Performance
NPipeline is designed for throughput. The execution engine uses ValueTask fast paths, compiled node factories, execution plan caching, and object pooling to minimize per-item overhead.
Most pipelines are fast without tuning. When you need to squeeze out more, these guides cover the optimization techniques available - many of which are enforced automatically by build-time analyzers.
For pipelines processing millions of items per second, switch to PipelineOptimizationProfile.HighThroughput to enable zero-allocation context dictionaries and activate all performance analyzers. See Optimization Profiles for details.
In This Section
- Best Practices - do's and don'ts backed by analyzer rules
- Synchronous Fast Paths - eliminate Task allocations with
ValueTaskfor synchronous transforms - Execution Plan Caching - avoid reflection overhead on repeated pipeline runs