|
8 | 8 | Classes that implement all-purpose, rock-solid **random number generators**. |
9 | 9 |
|
10 | 10 | Library priorities: |
11 | | -- perfect match of the numbers on all platforms |
| 11 | +- generation of identical bit-accurate numbers regardless of the platform |
| 12 | +- reproducibility of the same random results in the future |
| 13 | +- high-quality randomness |
12 | 14 | - performance |
13 | | -- quality of the numbers produced |
14 | 15 |
|
15 | 16 | Algorithms are [Xoshiro](https://prng.di.unimi.it/) for **quality** and |
16 | 17 | [Xorshift](https://en.wikipedia.org/wiki/Xorshift) for **speed**. |
17 | 18 |
|
18 | 19 | # Speed |
19 | 20 |
|
20 | | -Generating 50 million of random numbers with AOT-compiled binary. |
| 21 | +Generating 50 million random numbers with AOT-compiled binary. |
21 | 22 |
|
22 | 23 | | Time (lower is better) | nextInt | nextDouble | nextBool | |
23 | 24 | |------------------------|---------|------------|----------| |
@@ -176,8 +177,8 @@ All the benchmarks on this page are from AOT-compiled binaries running on AMD A9 |
176 | 177 |
|
177 | 178 | The library has been thoroughly **tested to match reference numbers** generated by C algorithms. The |
178 | 179 | sources in C are taken directly from scientific publications or the reference implementations by the inventors of the algorithms. The Xorshift128+ results are also matched to reference |
179 | | -values from [JavaScript xorshift library](https://github.com/AndreasMadsen/xorshift), that tested |
| 180 | +values from [JavaScript xorshift library](https://github.com/AndreasMadsen/xorshift), which tested |
180 | 181 | the 128+ similarly. |
181 | 182 |
|
182 | | -Testing is done in the GitHub Actions cloud on **Windows**, **Ubuntu** and **macOS** in **VM** and **Node.js** modes. |
| 183 | +Testing is done in the GitHub Actions cloud on **Windows**, **Ubuntu**, and **macOS** in **VM** and **Node.js** modes. |
183 | 184 |
|
0 commit comments