[SOLVED] Why is C slower than Java?

[usernamer]

I'm sure it is generally, but with this example of a for loop and printing a number each time within that loop, it doesn't seem to be the case, which surprised me.

[r-senior]

I don't suggest this is a flawless benchmark, it's no more than idle amusement ...

$ make benchmark
time ./eratc
C

real	0m9.573s
user	0m9.557s
sys	0m0.008s
time ./eratcxx
C++

real	0m10.580s
user	0m10.557s
sys	0m0.012s
time ./eratp
Pascal

real	0m11.969s
user	0m11.941s
sys	0m0.016s
time java Erat
Java

real	0m11.220s
user	0m11.245s
sys	0m0.076s
time ./eratpy.py
Python (only 1/10 of the iterations)

real	0m36.453s
user	0m36.350s
sys	0m0.056s

Note that the Python version only runs 1/10 of the iterations, so you could extrapolate that to ~365s. It can be compiled of course, if anyone wants to try that (or to eliminate the sqrt). Source code attached -- you'll need Free Pascal installed to try the Pascal, or just remove it from the Makefile.

Comparison.tar.gz

[ofnuts]

C is so faster than java
java is the slowest programming language i have ever wrote program in

Then you haven't tried many...

[ofnuts]

Java usually benchmarks around 110-120% of the execution time of C and C++ for equivalent code.

I've got some Eratosthenes sieve programs somewhere that are consistent with this. Exact same algorithm in C, C++, Java, Pascal and Python. Still not a perfect benchmark but more realistic than an empty loop. I'll post them later if you want to play with them.

I have converted a prime factor extraction from C to Java and Java is more like 80% slower than C for this. But then there are different Java runtimes (using IBM JDK 1.6).

[woxuxow]

Then you haven't tried many...

If "C - C++ - C# - JAVA - Pascal - ASP.net - ASP classic" are not many
yeah, you are right.

[trent.josephsen]

I think ofnuts meant that you really don't have far to look to find a language implementation that underperforms Java (for some applications), so saying it's the slowest thing you've ever used says more about the range of your experience than the limitations of Java itself. (It probably also says something about the kind of programs you write; Java performs better with straight mathematics than heavy text manipulation, for instance.)

Python and LabVIEW come to mind as fairly popular languages that I've used that may underperform in certain areas when compared to Java.

[woxuxow]

dear trent.josephsen it dosn`t matter what our frient meant : the java, the program i written in java or ...
java is slow, maybe not the slowest but it is slower than C - C++ - Python - even C#

[KdotJ]

dear trent.josephsen it dosn`t matter what our frient meant : the java, the program i written in java or ...
java is slow, maybe not the slowest but it is slower than C - C++ - Python - even C#

Then perhaps you need to go back and redesign that Java program. I struggle to see how a python implementation of a program is faster than the same program implemented in Java.

[r-senior]

Could you give us a timed example where Python runs faster than an equivalent Java program written with a modern JDK and JVM?

[CptPicard]

I honestly would love to get some examples as well. I have written my fair share of number-crunchers and algorithm implementations in my time in all of Java, C and Python, and generally Java is almost always performant enough not to justify implementation in C because of Java's speed of development, while Python likewise tends to choke very quickly indeed...