DO NOT REPLY TO THIS EMAIL, BUT PLEASE POST YOUR BUG RELATED COMMENTS THROUGH THE WEB INTERFACE AVAILABLE AT <http://issues.apache.org/bugzilla/show_bug.cgi?id=28183>. ANY REPLY MADE TO THIS MESSAGE WILL NOT BE COLLECTED AND INSERTED IN THE BUG DATABASE.
http://issues.apache.org/bugzilla/show_bug.cgi?id=28183 [Patch] replace DocumentWriter with InvertedDocument for performance ------- Additional Comments From [EMAIL PROTECTED] 2004-04-07 03:35 ------- Yes, the patch is not the easiest to read. Sorry about that! I thought I'd send it in to get the conversation started. Unfortunately I needed to do a bit of refactoring before starting (to make it possible to subclass SegmentInfo), so it wouldn't be a very clean patch anyway. As far as memory usage goes, in my particular case I don't think it matters. I'm running java with 256M of memory on a machine with 768M, and java isn't making much use of the memory it has. (Reducing Java's memory to 128M results in only a 10% slowdown due to increased garbage collection.) Raising minMergeDocuments beyond about 500 seems to result in no performance improvement, either with or without the patch. The task is CPU-bound with extra memory and I/O bandwidth available, so it looks like trading memory for performance makes sense even if the memory usage is higher. However, perhaps this isn't true for other machines and/or other datasets? Do you have an example of a workload where memory is the bottleneck? Or perhaps I'm missing something about how to tune Lucene? --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
