From: Ken H. <kdh...@an...> - 2004-09-13 20:51:48
|
That's a great speedup. Is there any noticable deterioration in output quality? -----Original Message----- From: klu...@li... [mailto:klu...@li...] On Behalf Of Chris Thorp Sent: Monday, September 13, 2004 4:48 PM To: klu...@li... Subject: [Klustakwik-develop] Newest feature addition to the KlustaKwik engine Everyone using KlustaKwik for clustering large datasets will enjoy the new k-means preprocessing step. The new k-means preprocessing step gave an approximately 5x improvement in processing throughput. A dataset which took 60m6s without k-means (and, therefore, random initial cluster assignments) was processed with the k-means step in 11m58s. K-means is enabled with the new command line flag "-doKMeans 1". The default is for k-means to not be run, which means an implied "-doKMeans 0". The code for this new engine feature is checked in to the sf.net CVS repository. If there are any comments, questions, or concerns, please feel free to email me: th...@sp.... Benchmarking computer specs: Athlon 2800+ with 1GB RAM running WinXP SP1 -Chris Thorp ------------------------------------------------------- This SF.Net email is sponsored by: YOU BE THE JUDGE. Be one of 170 Project Admins to receive an Apple iPod Mini FREE for your judgement on who ports your project to Linux PPC the best. Sponsored by IBM. Deadline: Sept. 13. Go here: http://sf.net/ppc_contest.php _______________________________________________ Klustakwik-develop mailing list Klu...@li... https://lists.sourceforge.net/lists/listinfo/klustakwik-develop |