Skip to content

README.md

Latest commit

 

History

History
110 lines (75 loc) · 5.55 KB

README.md

File metadata and controls

110 lines (75 loc) · 5.55 KB

NetFramework.Tasks.Management — Benchmarks

BenchmarkDotNet suite that measures the cost of every public operation exposed by ITaskManagement, targeting all three supported runtimes: net8.0, net9.0, and net10.0.

Benchmark classes

LifecycleBenchmarks

Isolates each individual stage of the task lifecycle. [InvocationCount(1)] is applied at the class level because every method mutates the shared static ConcurrentDictionary — each iteration must start from a clean slate. Targeted IterationSetup / IterationCleanup drive the task to the required state before timing begins, then reset afterwards.

Benchmark What is measured
RegisterTask ConcurrentDictionary.TryAdd + Task constructor
StartTask TryGetValue + Task.Start (thread-pool enqueue)
CancelTask TryGetValue + CancellationTokenSource.Cancel
DeleteTask TryRemove + Task.Dispose + GC.Collect (see ordering invariant in CLAUDE.md)
FullLifecycle All four stages end-to-end, including the forced GC

GetTasksStatusBenchmarks

Measures GetTasksStatus() (a ConcurrentDictionary → Dictionary snapshot) as the number of registered tasks grows. Tasks stay alive across all iterations — no InvocationCount(1) needed because the operation is purely read-only.

Parameter Values
TaskCount 1, 10, 50

CancelAllTasksBenchmarks

Measures the Parallel.ForEach fan-out inside CancelAllTasks as the number of running tasks grows. [InvocationCount(1)] + IterationSetup re-registers and re-starts all tasks before each measurement so cancellation state is always fresh.

Parameter Values
TaskCount 1, 10, 50

DataflowComparisonBenchmarks

Head-to-head comparison of NetTaskManagement vs TPL Dataflow for the scenarios where both libraries can solve the same problem. The goal is to help developers choose the right tool, not declare a winner — each library has distinct strengths.

Benchmark pair What is measured
NTM_StartNWorkers / Dataflow_StartNWorkers Cost of getting N concurrent workers running. NTM: RegisterTask × N + StartTask × N (ConcurrentDictionary inserts + Task.Start). Dataflow: new ActionBlock + Post × N.
NTM_CancelNWorkers / Dataflow_CancelNWorkers Cancel N running workers. NTM: CancelAllTasks() uses Parallel.ForEach and returns per-task status. Dataflow: single CancellationTokenSource.Cancel() shared across all workers.
NTM_ProcessNItems / Dataflow_ProcessNItems Process N short-lived items end-to-end. NTM: StartTask × N → CheckTaskStatusCompleted × N → DeleteTask × N (includes GC.Collect per delete). Dataflow: Post × N → Complete → Wait.

[InvocationCount(1)] applies to all pairs because every method mutates shared state. A Thread.Sleep(50) in the cancellation setup ensures workers are actively spinning before the timed cancellation call — without it, some items may not have been dequeued yet and would understate the true cancellation cost.

Parameter Values
N 10, 50, 100

Running

Benchmarks must be run in Release configuration. Debug builds produce meaningless results.

Single framework

# net8.0
dotnet run -c Release -f net8.0 --project benchmarks/NetFramework.Tasks.Management.Benchmarks/

# net9.0
dotnet run -c Release -f net9.0 --project benchmarks/NetFramework.Tasks.Management.Benchmarks/

# net10.0
dotnet run -c Release -f net10.0 --project benchmarks/NetFramework.Tasks.Management.Benchmarks/

Filter to a specific class

dotnet run -c Release -f net8.0 --project benchmarks/NetFramework.Tasks.Management.Benchmarks/ -- --filter *Lifecycle*
dotnet run -c Release -f net8.0 --project benchmarks/NetFramework.Tasks.Management.Benchmarks/ -- --filter *GetTasksStatus*
dotnet run -c Release -f net8.0 --project benchmarks/NetFramework.Tasks.Management.Benchmarks/ -- --filter *CancelAll*
dotnet run -c Release -f net8.0 --project benchmarks/NetFramework.Tasks.Management.Benchmarks/ -- --filter *DataflowComparison*

All benchmarks

dotnet run -c Release -f net8.0 --project benchmarks/NetFramework.Tasks.Management.Benchmarks/ -- --filter *

Interactive menu (no filter)

dotnet run -c Release -f net8.0 --project benchmarks/NetFramework.Tasks.Management.Benchmarks/

BenchmarkDotNet will list all available benchmark classes and let you choose interactively.

Output

Results are written to BenchmarkDotNet.Artifacts/ in the project directory:

BenchmarkDotNet.Artifacts/
└── results/
    ├── NetFramework.Tasks.Management.Benchmarks.LifecycleBenchmarks-report.md
    ├── NetFramework.Tasks.Management.Benchmarks.GetTasksStatusBenchmarks-report.md
    ├── NetFramework.Tasks.Management.Benchmarks.CancelAllTasksBenchmarks-report.md
    └── NetFramework.Tasks.Management.Benchmarks.DataflowComparisonBenchmarks-report.md

[MemoryDiagnoser] is enabled on all classes so each report includes allocated bytes per operation in addition to execution time.

Notes

  • DeleteTask and FullLifecycle will be slower than the other operations because DeleteTask calls GC.Collect() + GC.WaitForPendingFinalizers() by design (see the ordering invariant in CLAUDE.md).
  • CancelAllTasks uses Parallel.ForEach internally, so its cost scales sub-linearly with TaskCount on machines with multiple cores.
  • GetTasksStatus allocates a new Dictionary<string, TaskStatus> on every call — the allocation column in the report reflects this.