Speed Results of C++ vLib XML Parsing

This is what I’ve got so far:

The test: Getting all the diffuse, specular, and normal maps used in a chunk file that has 669859 lines.

Note – I’m not checking for duplicate entries when appending to the array, the number of textures is basically a count of how many times “Diffuse_Map”, “Normal_Map” and “Specular_Map” show up in the file.

Another thing to note – Initializing vlib adds .5 second overhead to the first run of the function. When parsing a small file, using this module may be overkill. You’ll see the most performance gain in larger files or when parsing multiple files in a single instance of an application.

Using SAX:
Number of textures 351
Total time - 20.9679999352

Using RE Parsing:
Number of textures 351
Total time - 9.78900003433

Using the wrapped parser:
Number of textures 351
Total time - 1.47599983215


In summary: