Post-reader changes benchmarks

Post-reader changes benchmarks

[Andy Wingo]

Hi all,

I have re-run my benchmarks against the released Guile 1.8.2, Guile HEAD
before the reader changes, and Guile HEAD after the changes.

The times in this benchmark series are lower because I changed the way I
run them: I use emacs' M-! instead of running the tests in
gnome-terminal. Probably the timings included the output of the process
also, and gnome-terminal is slow. The variation is down now.

PNG: http://wingolog.org/pub/guile-benchmarks-04-08-07.png
Data file: http://wingolog.org/pub/guile-benchmarks-04-08-07.scm

Executive summary: Ludovic is my favorite person :)

To reproduce the chart, you will need guile-charting from bzr.

Cheers,

Andy
-- 
http://wingolog.org/


_______________________________________________
Guile-devel mailing list
Guile-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/guile-devel

Re: Post-reader changes benchmarks

[Ludovic Courtès]

Hi Andy,

Andy Wingo <wingo@pobox.com> writes:

> I have re-run my benchmarks against the released Guile 1.8.2, Guile HEAD
> before the reader changes, and Guile HEAD after the changes.

Thanks!  Good to see we're doing even slightly better than SCM.

That 1.6 is significantly faster than 1.8 is most likely due to the
`pthread_getspecific ()' calls that we do for each and every allocation
in 1.8, I think.

Thanks,
Ludovic.



_______________________________________________
Guile-devel mailing list
Guile-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/guile-devel