readme.md

Core operations evaluation

These scripts compare the performance of core N-dimensional array operations of Owl with NumPy and Julia. The evaluation results on one of our tested machines and detailed analysis can be seen at one chapter of Owl's documentation. Here we briefly introduce how to reproduce these results:

1. Run the three scripts (op_eval.ml, op_eval.py, and op_eval.jl) separately. Each will generate several csv files in current directory.

2. Run the python script draw_figure.py. It will create a ./fig directory if it does not exist, and save generated result figures there.