search

BitFunnel / mg4j-workbench

Java tools for evaluating BitFunnel performance compared to an mg4j baseline.
GNU Lesser General Public License v3.0
1 stars 2 forks source link

Investigate whether mg4j has a mode that counts matches. #2

Closed MikeHopcroft closed 7 years ago

MikeHopcroft commented 7 years ago

The act of returning a list of query results may cause GC-related performance problems. Might want to consider counting the matches instead of returning them.

One way to do this would be to modify QueryEngine.getResults() (starts on 387 of QueryEngine.java) to count the results, but not return them.

MikeHopcroft commented 7 years ago

On the other hand, BitFunnel returns the list of all matches, so an apples-to-apples comparison should probably collect matches in mg4j as well.

MikeHopcroft commented 7 years ago

Closing since we'd like the experiment to be apples-to-apples as much as possible. This will also save space in the paper, as we won't have to explain the differences and their rationale.