Sorting Algorithm Benchmark

A C++ benchmark comparing insertion sort, merge sort, and quicksort across increasingly large randomized inputs. The program counts comparisons and swaps, exports results, and supports Gnuplot visualizations on linear and logarithmic scales.

Key finding

At 10,000 elements, insertion sort required roughly 49.8 million measured operations, compared with approximately 254 thousand for merge sort and 253 thousand for quicksort. The experiment makes the practical difference between quadratic and n log n growth visible.

Measured results

Input size	Insertion total operations	Merge total operations	Quick total operations
100	4,518	1,221	1,052
1,000	486,487	18,675	18,488
5,000	12,480,130	117,019	105,933
10,000	49,784,822	254,072	252,819

The measured growth illustrates insertion sort's quadratic behavior and the substantially better scaling of merge sort and quicksort.

Visualizations

Build and run

g++ -std=c++17 -O2 main.cpp -o sorting-benchmark
./sorting-benchmark

With Gnuplot installed, run generate_all_plots.sh to regenerate the charts.

Testing and verification

The benchmark was verified by compiling with a C++17 compiler and checking that each algorithm produces sorted output for the generated input sizes. The committed sorting_results.csv records the comparison, swap, and total-operation counters used in the charts.

g++ -std=c++17 -O2 -Wall -Wextra main.cpp -o sorting-benchmark
./sorting-benchmark

This is an independent algorithm-analysis portfolio project; no formal course attribution is claimed in the source materials.

Testing and verification

The benchmark was verified by compiling with a C++17 compiler and checking that each algorithm produces sorted output for the generated input sizes. The committed sorting_results.csv records the comparison, swap, and total-operation counters used in the charts.

g++ -std=c++17 -O2 -Wall -Wextra main.cpp -o sorting-benchmark
./sorting-benchmark

This is an independent algorithm-analysis portfolio project; no formal course attribution is claimed in the source materials.

Skills demonstrated

• C++ algorithm implementation
• Experimental algorithm analysis
• Comparisons and swap instrumentation
• CSV export and Gnuplot visualization
• Relating measured behavior to Big-O complexity

About the author

Built by Ahmed Balde to connect algorithm theory with measured runtime behavior. See more C++, Python, data, and software-engineering work on GitHub.