keith-turner commented on PR #96:
URL: https://github.com/apache/accumulo-access/pull/96#issuecomment-3716835274

   Ran the benchmark against commit dd2ef2dfb0124b96960df3de82ee6ae94c7fb07d 
from this PR and saw the following.
   
   ```
   AccessExpressionBenchmark.measureBytesValidation           thrpt   12  
14.172 ± 0.031  ops/us
   AccessExpressionBenchmark.measureLegacyEvaluationOnly      thrpt   12  
22.322 ± 0.150  ops/us
   AccessExpressionBenchmark.measureLegacyParseAndEvaluation  thrpt   12   
9.637 ± 0.384  ops/us
   AccessExpressionBenchmark.measureParseAndEvaluation        thrpt   12  
10.821 ± 0.112  ops/us
   AccessExpressionBenchmark.measureStringValidation          thrpt   12  
17.741 ± 0.429  ops/us
   ```
   
   Then ran against c70d418ef10c008344df860c787fe947d5ef8a3a from main and saw 
the following.
   
   ```
   AccessExpressionBenchmark.measureBytesValidation           thrpt   12  
35.619 ± 2.259  ops/us
   AccessExpressionBenchmark.measureLegacyEvaluationOnly      thrpt   12  
22.444 ± 0.409  ops/us
   AccessExpressionBenchmark.measureLegacyParseAndEvaluation  thrpt   12   
9.495 ± 0.021  ops/us
   AccessExpressionBenchmark.measureParseAndEvaluation        thrpt   12  
16.947 ± 0.328  ops/us
   AccessExpressionBenchmark.measureStringValidation          thrpt   12  
28.937 ± 0.013  ops/us
   ```
   
   Improved the benchmark times with the changes in 
22af52dae335f8ba9d9b94cb7273252df6556cd3 .  These changes use a char[] array 
instead of String internally to avoid calling with charAt() on string so 
frequently.
   
   ```
   AccessExpressionBenchmark.measureBytesValidation           thrpt   12  
18.068 ± 0.386  ops/us
   AccessExpressionBenchmark.measureLegacyEvaluationOnly      thrpt   12  
22.262 ± 0.165  ops/us
   AccessExpressionBenchmark.measureLegacyParseAndEvaluation  thrpt   12   
9.413 ± 0.164  ops/us
   AccessExpressionBenchmark.measureParseAndEvaluation        thrpt   12  
12.825 ± 0.253  ops/us
   AccessExpressionBenchmark.measureStringValidation          thrpt   12  
22.166 ± 1.012  ops/us
   ```
   
   
   Experimented w/ adding char[] methods to the public API (in addition to 
string methods) and those are faster when you already have a char array.  
However using the java CharDecoder class to create char[] is much slower than 
string, so its not a net win.


-- 
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.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to