Parallelized implementation of the Smith-Waterman algorithm for PiA GPU101 @PoliMi.

• Features
• How It Works
• Build and Run Instructions

• Parallelized Score and Direction Matrix Computation;
• GPU Backtrace Implementation;
• CPU-GPU Comparison;
• Pseudo-Random Sequence Generation using the Collatz-Weyl Generator (CWG128).

1. Sequence Generation:

   • Sequences are randomly generated using the Collatz-Weyl Generator, producing nucleotide sequences of given length (S_LEN) for alignment.

2. Parallelized Computation:

   • The CUDA kernel sw_parallelized computes the score matrix, direction matrix, and performs the traceback. Each CUDA block handles one sequence pair.

3. CPU Baseline:

   • The sw_serialized function implements a single-threaded CPU version of Smith-Waterman for comparison.

1. Clone the Repository:

   git clone https://github.com/Penzo00/Smith-Waterman.git
   cd Smith-Waterman

2. Build the Program: Use nvcc to compile the CUDA code:

   nvcc -o sw_parallel_final_true sw_parallel_final_true.cu -O3

3. Run the Executable. Optionally, pass a seed for the Collatz-Weyl Generator:

   ./sw_parallel_final_true [seed]