crossbeam-arccell

Crates.iocrossbeam-arccell
lib.rscrossbeam-arccell
version0.6.4
sourcesrc
created_at2018-08-07 04:18:34.181365
updated_at2018-08-07 04:28:45.161372
descriptionAtomically updatable Arc
homepagehttps://github.com/k3d3/crossbeam-arccell
repositoryhttps://github.com/k3d3/crossbeam-arccell
max_upload_size
id77967
size29,258
Keith (k3d3)

documentation

https://docs.rs/crossbeam-arccell

README

Crossbeam-ArcCell

crates.io crossbeam-arccell docs

Crossbeam-ArcCell is an updatable Arc implementation using crossbeam-epoch for memory reclamation. It is meant to be as fast and consistent as possible for load speed, at the expense of having inconsistent-timed and potentially very slow writes.

The idea behind this data structure is that we have an atomic pointer that is always pointing to valid data. This data should always be an atomic pointer dereference (and a pin) away.

When the ArcCell needs to be updated, the old value cannot be modified because it might be in use by other threads. Instead, the existing data must be cloned to a new location and modified there. Once a new value is ready, a compare-and-swap is performed on the atomic pointer, so all threads requesting data after that point will receive the newly-updated data. Eventually, the old memory will be reclaimed.

Loads should always be constant-time, even in the face of both load and update contention.

Updates might take a long time, and the closure passed to it might run multiple times. This is because if the "old" value is updated before the closure finishes, the closure might overwrite up-to-date data and must be run again with said new data passed in. Additionally, memory reclamation of old ArcCell values is performed at this point, unless the _no_reclaim() methods are used. If these methods are used, be sure to use the reclaim() method to prevent a memory leak.

If you want to set the ArcCell data regardless of the current value, you can use the set() method. This call should be a bit quicker than update.

Example

extern crate crossbeam_arccell;

use crossbeam_arccell::ArcCell;

// Create a new ArcCell with a Vec of numbers
let arc = ArcCell::new(vec![1,2,3,4]);

// Read from the ArcCell
{
    let data = arc.load();
    println!("Current ArcCell value: {:?}", data);
}

// Update the ArcCell to add a new number
arc.update(|old| {
    let mut new = old.clone();
    new.push(5);
    new
});

// Read the new data
{
    let data = arc.load();
    println!("Current ArcCell value: {:?}", data);
}

// Set the ArcCell value
let data = vec![9,8,7,6];
arc.set(data);

// Read the new data, again
{
    let data = arc.load();
    println!("Current ArcCell value: {:?}", data);
}

Benchmarks

Note that these benchmarks exist without any contention.

Under contention:

  • Crossbeam-ArcCell's load will always be constant-time.
  • Crossbeam-ArcCell's update will slow down if multiple threads attempt to write at the same time.

The following benchmarks are available under the benches/ directory:

// Crossbeam-ArcCell
test cb_arc_load              ... bench:           9 ns/iter (+/- 0)
test cb_arc_update            ... bench:         456 ns/iter (+/- 4)
test cb_arc_update_no_reclaim ... bench:          74 ns/iter (+/- 0)
test cb_arc_set               ... bench:         444 ns/iter (+/- 6)
test cb_arc_set_no_reclaim    ... bench:          64 ns/iter (+/- 0)

// Arc in stdlib
test arc_load                 ... bench:          11 ns/iter (+/- 0)

// RwLock in stdlib
test rwlock_load              ... bench:          26 ns/iter (+/- 0)
test rwlock_update            ... bench:          25 ns/iter (+/- 0)

// RwLock in parking_lot
test pl_rwlock_load           ... bench:          16 ns/iter (+/- 1)
test pl_rwlock_update         ... bench:           9 ns/iter (+/- 0)

License

Licensed under the terms of the MIT license and the Apache License (Version 2.0)

See LICENSE-MIT and LICENSE-APACHE for details.

Contribution

Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in the work by you, as defined in the Apache-2.0 license, shall be dual licensed as above, without any additional terms or conditions.

Commit count: 41

cargo fmt