This repository serves as a benchmarking for different implementations of two-dimensional arrays in Rust - specifically two dimensional arrays with the dimension of (n, 3) used to describe three-dimensional points in space. It compares the execution time of accessing and modifying two-dimensional array representations that include:
Array2<f32>from the ndarray crate- A flattened vector with the standard library
Vec<f32> - A vector of arrays
Vec<[f32; 3]> - A vector of vectors
Vec<Vec<f32>>
The current test results for each array representation are outlined below:
| Test | Execution Time (ns/iter) |
|---|---|
benchmark_add_one_to_each_element_array2 |
102,033 ns/iter (+/- 27,931) |
benchmark_add_one_to_x_value_array2 |
18,199 ns/iter (+/- 1,289) |
benchmark_get_sub_array2 |
273 ns/iter (+/- 19) |
benchmark_add_one_to_each_element_flattened_vec |
17,347 ns/iter (+/- 12,005) |
benchmark_add_one_to_x_value_flattened_vec |
25,115 ns/iter (+/- 8,057) |
benchmark_get_sub_flattened_vec |
4,249 ns/iter (+/- 8,985) |
benchmark_add_one_to_each_element_vec_of_arrays |
40,618 ns/iter (+/- 29,715) |
benchmark_add_one_to_x_value_vec_of_arrays |
21,157 ns/iter (+/- 24,437) |
benchmark_get_sub_vec_of_arrays |
160 ns/iter (+/- 166) |
benchmark_add_one_to_each_element_vec_of_vecs |
163,413 ns/iter (+/- 183,558) |
benchmark_add_one_to_x_value_vec_of_vecs |
91,462 ns/iter (+/- 164,705) |
benchmark_get_sub_vec_of_vecs |
606 ns/iter (+/- 402) |