Comparing execution speed of Java to other interpreted languages.

[Dave Rudolf]

Hey all,

Just a curiosity of mine: Does anyone know of any benchmark-type comparisons
between the JVM compared to other interpreted languages, like C#, Perl,
Python, VB, etc? I know that any such comparisons can be skewed (e.g. the
JVM is relatively slow at first until the classloader has read all the class
files, etc). I would love to see what has been found on the subject, and who
said it

Thanks,

Dave

 

[Peter Ashford]

Hey all,

Just a curiosity of mine: Does anyone know of any benchmark-type comparisons
between the JVM compared to other interpreted languages, like C#, Perl,
Python, VB, etc? I know that any such comparisons can be skewed (e.g. the
JVM is relatively slow at first until the classloader has read all the class
files, etc). I would love to see what has been found on the subject, and who
said it

Thanks,

Dave

OS News recently did a performance comparison between 9 languages, which
included java and some CLR languages

 

[Peter Ashford]

OS News recently did a performance comparison between 9 languages, which
included java and some CLR languages

heres a link: http://osnews.com/story.php?news_id=5602

 

[Roedy Green]

Just a curiosity of mine: Does anyone know of any benchmark-type comparisons
between the JVM compared to other interpreted languages, like C#, Perl,
Python, VB, etc?

Java is not an "interpreted language". It as a language that CAN be
interpreted. More often it is JITed, Hotspotted, or natively
compiled.

It depends on your political agenda. If you want to make Java look
bad, compare the rival with classic Java interpretation. If you want
to make it look good, compare with natively compiled or hotspotted.

Java tends to have longer start up times, then picks up speed. You
can arrange your benchmark to make Java look good or bad. The longer
running it is, the better Java will look.

 

[Tim Ward]

Java tends to have longer start up times, then picks up speed. You
can arrange your benchmark to make Java look good or bad. The longer
running it is, the better Java will look.

.... unless your benchmark runs long enough to invoke the garbage collector,
and unless you're counting memory freed back to the OS as a performance
indicator.

 

[Tor Iver Wilhelmsen]

Java tends to have longer start up times, then picks up speed. You
can arrange your benchmark to make Java look good or bad. The longer
running it is, the better Java will look.

And preferably you should call a method multiple times, since most
optimalization happens after a method run.

 

[Mark Thornton]

... unless your benchmark runs long enough to invoke the garbage collector,
and unless you're counting memory freed back to the OS as a performance
indicator.

Java can free memory back to the OS. The garbage collector can also out
perform traditional memory allocators depending on the nature of your
application.