Re: New LuaJIT benchmark results (was Re: [ANN] LuaJIT-2.0.0-beta3)

[Isaac Gouy <igouy2@...>]

Geoff Leyland <geoff_leyland <at> fastmail.fm> writes:

> 
> On 10/03/2010, at 7:51 AM, Mike Pall wrote:
> > I wrote:
> >> This is the third beta release of LuaJIT 2.0.0.
> > 
> > The cross-language benchmark shootout results have been updated
> > for beta3 (thanks Isaac!):
> > ...
> > Ok, so Java received a grace period ... until the next beta. 
> 
> Given that you said
> 
> > This release includes many fixes and performance enhancements,
> > e.g. for recursive code
> 
> what's the story with the binary-trees benchmark?  Is it a GC thing, a
recursive thing or is it a bad benchmark?
> 
> -joff makes it a bit slower, and collectgarbage("stop") at the top (for
instances that fit in memory) make
> it a bit faster, but neither's a huge difference.
> 
> On the other hand, since the shootout appears to judge on the median, it's
probably knucleotide you want to
> make faster.


The "shootout" sorts on the median.

The "shootout" shows the interquartile range because that tells us more.

The "shootout" shows box plots because that tells us more.


http://shootout.alioth.debian.org/u32/which-programming-languages-are-fastest.php?luajit=on&java=on&calc=chart


But the "shootout" can't make you read them :-)