A "quick" microbenchmark update PHP 5.4
This is a follow up to A "quick" microbenchmark... you may want to go and read that first.
With the release of PHP 5.4 being toted as a "significant performance improvement" I thought it might be fun to go back and run the quicksort benchmark against the new release. I upgraded to 5.4.3 on my local machine and ran the tests.
Results
| Language | Version | Time in milliseconds | |
|---|---|---|---|
| 1 | C++ | MS 16.00.30319.01 for 80x86 | 75 |
| 2 | C# Array.Sort() | .NET 3.5 | 100 |
| 3 | Java | 1.6.0_23 | 105 |
| 4 | Groovy api (aka java) | 1.8.5 | 110 |
| 5 | Java Arrays.sort() | 1.6.0_23 | 121 |
| 6 | C# | .NET 3.5 | 126 |
| 7 | Scala | 2.9.1.final | 128 |
| 8 | Node.js (v8) | 0.6.7 | 175 |
| 9 | Chrome JavaScript | 16.0.912.77 | 182 |
| 10 | Ruby api array.sort! | 1.9.2p290 | 250 |
| 11 | Firefox JavaScript | 10.0 | 271 |
| 12 | IE JavaScript | 9.0.8112.16421 | 307 |
| 13 | IE JavaScript api sort() | 9.0.8112.16421 | 375 |
| 14 | Node.js (v8) api sort() | 0.6.7 | 480 |
| 15 | Chrome JavaScript api sort() | 16.0.912.77 | 520 |
| 16 | Python api sort() | 3.1.3 | 814 |
| 17 | PHP api sort() | 5.3.8 | 1441 |
| 18 | PHP api sort() | 5.4.3 | 1516 |
| 19 | Firefox JavaScript api sort() | 10.0 | 3490 |
| 20 | Ruby | 1.9.2p290 | 3520 |
| 21 | Groovy | 1.8.5 | 4100 |
| 22 | PHP | 5.4.3 | 5302 |
| 23 | Python | 3.1.3 | 8100 |
| 24 | PHP | 5.3.8 | 9700 |
Performance improvement? Yes!
The PHP api version was the same as expected. I didn't look at the source but I would not have expected the sort algorthim to have been changed. As you can see though, the implemented version was "significantly" faster as promised... a whopping 45% faster! That is quite the improvement indeed. There are some other breaking compatibility changes though to be aware of. Anyone still relying on register_globals or magic_quotes please step forward !!
As before you can view the code for each implementation on github.
