该项目实现了不同语言下的k-means算法。
其特点在于实现了cuda及cuda-numba的k-means代码。
我们在以下环境中进行了实验：
- Intel Core i5-4200H
- NVIDIA Geforce GTX850M
- Ubuntu 16.04
对其代码的运行速度进行了比较，并得到了以下运行结果(仅供参考)：
- cuda (9.0): 5.6ms
- cuda-numba (0.38.0): 13.2ms
- numba (0.38.0): 75ms
- openmp (2 threads): 119ms
- c++ (g++ 5.4.0): 200ms
- python (3.6.5): 10000ms
该项目代码使用了Jansson依赖包用于处理JSON文件，并使用了CMake作为编译工具。
该项目参考了 https://github.com/andreaferretti/kmeans ，由 Romic Huang 在2018年春季完成。
This project implements k-means algorithm in different languages.
Specially, it realize k-means' codes of CUDA and cuda-numba.
We conducted experiment in the following environments:
- Intel Core i5-4200H
- NVIDIA Geforce GTX850M
- Ubuntu 16.04
We compares the speed of the codes and obtains the following results (for reference only):
- cuda (9.0): 5.6ms
- cuda-numba (0.38.0): 13.2ms
- numba (0.38.0): 75ms
- openmp (2 threads): 119ms
- c++ (g++ 5.4.0): 200ms
- python (3.6.5): 10000ms
This project uses Jansson to handle JSON file and uses CMake as a compilation tool.
This project reference https://github.com/andreaferretti/kmeans, and was completed by Romic Huang in the spring of 2018.