Simple SPMV

This project is a simple implementation of Sparse Matrix-Vector Multiplication (SPMV) with various sparse matrix formats on different devices (e.g., GPU and ARM). The project is implemented in C++ and CUDA.

ToDoLists

GPU naive kernel:

• DIA
• ELL
• CSR(scalar)
• CSR(vector)
• COO(segment)
• CSR-Stream
• CSR5

ARM naive kernel:

• DIA
• ELL
• CSR(scalar)
• CSR(vector)
• COO(segment)
• CSR-Stream
• CSR5

Requirements

• CUDA Toolkit
• C++ compiler (e.g., g++)
• GNU make
• CMake

Building

To build the project, you can use the following commands:

mkdir build && cd build
cmake ..
make -j8 && make install

Usage

To run the program, use the following command:

# If you want to run the program on the NVIDIA GPU.
./bin/cuda_test_all <M> <K>

• <M>: The dimension M of the matrix.
• <K>: The dimension K of the matrix.

License

This project is licensed under the MIT License.