Benchmarks

TaskShed was benchmarked against two established asynchronous and persistent task scheduling libraries - APScheduler and ARQ to assess performance in selected areas.

The benchmarks were run on a MacBook Air (M1, 2020) with macOS Sonoma. They have not been independently verified. The primary aim was to provide a transparent, reproducible comparison of TaskShed’s performance under specific test conditions.

All code for these benchmarks is available in the tests/benchmarks directory of the TaskShed repository.

Time to Schedule 500 Tasks (s)

This test measures how long each framework takes to schedule 500 tasks. Faster scheduling is particularly relevant for applications that dynamically create large numbers of tasks.

In these tests, TaskShed scheduled tasks in less time than APScheduler and ARQ. The batched scheduling features of TaskShed (using either Redis or MySQL) showed especially low scheduling times.

Total CPU Time (s) to Process 50,000 Tasks

This benchmark measures the total CPU time required to process 50,000 tasks. Lower CPU time can indicate higher resource efficiency, which may be beneficial in terms of operational cost and energy use.

In these measurements, TaskShed used less CPU time than the other tested frameworks.

Task Execution Lag (ms)

Execution lag refers to the time difference between a task's scheduled execution and its actual start time. Lower lag indicates that tasks are executed closer to their intended schedule.

Again, TaskShed recorded lower execution lag than APScheduler and ARQ.