RFR 8005698 : Handle Frequent HashMap Collisions with Balanced Trees
brent.christian at oracle.com
Wed May 22 20:50:37 UTC 2013
I have updated the changes as follows:
* TreeBin.createEntryForNode() + the anonymous TreeBin subclass in
[Linked]HashMap have been replaced by a TreeBin.map reference back to
the containing map, enabling TreeBin to just call newEntry() directly.
* TreeNode.entry was made final
* Added a top-level comment in HashMap giving a brief overview of how
balanced trees are used
* Updated the TreeBinSplitBackToEntries test for TREE_THRESHOLD=16
* Test code was updated to not need to be on the bootclasspath.
LaunchOnBootClass.java was removed from the changes (and would not have
been needed anyway, due to the recent jtreg update)
The new webrev is here:
On 5/13/13 9:48 AM, Brent Christian wrote:
> Please review my changes for 8005698 : Handle Frequent HashMap
> Collisions with Balanced Trees.
> For HashMap and LinkedHashMap, we would like to duplicate the technique
> (and some of the code) that the latest ConcurrentHashMap uses for
> handling hash collisions: use of balanced trees to store map entries in
> hash bins containing many collisions.
> Webrev is here:
> Some high-level details copied from the JEP:
> Earlier work in this area in JDK 8, namely the alternative
> string-hashing implementation, improved collision performance for
> string-valued keys only, and it did so at the cost of adding a new
> (private) field to every String instance.
> The changes proposed here will improve collision performance for any key
> type that implements Comparable. The alternative string-hashing
> mechanism, including the private hash32 field added to the String class,
> can then be removed.
> The principal idea is that once the number of items in a hash bucket
> grows beyond a certain threshold, that bucket will switch from using a
> linked list of entries to a balanced tree.
> We will not implement this for the Hashtable class - some legacy code
> that uses it is known to depend upon iteration order. Instead, Hashtable
> will be reverted to its state prior to the introduction of the
> alternative string-hashing implementation, and will maintain its
> historical iteration order.
> We also will not implement this technique in WeakHashMap. An attempt was
> made, but the complexity of having to account for weak keys resulted in
> an unacceptable drop in microbenchmark performance. WeakHashMap will
> also be reverted to its prior state.
> 2. http://openjdk.java.net/jeps/180
> 3. http://hg.openjdk.java.net/jdk8/jdk8/jdk/rev/43bd5ee0205e
More information about the core-libs-dev