This repository contains a versatile CSV merging tool with both GPU (CUDA) and CPU versions for efficient consolidation of multiple CSV files. The tool leverages RAPIDS cuDF and Dask libraries for GPU acceleration.

• 中文
• Deutsch

• GPU (CUDA) and CPU Versions: Choose the appropriate version based on your hardware configuration.
• Efficient Merging: Consolidate multiple CSV files into a single merged CSV file.
• Flexible Naming: Customize the naming convention for the merged CSV file.
• Support for Large Files: Handles large CSV files with ease.

Additional Features for GPU Version:

• CUDA Acceleration: Utilizes the power of CUDA for significantly faster CSV merging operations compared to CPU computation.

python merge_gpu.py --folder_path /path/to/csv/files

python merge_cpu.py --folder_path /path/to/csv/files

Note: The folder specified is relative to the current directory, not an absolute path. If you need to modify the way the directory is specified, simply adjust the following lines:

# Specify a specific folder as the directory and call the function
specified_folder = 'Example0'
folder_path = os.path.join(os.getcwd(), specified_folder)
merge_csv_files(folder_path)

• GPU Version:

  • Python 3.x
  • RAPIDS cuDF
  • Dask (for Dask cuDF support)
  • Linux environment
  • NVIDIA GPU with CUDA support

• CPU Version:

  • Python 3.x

Note:

• The GPU version requires a Linux environment with an NVIDIA GPU that supports CUDA.
• RAPIDS is not compatible with Windows.
• CUDA does not support macOS.
• For detailed instructions on installing RAPIDS, please refer to the RAPIDS documentation.

For demonstration purposes, a folder named Example0 has been provided, containing three CSV files: file0.csv, file1.csv, and file2.csv. You can run the tool on this example by specifying the folder path.

python merge_gpu.py --folder_path Example0