@ -14,8 +14,8 @@ were performed in ***8ms*** on a commodity PC.<sup>2</sup> The next fastest con
| HikariCP | 8 | 7 | 13 |
<sub><sup>1</sup>We contend HikariCP is near the theoretical maximum on current JVM technology.</sub><br/>
<sub><sup>2</sup>400 threads, 50 connection pool. Measurements taken in *nanoseconds* and converted to *milliseconds*.
See benchmarks [here](https://github.com/brettwooldridge/HikariCP/wiki/Benchmarks)</sub><br/>
<sub><sup>2</sup>400 threads, 50 connection pool. Measurements in *nanoseconds* and converted to *milliseconds*.
See benchmarks [here](https://github.com/brettwooldridge/HikariCP/wiki/Benchmarks). See how we do it [here](https://github.com/brettwooldridge/HikariCP/wiki/Down-the-Rabbit-Hole).</sub><br/>