orx-concurrent-bag

An efficient, convenient and lightweight grow-only concurrent collection, ideal for collecting results concurrently.

convenient: the bag can be shared among threads simply as a shared reference, not even requiring Arc,
efficient: for collecting results concurrently:
- rayon is significantly faster when the elements are small and there is an extreme load (no work at all among push calls);
- ConcurrentBag is starts to perform faster as elements or the computation in between push calls get larger (see E. Benchmarks for the experiments).
lightweight: due to the simplistic approach taken, it enables concurrent programs with smaller binary sizes.

The bag preserves the order of elements with respect to the order the push method is called.

Examples

Safety guarantees to push to the bag with an immutable reference makes it easy to share the bag among threads.

Using `std::sync::Arc`

We can share our bag among threads using Arc and collect results concurrently.

use orx_concurrent_bag::*;
use std::{sync::Arc, thread};

let (num_threads, num_items_per_thread) = (4, 8);

let bag = Arc::new(ConcurrentBag::new());
let mut thread_vec: Vec<thread::JoinHandle<()>> = Vec::new();

for i in 0..num_threads {
    let bag = bag.clone();
    thread_vec.push(thread::spawn(move || {
        for j in 0..num_items_per_thread {
            // concurrently collect results simply by calling `push`
            bag.push(i * 1000 + j);
        }
    }));
}

for handle in thread_vec {
    handle.join().unwrap();
}

let mut vec_from_bag: Vec<_> = unsafe { bag.iter() }.copied().collect();
vec_from_bag.sort();
let mut expected: Vec<_> = (0..num_threads).flat_map(|i| (0..num_items_per_thread).map(move |j| i * 1000 + j)).collect();
expected.sort();
assert_eq!(vec_from_bag, expected);

Using `std::thread::scope`

An even more convenient approach would be to use thread scopes. This allows to use shared reference of the bag across threads, instead of Arc.

use orx_concurrent_bag::*;

let (num_threads, num_items_per_thread) = (4, 8);

let bag = ConcurrentBag::new();
let bag_ref = &bag; // just take a reference
std::thread::scope(|s| {
    for i in 0..num_threads {
        s.spawn(move || {
            for j in 0..num_items_per_thread {
                // concurrently collect results simply by calling `push`
                bag_ref.push(i * 1000 + j);
            }
        });
    }
});

let mut vec_from_bag: Vec<_> = bag.into_inner().iter().copied().collect();
vec_from_bag.sort();
let mut expected: Vec<_> = (0..num_threads).flat_map(|i| (0..num_items_per_thread).map(move |j| i * 1000 + j)).collect();
expected.sort();
assert_eq!(vec_from_bag, expected);

Safety

ConcurrentBag uses a PinnedVec implementation as the underlying storage (see SplitVec and Fixed). PinnedVec guarantees that elements which are already pushed to the vector stay pinned to their memory locations unless explicitly changed due to removals, which is not the case here since ConcurrentBag is a grow-only collection. This feature makes it safe to grow with a shared reference on a single thread, as implemented by ImpVec.

In order to achieve this feature in a concurrent program, ConcurrentBag pairs the PinnedVec with an AtomicUsize.

len: AtomicSize: fixes the target memory location of each element to be pushed at the time the push method is called. Regardless of whether or not writing to memory completes before another element is pushed, every pushed element receives a unique position reserved for it.
PinnedVec guarantees that already pushed elements are not moved around in memory during growth. This also enables the following mode of concurrency:
- one thread might allocate new memory in order to grow when capacity is reached,
- while another thread might concurrently be writing to any of the already allocation memory locations.

The approach guarantees that

only one thread can write to the memory location of an element being pushed to the bag,
at any point in time, only one thread is responsible for the allocation of memory if the bag requires new memory,
no thread reads any of the written elements (reading happens after converting the bag into_inner),
hence, there exists no race condition.

Construction

As explained above, ConcurrentBag is simply a tuple of a PinnedVec and an AtomicUsize.. Therefore, it can be constructed by wrapping any pinned vector; i.e., ConcurrentBag<T> implements From<P: PinnedVec<T>>. Further, there exist with_ methods to directly construct the concurrent bag with common pinned vector implementations.

use orx_concurrent_bag::*;

// default pinned vector -> SplitVec<T, Doubling>
let bag: ConcurrentBag<char> = ConcurrentBag::new();
let bag: ConcurrentBag<char> = Default::default();
let bag: ConcurrentBag<char> = ConcurrentBag::with_doubling_growth();
let bag: ConcurrentBag<char, SplitVec<char, Doubling>> = ConcurrentBag::with_doubling_growth();

let bag: ConcurrentBag<char> = SplitVec::new().into();
let bag: ConcurrentBag<char, SplitVec<char, Doubling>> = SplitVec::new().into();

// SplitVec with [Recursive](https://docs.rs/orx-split-vec/latest/orx_split_vec/struct.Recursive.html) growth
let bag: ConcurrentBag<char, SplitVec<char, Recursive>> =
    ConcurrentBag::with_recursive_growth();
let bag: ConcurrentBag<char, SplitVec<char, Recursive>> =
    SplitVec::with_recursive_growth().into();

// SplitVec with [Linear](https://docs.rs/orx-split-vec/latest/orx_split_vec/struct.Linear.html) growth
// each fragment will have capacity 2^10 = 1024
let bag: ConcurrentBag<char, SplitVec<char, Linear>> = ConcurrentBag::with_linear_growth(10);
let bag: ConcurrentBag<char, SplitVec<char, Linear>> = SplitVec::with_linear_growth(10).into();

// [FixedVec](https://docs.rs/orx-fixed-vec/latest/orx_fixed_vec/) with fixed capacity.
// Fixed vector cannot grow; hence, pushing the 1025-th element to this bag will cause a panic!
let bag: ConcurrentBag<char, FixedVec<char>> = ConcurrentBag::with_fixed_capacity(1024);
let bag: ConcurrentBag<char, FixedVec<char>> = FixedVec::new(1024).into();

Of course, the pinned vector to be wrapped does not need to be empty.

use orx_concurrent_bag::*;

let split_vec: SplitVec<i32> = (0..1024).collect();
let bag: ConcurrentBag<_> = split_vec.into();

Write-Only vs Read-Write

The concurrent bag is write-only & grow-only bag which is convenient and efficient for collecting elements.

See ConcurrentVec for a read-and-write variant which

guarantees that reading and writing never happen concurrently, and hence,
allows safe iteration or access to already written elements of the concurrent vector,
with a minor additional cost of values being wrapped by an Option.

Benchmarks

You may find the details of the benchmarks at benches/grow.rs.

In the experiment, ConcurrentBag variants and rayon is used to collect results from multiple threads. You may see in the table below that rayon is extremely fast with very small output data (i32 in this case). As the output size gets larger and copies become costlier, ConcurrentBag starts to perform faster.

orx-concurrent-bag 1.3.0