Re: WDSC on Mac -- WDSCI-L

Aaron Bartell skrev den 06-12-2007 17:38:

Before bashing Java on its speed, you need to measure ;-) There is

nothing preventing a Java implementation from having reasonable performance.

I am not entirely following you here, or rather, don't want to drawn the
conclusion spinning in my head.

While we all know Java has gained some excellent performance in recent
releases (I remember hearing of some big ones in 1.5) I don't think it can
hold a candle to a same process written in a lower level language, correct?
Sure, you can beef up your hardware to appease the Java gods running apps on
your machine, but wouldn't the equivalent piece of C/C++ more often than not
run much faster?

Very short: Modern Java implementations is comparable with C and C++ in terms of speed.

Jacob Marner researched the subject in 2002 and wrote a report named "Evaluating Java for Game Development" (Google knows where to find the PDF) where he discuss the state back then (Java 1.4 with HotSpot is included).

His conclusion was:
==
"In contrary to popular belief Java applications are in fact not much slower than C++ applications when they have been tuned for performance. A rough estimate based on the various benchmarks would say that tweaked Java code is a little slower than C++, typically 20-50% on the average, but this is hard to tell for certain because of the large variations in the speed seen in the benchmarks. The slowdown is less in 3D applications where performance mostly depends on the performance of the 3D hardware and not on the programming language used.

For untweaked code Java is much slower than C++, often a factor of three and four. This makes it vital to tweak the performance critical sections of Java code.

Java increases the overall productivity of a software project with about 30% and the productivity of the code phase with about 65%. This is quite a significant increase."
==

I believe that many of the issues that needs to be tweaked has been improved in the last 5 years, as they were issues in HotSpot.

There is still the warmup time which should be dealt with similar to the way that the classic OS/400 JVM do - namely by storing the result for the next runs.

Memory usage is quite another chapter - perhaps this will shrink dramatically if Hotspot is only needed once when changing and not for additional runs.

When all the above is said and done, it makes MY life much easier that I do not explicitly have to deal with memory. I just create objects when needed, and set the variables to null when done with them, and they magically get cleaned.

Only problem is that it is not trivial to port good C to good Java.

Couldn't they just port the C? I would guess that would be much less work,
but I could be wrong.

They probably could, but that turns the problem into maintaining some C code on numerous platforms instead of just the Java code.

Given the choice I know what I prefer. Besides - IBM probably has a C->Java compiler hidden in the labs somewhere anyway :)