On Oct 8, 6:54 pm, "Simon Ask Ulsnes" <[EMAIL PROTECTED]> wrote:
>
> Can you provide the source code for your benchmarks?
Absolutely not! :-))))))
I'm a worldly-wise old owl who's been in too many decades of language
and benchmarking wars, so I'm not going there, hehe.
These numbers are only meaningful to me, on specific x86 hardware, for
the specific algorithm used and no other, and I only use the tests to
see how the various implementations evolve over time, and to give a
general idea of speed. They're dependent on numerous things, and are
only performed fairly to my own satisfaction, nobody else's.
Benchmarks they are not. I am language-agnostic though, so I don't
mind how the rankings turn out, may the fastest one win. I like
efficiency.
In any case, you should never believe anyone else's benchmarks nor
trust their methods, do your own measurements. :-)
That said, how many possible ways are there to write recursive
floating-point factorials? Not many. :-) Just aim for consistency
and you'll be OK.
This was really just intended as a "Well done folks!" to the V8 devs,
as it's cheaper than sending them a beer. ;-)
Morgaine.
--~--~---------~--~----~------------~-------~--~----~
v8-users mailing list
[email protected]
http://groups.google.com/group/v8-users
-~----------~----~----~----~------~----~------~--~---
