Skip to end of metadata
Go to start of metadata

Performance

Retained Size Approximation

Do a very fast approximation of the retained size for the whole table or table rows.

The calculation is about 10.000 times faster than computing the precise retained size and presents a lower bound for the retained size. The margin of error is less than 3 to 5 percent. That's good enough to spot the big memory chunks and now cheap enough to use it on every result to assess its impact on the heap!

New Features

Queries

Queries inspect aspects of the heap dump like collections, duplicate classes, grouping of values etc. They replace the former console and its commands - new queries have been added (please see below).

Do a full text search on queries using the Ctrl-Q.

Below the dialog you see context help.

Enter parameters to specifiy what you want to see. Of course, most queries are also available from the context menu and can be run on arbitrary object sets.

Array Fill Ratios

Print a frequency distribution of fill ratios of non-primitive arrays.

Here you see 37.251 arrays which have only up to 20 percent of their elements filled.

Collections Groups By Size

Creates a distribution histogram of given collections by their size.

Here you see 17.703 array lists which contain only 4 elements but retain more than 10MB.

Hash Entries

Extract key value pairs from hashed collections, for example the Java System Properties.

Find Strings

Find strings matching a regular expression.

Here all strings matching .java. are listed.

Group By Value

Group objects by their value, either any primitive field or the string representation.

Here Threads are grouped by their name.

Local Variables

Print local variables (Java Locals and JNI Locals) which are currently on the stack of a thread.

  • No labels