Skip to content

Performance Benchmarks

Nalix is engineered for high-throughput, low-latency real-time applications. This documentation provides a comprehensive report of 67 performance test suites executed on 2026-04-10.

Performance Philosophy

The framework achieves exceptional performance through several core design principles:

  • Lock-Free Abstractions: Minimizing contention using interlocked and thread-local patterns.
  • Zero-Allocation Pipelines: Reusing memory via advanced pooling for data transfers.
  • Pre-computed Metadata: Avoiding reflection and string lookups in the hot path.
  • Hardware-Aware Optimizations: Leveraging SIMD, aggressive inlining, and memory-safe Span<T> layouts.

Benchmark Categories

Explore detailed metrics by subsystem:

🚀 Industrial-Grade Benchmarking

Beyond micro-benchmarks, Nalix includes a high-concurrency stress testing tool located in tools/Nalix.Bench. This tool is used to validate system stability under localized "worst-case" scenarios.

Features

  • High Precision: Measures Round-Trip Time (RTT) with nanosecond precision.
  • Statistical Tail Latency: Reports P50, P90, P99, and P99.99 to identify jitter.
  • Concurrent Load: Simulations with thousands of sessions and millions of operations.
  • Zero-Allocation Validation: Monitoring Gen 0/1/2 GC counts during the test.

Usage:

# Ping-Pong Latency Test (1000 sessions, 100k pings)
./tools/benchmarks.cmd --mode ping --sessions 1000 --count 100000

# Throughput Test (1000 sessions, large payload)
./tools/benchmarks.cmd --mode throughput --sessions 1000 --payload 1024

Environment Details

Benchmarks were executed in the following environment:

  • CPU: 13th Gen Intel Core i7-13620H (2.40GHz)
  • Cores: 10 Physical, 16 Logical
  • Runtime: .NET 10.0.5 (X64 RyuJIT)
  • Environment: Performance Power Plan, Server GC Enabled
  • Toolchain: BenchmarkDotNet v0.15.8