Friday, November 2, 2007

How much cores are using WF solutions

Tim Bray published WF XI: Results and I would like to know how much of all these CPU cores each solution uses. Than I compute this table:
Name            Language    Elapsed     User        System  Parallel CPU work
-----------------------------------------------------------------------------
clv5            Gawk        46.73       40.63       6.1         1
tbray5          Erlang      01:04.32    35:33.35    00:45.84    33.88
wfinder1_1      Erlang      6.46        34.07       8.02        6.52
report-counts   Ruby        01:43.71    01:27.11    00:16.60    1
?               Groovy      02:21.83    02:22.97    00:19.95    1.15
wf_p            Ruby        50.16       37.58       12.5        1
wf-2            Python      41.04       34.8        6.24        1
wf-6(2)         Python      16.91       3.62        1.86        0.32
wf-6(4)         Python      9.08        3.66        1.89        0.61
wf-6(8)         Python      5.81        *           *           *
wf-6(16)        Python      4.38        *           *           *
wf              OCaml       49.69       41.94       7.75        1
widefinder      PHP         01:29.81    01:23.10    00:06.71    1
wf_pichi3       Erlang      8.28        51.98       9.38        7.41
tbray5          Erlang      00:20.74    03:51.33    00:08:00    34.3
Nice, that erlang implementations can use cores well, but in this task is not so much good generally. Erlang manages parallel processes well, but those processes can be better written in other languages and used as ports. Especially when this task is string operations on big amount of data.