*edit* Sorry – jumped the gun with my original test code here – need to close the IndexWriter after the optimize! The gains are only with multi segment indexes. Corrected entry follows:

Lets do a little test. We will load up a FieldCache with 5,000,000 unique strings and see how long it takes Lucene 2.4 in comparison to Lucene 2.9.

Lets use my quad core laptop and the following test code:

public class ContrivedFCTest extends TestCase {
  public void testLoadTime() throws Exception {
    Directory dir = FSDirectory.getDirectory(System.getProperty("java.io.tmpdir") + File.separator + "test");
    IndexWriter writer = new IndexWriter (dir, new SimpleAnalyzer(), true, IndexWriter.MaxFieldLength.LIMITED);
    for(int i = 0; i < 5000000; i++) {
      Document doc = new Document();
      doc.add (new Field ("field",  "String" + i, Field.Store.NO, Field.Index.NOT_ANALYZED));

    IndexReader reader = IndexReader.open(dir);
    long start = System.currentTimeMillis();
    FieldCache.DEFAULT.getStrings(reader, "field");
    long end = System.currentTimeMillis();
    System.out.println("load time:" + (end - start)/1000.0f + "s");

The results?

Lucene 2.4: 150.726s
Lucene 2.9: 9.695s

We discovered early this year that in the past, Lucene has been terribly inefficient when loading FieldCaches over multiple segments. Lucene 2.9 addresses this at the MultiReader level (thank you Yonik!). Also, internal FieldCache usage is now per segment, which sidesteps loading FieldCaches over mutiple segments all together – each segment has its own FieldCache.