NeuroCalc 🧠

NeuroCalc is a high-performance C++ extension library designed to accelerate common mathematical operations used in neural network computations. By offloading vectorized math to native C++, we bypass the Python Global Interpreter Lock (GIL) and overhead.

🚀 Key Features

• Vectorized Hypotenuse: Optimized Euclidean distance calculation using -O3 compiler flags.
• Dot Product Kernel: High-speed floating-point multiplication and summation.
• Memory Optimization: Uses std::vector::reserve to minimize reallocations during runtime.

📊 Performance Benchmarks

We tested NeuroCalc against standard Python list comprehensions with 100,000 elements:

Operation	Python Time	C++ Time	Speedup
Hypotenuse	1.53s	0.53s	~2.8x
Dot Product	0.12s	0.08s	~1.5x

🛠️ Installation

NeuroCalc uses scikit-build-core for a seamless C++ build experience within Python.