Tag Archives: multithreading

java.util.Random and java.util.concurrent.ThreadLocalRandom in multithreaded environments

by Mikhail Vorontsov

Generating pseudorandom data

There are 2 types of random generators in Java: pseudorandom and secure. Pseudorandom generators are transforming a seed into a new portion of pseudorandom data based on some formula. Secure random generators are using some machine specific sources of actually random events (file/socket access, for example) to generate its data.

Secure random generators:

  • should be used only when cryptographically strong random data is required
  • are slow
  • could be waiting for external events (“stuck”) if you have requested a lot of random data (Linux /dev/random is an example of such generator)

Pseudorandom generators, on the other hand, depend only on the initial “seed” value, so you can generate the same sequence of pseudorandom events if you will provide the same seed to the algorithm. In general case, such generators are fast because they are CPU bound only (do not rely on any IO). We will review the evolution of pseudorandom generators in Java and the reasons behind these changes.


java.util.Random is available from Java 1.0. It is a thread safe class, so you may share (in theory) instances of this class between several threads and do not expect to get the same random data in 2 threads at the same time. Such thread safety is achieved via using an AtomicLong for the generator seed.

Random uses AtomicLong CAS (compare-and-set) operations for updating its seed. Despite being a light non-blocking primitive used in a lot of non-blocking algorithms, CAS behaves really poorly under the high contention. Wait for test results to see how poorly it behaves.


java.util.concurrent.ThreadLocalRandom was added in Java 7 as an attempt to overcome all performance issues associated with java.util.Random. This new class extends java.util.Random, so you may pass it to all logic expecting the old java.util.Random.

Here are main ThreadLocalRandom implementation details:

  • Internally it does not use an AtomicLong seed from Random. Instead it uses an ordinary long.
  • You can not create an instance of ThreadLocalRandom yourself, because its constructor is not public. Instead you should use its static factory ThreadLocalRandom.current(). This factory method queries internal ThreadLocal<ThreadLocalRandom>
  • It is CPU-cache aware, so it uses 8 long dummy fields for padding, thus pushing anything else out of its 64 byte L1 cache line.

All of these changes are important, which will be revealed by the following tests.

Continue reading

String.intern in Java 6, 7 and 8 – multithreaded access

by Mikhail Vorontsov

This article follows the initial String.intern in Java 6, 7 and 8 – string pooling article describing the implementation and benefits of using String.intern() method in Java 7 and 8. The original article was already getting too long, so I had to write this article in order to describe the performance characteristics of the multithreaded access to the String.intern.

The tests we will perform today are calling String.intern() from the multiple threads. They will emulate the behavior of a lot of modern server applications (for example, specialized web crawlers). For tests I will use my new workstation with Intel Xeon E5-2650 CPU (8 physical, 16 virtual cores @ 2 Ghz) and 128 Gb RAM. It will allow us to test the rather high contention scenarios. In this article we will create 8 threads in order to utilize all physical cores.

There will be 4 tests:

  1. The benchmark one – a single thread calling String.intern() using testLongLoop method from the previous article. It will show us how fast this configuration is without any contention.
  2. All 8 threads are calling String.intern() with unique values – an own prefix will be added by each thread to the interned string. This test will show us the synchronization overhead of String.intern(). It should be the theoretical worst case: it is highly unlikely that the only thing the actual application would do is calling String.intern in a loop from many threads.
  3. Initially we start the first thread interning the set of strings. After 2 seconds delay we will start a second thread interning the same set of strings. We expect that the following assumptions will be true: str.intern()==str for the first thread; str.intern()!=str for the second thread. It will allow us to prove that there are no thread local JVM string pools.
  4. All 8 threads will intern the same set of values. This scenario will be closer to the real situation – it will provide us the rather likely mix of adding strings to the JVM pool and querying the strings from it. Nevertheless, such a high read contention on the JVM string pool is still an unlikely event.

Continue reading