viswanathk commented on issue #14002: URL: https://github.com/apache/lucene/issues/14002#issuecomment-2498525370
> Interesting. The "theoretical" max depth of this stack would be the size of this graph. I suppose the stack does get large, which would explain a high no. of Array::grow calls? Would be interesting to see the impact of just setting initial capacity to half of the size of graph. Maybe `sqrt` might be a good start? On next grow it should be large enough, and is probably small enough starting point. > I do think it's worth improving. Another way could be to measure empirically the stack depth - maybe it scales in a predictable way with total number of vectors? And then we can use that prediction as an initial size. We could also think of using a primitive array that grows (implement our own primitive ArrayDeque). Should we be concerned about the chance of a humongous allocation (if the stack is really deep) with primitives? If not, I can give it a shot with running the same benchmarks with primitives and starting at `sqrt` of the graph size at that level. -- 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 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
