By Volume4
via infoworld.com
Published: Mar 28 2008 / 11:44
Java is not the slowpoke of old days and performance now matches or exceeds applications developed in C, technologists stressed Wednesday during a presentation at TheServerSide Java Symposium in Las Vegas.
Java Virtual Machines and garbage collection, for example, have improved and Java itself has come along way, the technologists argued. Early Java Virtual Machines were indeed slow, said Brian Goetz, Sun senior staff engineer. But that has changed.
"The reality is, Java is pretty fast these days," Goetz said. Java performance now measures up to C but lags behind in memory consumption, he said.
Comments
rd112681 replied ago:
First comment claims 'CPython very long time ago has JIT and it faster than Java up to 5-10'.
True or False?
andrewm replied ago:
>First comment claims 'CPython very long time ago has JIT and it faster than Java up to 5-10'.
>True or False?
err, false. cpython on its own is roughly 10-15x slower than Java. of course it has some characteristics which Java doesn't, such as reference counting (and background garbage collection) which mean that resources can often be freed up in a more timely fashion. i changed my current phd project from python to java for performance reasons.
psyco is a way of running python faster, although when i looked at it there were limitations:
http://psyco.sourceforge.net/
Basically, i think it is an I386-only JIT with a not great memory footprint.
ironpython for me runs about 3x faster than cpython (windows, same hardware) so we are perhaps getting within 5x Java speeds. However this is clearly relying on the CLR to do the heavy lifting. interestingly we got similar performance between ironpython and interpreted scheme (without JIT). hmmm...
in my Java phd work (an interactive case tool with complex behaviour derived from a formal model) i do millions of computations between interactions and allocate many, many objects. i thought i'd have to optimise it, but the lag is barely perceptible in practice. must be that hotspot thingo ;-)
I think the commenter's 5-10x was referring to speed of development. in my experience i found python to be faster for development, but nothing like that order! more like 50%. kinda reminds me of the richard gabriel experiment where the lisp bigots on being tested found that they only got 20% more productivity than a good C++ coder...
some of the comments on that infoworld site are amazingly clueless...
Andrew
rd112681 replied ago:
Many thanks for this.
Voters For This Link (14)
Voters Against This Link (2)