[erlang-questions] widefinder update

Hynek Vychodil <>
Sun Oct 28 16:21:57 CET 2007


>
> I wrote some code to try parallel splitting, but it is still with
> single dispatcher, because you must have single reader. Sequential
> reading is about twenty times faster than random and you must keep
> information about block order. Let you see chunk_reader and
> nlt_reader.
>
> Crazy think that chunk_reader is slower when compiled with native
> option. I don't know why.
>
> -- Hynek (Pichi) Vychodil
>
>
And complete with this file_map_reduce.erl and wf_pichi1.erl code.
In wf_pichi1.erl is some my very old code but there is also much
better Anders finder from wfbm4_ets1 and works well. This is code with
one ets which can have bottle neck with many CPUs, of course. I know
it.

-- Hynek (Pichi) Vychodil
-------------- next part --------------
A non-text attachment was scrubbed...
Name: file_map_reduce.erl
Type: application/octet-stream
Size: 1931 bytes
Desc: not available
URL: <http://erlang.org/pipermail/erlang-questions/attachments/20071028/207cb882/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: wf_pichi1.erl
Type: application/octet-stream
Size: 5599 bytes
Desc: not available
URL: <http://erlang.org/pipermail/erlang-questions/attachments/20071028/207cb882/attachment-0001.obj>


More information about the erlang-questions mailing list