On Aug 21, 2006, at 8:25 AM, SASADA Koichi wrote:

> Good results :)
>
> Can you try without a following optimization?
> -> edit "yarv-build/vm_opts.h" and
>
> #define OPT_SPECIALISED_INSTRUCTION  1
>
> to
>
> #define OPT_SPECIALISED_INSTRUCTION  0
>
> and rebuild.
>
> This optimization boost simple methods such as Fixnum#+.  Recent YARV
> turns on this option from default.
>
> --  
> // SASADA Koichi at atdot dot net

$ ruby-yarv -v
ruby 2.0.0 (Base: Ruby 1.9.0 2006-04-08) [powerpc-darwin8.7.0]
YARVCore 0.4.1 Rev: 540 (2006-08-20) [opts: [direct threaded code]  
[inline method cache] ]

#define OPT_SPECIALISED_INSTRUCTION  1
$ ruby-yarv MatrixBenchmark.rb 256
Matrix of dimension 256 times its inverse = identity? true
29.580000   0.260000  29.840000 ( 37.981190)

#define OPT_SPECIALISED_INSTRUCTION  0
$ ruby-yarv MatrixBenchmark.rb 256
Matrix of dimension 256 times its inverse = identity? true
45.720000   0.330000  46.050000 ( 60.869100)

For reference, default ruby with OS X 10.4
$ ruby -v
ruby 1.8.2 (2004-12-25) [powerpc-darwin8.0]

$ ruby MatrixBenchmark.rb 256
Matrix of dimension 256 times its inverse = identity? true
144.020000   0.980000 145.000000 (185.637535)

--
ML: yarv-dev-en quickml.atdot.net
Info: http://www.atdot.net/~ko1/quickml