Skip to content

Latest commit

 

History

History
39 lines (30 loc) · 2 KB

README.md

File metadata and controls

39 lines (30 loc) · 2 KB

Benchmark Results and Artifacts

This directory contains experimental benchmark data, system provenance metadata, statistical summaries, and performance scaling visualizations generated by ParallelComputeLab.


1. Directory Contents

File Type Description
system_metadata.json JSON Hardware topology, core counts, cache sizes, OS, kernel, compiler flags, and CPUID features.
raw_benchmark_results.json JSON Raw Google Benchmark output containing all individual repetition timings, iteration counts, and byte counters.
raw_benchmark_results.csv CSV Comma-separated export of benchmark results for spreadsheet or R/Pandas analysis.
summary_metrics.json JSON Computed statistical aggregates: median latency, IQR, standard deviation, bandwidth (GB/s), and speedup.
summary_table.md Markdown Formatted summary tables comparing execution times, parallel speedups, and efficiencies.
scaling_threads.png PNG Chart Multi-thread strong scaling speedup $S(p)$ and parallel efficiency $E(p)$ for $p \in {1, 2, 4, 8}$.
workload_comparison.png PNG Chart Execution time comparison across Scalar, Compiler-Optimized, SIMD, and Parallel implementations.
memory_bandwidth.png PNG Chart Latency degradation vs access stride (1 to 64) demonstrating cache-line eviction costs.
scheduling_comparison.png PNG Chart Runtime comparison of OpenMP static, dynamic, and guided scheduling on uniform vs imbalanced workloads.

2. Reproducing Results

To execute the benchmarks afresh and update all results:

# 1. Run benchmarks with 3 repetitions
python3 ParallelComputeLab/scripts/run_benchmarks.py \
  --binary build/release/parallel_compute_benchmarks \
  --output-dir ParallelComputeLab/results \
  --repetitions 3 \
  --min-time 0.2

# 2. Re-generate analysis tables and plots
python3 ParallelComputeLab/scripts/analyze_results.py \
  ParallelComputeLab/results/raw_benchmark_results.json \
  ParallelComputeLab/results