[ruby-talk:15837] Re: java based interpreter and regexes

> Well, Benoit's results match yours, although it looks as if his machine
> configuration is nearly identical to yours (same CPU, same VM, etc).
>
> If this true, the 1.4 VM is literally an order of magnitude faster than
the
> 1.3 VM, or else the Windows VM is a better implementation than the Linux
> VM.  I find this last option a bit odd, because IBM's 1.3 VM for Linux was
> supposed to be the fastest for non-GUI apps, although I last checked this
> several months ago.
>
Frankly I wouldn't be surprised, your results indicated that the IBM 1.3 VM
was
slightly slower than sun's while my experience on a different type of non
gui program
(it was a code generator) indicated that the IBM VM was about 3 times faster
than
sun's (and slightly slower than IBM 1.1.8 VM maybe 5%) on that particular
test
I was very disapointed by HotSpot results.




> How does the 1.4 VM perform memory-wise?  The 1.3 VM is a memory hog under
> Linux.
>

didn't check but it probably is a memory hog.  This has been a big problem
with java and
it leads to big startup times, for short programs which don't do much this
is always penalizing:
using the 1.4 VM a simle Hello world program feels sticky (I'd say .5s from
calling till the string
hits the console even when calling the program several times, didn't time
it) while ruby is
instantaneous (at least the second time it is called)
> Yeah, but we aren't.  We're talking about a single, tight loop on static
> stringes.  Although, Benoit's results on a larger itteration, which
narrows
> the gap between Java 1.4 and Ruby runtimes, illustrate the garbage
collection
> point.  At 10,000 itterations, the Java code probably wasn't getting a
chance
> to invoke GC.  On the other hand, Ruby uses GC as well... is Ruby's GC
> mechanism more efficient than Java's?
>
Can you explain this bit, it seems to me that the gap was slightly widening
and probably due
to java's startup time not being as significative, I'll recall my numbers

10000
     ruby  utime=3.665, stime=0.02, cutime=0.0, cstime=0.0
    java   1703 ms,
Also I redid the same tests but with 100000 iteration instead of 10000 and
got some coherent results:
    ruby  utime=36.662, stime=0.05, cutime=0.0, cstime=0.0
    java 15072

ruby scaled by a factor of 10, not java, like I said this is probably due to
the high startup cost of the VM

Benoit