From: "M. Edward (Ed) Borasky" <znmeb@...>
Date: 2006-09-14T23:45:34+09:00
Subject: Re: Metaruby, BFTS, Cardinal and Rubicon - State of play?

Robert Dober wrote:
> And you benchmark algorithms written in the same language, BUT the shootout
> benchmarks *different languages* and I have looked at the algorithms they
> use just once, that was enough.
> 
> The point is use it as a tool if you find it useful, but here it is used
> for
> advocacy.
> 
> Cheers
> Robert
It's a perfectly natural desire to want to compare languages. To do that
requires micro-benchmarks that will execute in all of the languages. But
I think you're missing my point, which is that Ruby is slower than the
other dynamic languages on microbenchmarks because the implementation of
Ruby hasn't been performance-tuned to the extent that Perl, Python and
PHP have been tuned. It's not, as far as I can tell, because of anything
fundamental in the syntax or semantics of the Ruby language that
prohibits that tuning.

So rather than whine about advocacy or say "I looked at the algorithms
they use just once, that was enough", why not look at the algorithms
they use and tune the Ruby interpreter so it executes those algorithms
as efficiently as Perl, PHP and Python? Benchmarketing is a fact of life
in the "computer industry". Fortunes are made and lost because one gizmo
is faster than another gizmo on some "meaningless benchmark".

In short:

1. I don't see any fundamental reason why Ruby can't be as fast as Perl,
Python, or PHP.
2. It isn't there yet.