easyice opened a new issue, #12514: URL: https://github.com/apache/lucene/issues/12514
### Description Currently in BKD LeafNode, we scan all 512 values and call `visitor.visit ` in `CELL_CROSSES_QUERY` case, this is usually not a issue for range query, but for point query, such as `PointInSetQuery`, maybe only one value hit at a leaf node, but it still scan all the block. in 1D case, values will visited in increasing order. maybe we can create a special index structure for these 512 values, for instance, write the point value/offset per 32 values, used to skip some point values that will not match, i had wrote some POC code, it will improve query performance by 20% in high cardinality(10 million), but no improve in lower cardinality -- 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: issues-unsubscr...@lucene.apache.org.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org For additional commands, e-mail: issues-h...@lucene.apache.org
