On Tue, Dec 26, 2006 at 01:39:14AM -0500, Joel E. Denny wrote: > On Sat, 23 Dec 2006, Bob Rossi wrote: > > > If you have a lot of patches that make up the final patch, you should > > run the final patch through the benchmark script we were using. Just to > > make sure that all of the new modes don't effect the speed of the > > pure/unpure pull parsers. > > Thanks for that suggestion. What do you think of the attached uncommitted > patch? Also attached are the results that I get from etc/bench.pl after > applying all of this patch except the push.c portion. Finally, attached > are the results after applying all of the patch. Notice that, for pull > mode, switching from yacc.c to push.c has little impact. However, the > impact of switching to push mode is improved by this patch.
These timing results look great to me. The key component was obviously making sure the yacc impure/pure parser was the same as the push impure/pure parser. These results obviously show that. Thanks! Bob Rossi
