Featured White Papers

• Utilizing Fast Testing to Transform Java Development
• Do You Really Get Classloaders?
• Pragmatic Continuous Delivery with Jenkins, Nexus, and LiveRebel
• Your Next Java Web App: Less XML, No Long Restarts, Fewer Hassles
• Coding with JRebel, Java Forever Changed
• 2012 Developer Productivity Report: Java Tools, Tech, Devs, and Data

Sponsored Links

• Want to build HTML5 web applications? Check out Kendo UI JSP Server Wrappers!

  Learn More!

• Build modern HTML5 web apps using Java. Try Kendo UI Complete for JSP!

  Learn More!

• Graph databases are 1000x faster than relational.

  Learn More!

• Career opportunities at Harman- Careers That'll Rock Your Socks Off.

  Learn More!

• Developer-friendly Java PDF libraries by Qoppa Software

  Live Demos

• Stop blaming your app! Diagnose Java performance issues.

  Free tool >

• UrbanCode: Go From Continuous Integration to
  Continuous Delivery

  Learn More!

Sponsored Links

Optimize with a SATA RAID Storage Solution
Range of capacities as low as $1250 per TB. Ideal if you currently rely on servers/disks/JBODs

HotSpot: A new breed of virtual machine

Sun's next-generation dynamic compiler generates bytecodes that <em>scream</em>

By Eric Armstrong, JavaWorld.com, 03/01/98

Starting line: Warming up the engines

Lord knows, interpreters are slow. Pre-compiling Java source code into portable bytecodes saves the time needed for translating syntax, but interpreting bytecodes in the Java virtual machine (JVM) is still exceedingly slow compared to the native code produced by a compiler. Thus, Java performance is generally deemed acceptable for small applets but not for any sizable application.

A just-in-time compiler (JIT) runs many times faster than an interpreter. It makes a drastic, noticeable difference when you're running even an interactive application. Although a JIT falls short of compiled-code speeds, it greatly extends Java's applicability. But a reasonably-performing application is still compute-limited. As long as the app mostly interacts with the user or does I/O, it's fine. But if it starts doing a lot of graphic processing or extensive computing, performance rapidly drops to unacceptable levels.

Sun's new HotSpot technology, due to ship this summer (with a developer release due sooner), is a dynamic compiler -- a compiler built into the virtual machine that promises to run as fast or faster than compiled code in the majority of applications. HotSpot promises to extend Java's applicability into a wide variety of areas, from servers to mainstream desktop applications -- without sacrificing portability. Sun's Java roadmap indicates a HotSpot JVM developer's release was scheduled to be available early in the first quarter of this year; a deployable HotSpot JVM implementation is scheduled to ship this summer.

This article explores the inner workings of Sun's dynamic compiler, HotSpot, and details the kinds of applications in which HotSpot is -- and is not -- more effective than the fastest JIT. (For more information on compilers and interpreters, see the sidebar How compilers and interpreters work. To find out how a JIT improves performance, see the sidebar How a JIT works.)

They're off!: Burning rubber with dynamic compilation

HotSpot is a dynamic compiler. It combines the best features of a JIT compiler and an interpreter, combining the two in a single package. That two-in-one combination provides a number of important benefits for code performance, as you will shortly see.

For all of its apparent newness, the concept of dynamic compilation is based on research done over the past 10 years at Stanford University and the University of California, Santa Barbara (UCSB). So the technology is actually fairly stable, but it's HotSpot that has propelled dynamic compilation into the limelight. And, in addition to dynamic compilation, there are two other major time-sinks that HotSpot addresses: garbage collection and synchronization.

The cost of computing: garbage collection and synchronization

Unfortunately, in the majority of applications, optimizing Java code is not all there is to improving performance. Some of the same features that help Java developers increase productivity and write more versatile, bug-free applications also consume mass quantities of runtime compute cycles. The major cycle-stealing features are garbage collection and threads. As the pie chart below shows, together these two features take up about 40 percent of the average application's run time.