Many people complain that Java is far too slow to use in most applications that are traditionally written with C/C++. Well, this is probably never going to change unless Java allows itself to be completely compiled instead of bytecode, but IBM is trying to make the state that Java's in better. Now, IBM has released a Java 6 SDK that has increased performance over Sun's Java SDK.

It's amazing what open source can work. Although the idea of it making money seems ridiculous, just look at some of the more-successful applications created with this. Mozilla Firefox for example, has brought in billions of dollars. And so Java will likely do so, even though it's free.

This new version of the Java SDK from IBM includes data sharing between virtual machines, which in theory reduces the amount of memory needed because applications can share the data. Revamped documentation will allow users to get better information.

Quoting from IBM's website, here are the platforms supported:

  • Linux® on x86
  • Linux® on PowerPC® 32-bit
  • Linux® on PowerPC® 64-bit
  • Linux® on AMD64/EM64T
  • IBM AIX® on PowerPC 32-bit
  • IBM AIX® on PowerPC 64-bit

I wonder why Windows and Mac aren't listed ;).

Java is getting better day by day. Maybe someday I'll actually start using it, but I'm certainly very interested. I'm hopeful that a compiler that compiles Java into machine code will be created eventually, which shouldn't be too hard, given that fact that Java is open-source.

Way to go, Sun!

Hi,

I'm sorry, but as far as I know IBM has been releasing optimized JDKs for many years through a special agreement with Sun.

I still believe that open sourcing java could be good for Sun.

The only real question is: "Is it too late?"

shuri

The "complaints" that "Java is slow" have been inaccurate since the release of the 1.4 JVMs ages ago, but will never die because there are too many people who have vested interest in seeing the end of Java (not the least of which is a good portion of the open source zealots).
Strangely those same people do NOT have a mantra that "Ruby is slow" or "Python is slow" despite both being an order of magnitude slower than is Java (ironically it has been found that the latest version of JRuby is far faster than its C based counterpart).

It's quite possible to write extremely performant applications in Java, applications that in many cases outperform their counterparts written in C or especially C++.
But it's also quite possible to write tests that are rigged to show C++ as heavily outperforming Java.

I've seen some tests that were so clearly rigged it was silly that anyone fell for them at all.
Make it run for a very short period (so the JVM startup time makes up a significant portion of the runtime, a time cost native compiled applications don't have), use highly optimised math modules written in Assembler for the C++ program, but not the equivalent modules for the Java program, then deliberately select to "test" only those parts of the language where you know Java is relatively slow (goniometric functions mostly).
No wonder Java comes out low in such rigged "tests", but that's exactly the type of "test" that these people do.
And they do it deliberately because of their anti-Java agenda, an agenda that for many of them will not change now that they can mess around with and destroy the language.

In fact that's what I fear will happen, Java haters releasing deliberately broken versions and calling them Java, releasing them in such a way that the average user won't notice the difference with the official release but encounter problems whenever he tries to run a Java application or applet, and blames Sun (and Java in general) for that failure.

Comments
nice comment :)

In my experience most people who go out of their way to say a language is bad haven't used it much. I was slagging off PHP on some forums a few weeks ago. Have I ever used it? No. But seriously, that a languages implementation is not as fast as it could be of course doesn't mean the language itself is bad. The Unununium team choose to build an OS with Python. Their attitude was, sooner or later someone will come out with a faster implementation. I expect this will be true for Java.

Steven.

Java is plenty fast. It's people who have an anti-Java agenda that say otherwise, using data a decade old (when it was true) and exquisitely crafted "tests" to make their point.

In practice Java is quite fast. If it weren't we wouldn't need to build in a deliberate delay in our core Java applications in order to prevent the data from flowing faster than the hardware can handle...

The article starter has earned a lot of community kudos, and such articles offer a bounty for quality replies.