FIFO order. See scope_fifo() for more information.
ThreadBuilder: Thread builder used for customization via ThreadPoolBuilder::spawn_handler.
ThreadPool: Represents a user created thread-pool.
ThreadPoolBuildError: Error when initializing a thread pool.
ThreadPoolBuilder: Used to create a new ThreadPool or to configure the global rayon thread pool.

Enums§

Yield: Result of yield_now() or yield_local().

Functions§