From: tmoran@acm.org
Subject: Parallel Text Corpus Processing with Ada?
Date: Sat, 10 Nov 2007 18:11:57 -0600
Date: 2007-11-10T18:11:57-06:00 [thread overview]
Message-ID: <HfKdnVJtQa1Q1avanZ2dnUVZ_u3inZ2d@comcast.com> (raw)
In-Reply-To: 1194735959.240323.38210@v2g2000hsf.googlegroups.com
> E.g., one could imagine a parallel grep.
A year or something ago I tried a parallel word counter and as I recall
it ran pleasing faster on a dual-core than a single-core CPU. It had a
subroutine that accepted a buffer load of text updated a count parameter.
So parallelizing just consisted in making more than one counter task,
then having the read task give each a buffer load in turn, and after all
was done, harvesting the several counts. Of course there was also a
bit of special handling for words that were split across buffer loads,
but for a decent size buffer that took an inconsequential amount of time.
So yes, there are lots of opportunities to utility multiple cpus using
very simple Ada tasking.
next prev parent reply other threads:[~2007-11-11 0:11 UTC|newest]
Thread overview: 17+ messages / expand[flat|nested] mbox.gz Atom feed top
2007-11-10 23:05 Parallel Text Corpus Processing with Ada? braver
2007-11-11 0:11 ` tmoran [this message]
2007-11-11 1:10 ` Georg Bauhaus
2007-11-11 8:23 ` Dmitry A. Kazakov
2007-11-11 15:54 ` Georg Bauhaus
2007-11-11 16:13 ` Georg Bauhaus
2007-11-12 13:31 ` Dmitry A. Kazakov
2007-11-12 15:07 ` Georg Bauhaus
2007-11-12 16:11 ` Dmitry A. Kazakov
2007-11-11 22:49 ` braver
2007-11-12 16:17 ` Dmitry A. Kazakov
2007-11-13 22:45 ` Simon Wright
2007-11-14 23:38 ` braver
2007-11-15 9:39 ` Ludovic Brenta
2007-11-15 11:12 ` Dmitry A. Kazakov
2007-11-15 21:11 ` Simon Wright
2007-11-17 1:05 ` Randy Brukardt
replies disabled
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox