Package jline.bench
package jline.bench
Benchmarking utilities and performance test models for LINE solvers.
This package provides a comprehensive benchmarking framework for evaluating the performance and accuracy of LINE's various solvers across different model types and problem sizes. It includes predefined benchmark models, performance measurement utilities, and comparison tools.
Benchmark Categories
jline.bench.cqn- Closed Queueing Network benchmarksjline.bench.fj- Fork-Join model benchmarksjline.bench.lqn- Layered Queueing Network benchmarks
Features
- Standardized benchmark models for solver comparison
- Performance timing and memory usage measurement
- Accuracy validation against reference solutions
- Scalability testing with varying model sizes
- Statistical analysis of solver performance
Usage
Benchmark classes typically provide static methods to generate test models of various sizes and complexities. These can be used to evaluate solver performance across different scenarios.
Example usage:
// Run CQN benchmark
Network model = CQNBench.generate(10, 3); // 10 stations, 3 classes
long startTime = System.nanoTime();
NetworkAvgTable result = new SolverMVA(model).getAvgTable();
long duration = System.nanoTime() - startTime;
- Since:
- LINE 2.0
- See Also:
-
ClassesClassDescriptionMain benchmark runner - runs all benchmark suitesFormats benchmark results in MATLAB-style tablesHandles benchmark regression testing by storing and comparing against baseline scoresContainer for benchmark resultsContainer for regression baseline dataContainer for regression comparison resultsBenchmark solver initialization utilityUtility functions for benchmarkingQuick test of benchmark package functionality