stevedlawrence commented on issue #318: Add new TimeTracker for more advanced performance analysis URL: https://github.com/apache/incubator-daffodil/pull/318#issuecomment-581434686 An example of the output, adding the TImeTracker around the parse() call like mentioned in the commit message and parsing a large CSV-like data format, we get the following stats: ``` Name Time Pct Average Count LiteralNilDelimitedEndOfDataParser 3.330 34.03% 4030 826140 StringDelimitedParser 2.455 25.09% 4184 586640 DelimiterTextParser 1.038 10.61% 879 1180480 SimpleNilOrValueParser 0.985 10.07% 1192 826140 OrderedSeparatedSequenceParser 0.806 8.23% 10232 78720 ElementParser 0.404 4.13% 342 1180520 DelimiterStackParser 0.308 3.15% 244 1259220 ChoiceParser 0.226 2.31% 5750 39360 SeqCompParser 0.113 1.15% 318 354300 ConvertTextNumberParser 0.060 0.61% 1489652 40 OrderedUnseparatedSequenceParser 0.058 0.60% 2922016 20 ConvertTextCombinatorParser 0.000 0.00% 8825 40 Total Time: 9.783 ``` So we can see we spend a third of the time trying to parse delimted Nils, a quarter of the time trying to parse delimited strings, a thenth parsing actual delimiters, and so on. So to get any real benefits, we really just need to focus on speeding up these parsers. Note that this isn't too surpsing, we would expect delimiter scanning to take up a good chunk of the time, but it's helpful to see that in number form.
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services
