Fujii Masao
masao****@gmail*****
2013年 10月 17日 (木) 16:30:38 JST
On Fri, Oct 11, 2013 at 2:41 PM, Beena Emerson <memis****@gmail*****> wrote: > Hello, > >> >> What feature calls CPBIGM() most often? Text similarity search? >> Normal text search? bigm_similarity()? Other? >> >> It could be an idea to use that in the benchmark of the patch. > > > CPBIGM is used while making bigrams of strings. > If a very long string is used in show_bigm fucntion then CPBIGM will be > called for each bigm generated. > Similarly in bigm_similarity, the CPBIGM will be called while generating > bigm for both the string argument. > > So, I think show_bigm with a long string argument can be used as the > benchmark. Thanks for the info! We could not see any performance improvement by this patch. So, I'm thinking not to apply this patch in conclusion. Thought? Regards, -- Fujii Masao
Fork