OpenJDK 14 Has Some Performance Improvements But OpenJDK 8 Still Strong

[bug77]

Michael you should test all those benchmarks running on the GraalVM.
The hyped up people say, it performs about 4x faster during stream processing.

Better yet, try to use the native image feature. And watch it fall flat on its face

[Vorpal]

Very interesting benchmark. However, there is one thing I'd like to see from the Java benchmarks in the future: Client applications. Minecraft obviously comes to mind here, but there are other ones, like JOSM (OpenStreetMap editor). Not sure how easy it would be to get numbers out of the performance of these. For games like Minecraft you can at least measure the FPS and frame times. Also initial program load time might be of interest there. For heavily modded minecraft that is on the order of minutes.

Though admittedly, apart from those two I can't really think of any client java programs that are still in use. They probably exist, just not in domains I know about.

[tildearrow]

For games like Minecraft you can at least measure the FPS and frame times.

What about Minecraft server startup time?

[tildearrow]

Typos:

OpenJDK was running at about 88% the speed of OpenJDK 8.

Which OpenJDK?

With DaCpao's H2 benchmark

In some of the other Renasissance benchmarks,

[yariv]

One of the reasons that Java 8 is still widely used is that Java 9 broke backwards compatibility.

[TemplarGR]

Java and Performance are two things that can't be mentioned in the same context.

I hate ignorant people, especially on technical subjects around my area of expertise.... Seriously kids, STFU. Java is NOT slow. It hasn't been slow for decades. Go play your latest video game and stop commenting on things only professionals care about.

[bug77]

One of the reasons that Java 8 is still widely used is that Java 9 broke backwards compatibility.

No, it's because Java 9 (and 10) were not LTS releases, they only got 6 months of support. Nobody's gonna bother putting that in production. Only Java 11 has been production ready since Java 8 and, as far as I can tell, people don't have a problem using that. You're still better ignoring the module system, but other than that it doesn't raise any particular problems.
And Java 9 didn not break any backwards compatibility. Java 9 simply put a mechanism in place to keep internal classes from being outside accessible. Several libraries were making use of those classes (despite there being a warning from day 1 that that may create compatibility issues). Module system was postponed from Java 7 to Java 8 and eventually Java 9 largely to give those libraries time to adapt.

[vb_linux]

Do OpenCL or OpenMP or OpenACC or OneAPI or SYCL or CUDA support JAVA?!
if No, Why?
Does JAVA compiler have the ability to generate optimised binary for the host machine to use its full power?

First question makes no sense. Why certain APIs are for certain language only ?

Java compiler for now no ? But Java VM yes. Advantage is you only have to optimize once. And mostly it will be optimized by people who know how.

[Michael_S]

Do OpenCL or OpenMP or OpenACC or OneAPI or SYCL or CUDA support JAVA?!

if No, Why?

You can use them all with JNI. But not all high performance is embarrassingly parallel processing. You want to say, "Java and parallelized program high performance should not be in the same sentence", maybe you have an argument.

Does JAVA compiler have the ability to generate optimised binary for the host machine to use its full power?

You didn't answer those benchmarks. Does it need an optimized binary when it beats C++ in 3 of 6 benchmarks? Does it need an optimized binary when only one other programming language, Rust, consistently beats it?

By your logic, C sucks for performance.

[onlyLinuxLuvUBack]

What about this benchmark:

lang	production use license cost
c++	0 rupee forever
rust	0 rupee forever
java	10 billion rupees , and pray oracle doesn't change the agreement.

rust and c are lookin better to me