Matrix Multiplication Profiling

A performance comparison project demonstrating how progressive optimizations improve matrix multiplication speed from naive Python loops to optimized C with SIMD, tiling, and OpenMP parallelization.

---

Screenshots

Runtime vs Size	OpenMP Scaling

---

Implementations

Each stage builds upon the previous to demonstrate the impact of algorithmic and hardware-level optimizations:

Step	Implementation	Description
1️⃣	python_naive	Simple triple-loop in pure Python
2️⃣	naive_c	Direct translation of the naive algorithm into C
3️⃣	reorder_c	Loop reordering for better cache access patterns
4️⃣	omp_c	Adds multithreading with OpenMP
5️⃣	tiled_c	Cache blocking (tiling) for L2/L3 cache efficiency
6️⃣	simd_c	SIMD vectorization using AVX2 intrinsics

---

Requirements

• C++11/C11 compiler (GCC ≥ 9, Clang ≥ 10)
• CMake ≥ 3.16
• Python ≥ 3.8
• Matplotlib, pandas, seaborn

---

Build & Run

1.Build the C Benchmarks

cmake -S . -B build
cmake --build build -j

2.Run C Benchmarks

cd benchmark
bash bench.sh

3.Run Python Baseline

cd python
python3 bench_python.py

4.Generate Plots

cd ../benchmark
python3 plot_results.py ../data/results.csv