AsyncSemaphore 2.0.0

AsyncSemaphore

A fast, allocation-free async semaphore featuring:

• Automatic releasing without try/finally blocks by utilising the IDisposable using pattern
• Guarantee that release can only be called once per WaitAsync call
• A custom lock-free core that outperforms SemaphoreSlim (~2x faster uncontended, ~27% faster async handoff) and allocates zero bytes even for contended async waits (pooled IValueTaskSource waiters)
• Analyzers to help you implement the desired pattern
• An IAsyncSemaphore interface for if you need to mock

Install

dotnet add package AsyncSemaphore

Usage

private readonly AsyncSemaphore _asyncSemaphore = new AsyncSemaphore(1);

public async Task MyMethod()
{
    // Just assign the `IDisposable` returned from `WaitAsync` to a variable and use the using statement with it
    using var lockHandle = await _asyncSemaphore.WaitAsync();

    // Do whatever you want - Even if we throw exceptions, we'll release the semaphore once we leave this method's scope
    await DoSomethingInsideLock();
}

or scoped:

private readonly AsyncSemaphore _asyncSemaphore = new AsyncSemaphore(1);

public async Task MyMethod()
{
    // or create your own scope with {} braces - And after you leave that scope, your lock will be released
    using (await _asyncSemaphore.WaitAsync())
    {
        await DoSomethingInsideLock();
    }

    await DoSomethingAfterLockReleased();
}

Performance

AsyncSemaphore no longer wraps SemaphoreSlim — it has its own lock-free core:

• Uncontended waits are a single interlocked operation (no monitor lock): ~2x faster than SemaphoreSlim.
• Contended async waits use pooled (including thread-local cached) IValueTaskSource waiter nodes: ~27% faster handoff and zero bytes allocated per wait, versus 88 B per async waiter for SemaphoreSlim.

BenchmarkDotNet v0.15.8, Windows 11
12th Gen Intel Core i7-12700K 3.60GHz, 1 CPU, 20 logical and 12 physical cores
.NET SDK 10.0.400
  [Host]     : .NET 10.0.11, X64 RyuJIT x86-64-v3
  DefaultJob : .NET 10.0.11, X64 RyuJIT x86-64-v3

Uncontended is a wait/release cycle with the permit available. AsyncHandoff measures handing the permit to a queued async waiter. Parallel is 4 workers hammering a single-permit semaphore with an await Task.Yield() inside the lock — AsyncSemaphore trades ~5% mean time there for fully allocation-free waits.