On Mon, Jan 23, 2012 at 3:12 PM, John Elrick <john.elr...@fenestra.com>wrote:
> Brain hurts... > > Richard, more information for you. I rebuilt the call stack checking > system into pure Delphi and confirmed that yy_reduce appears to be the > malloc culprit. I further created a pair of procedures which I can use to > track the yyruleno from yy_reduce as though it were a call. Below is a > representative sample of the data.... > The parser does lots of little mallocs as it builds a parse tree. So I expect it to generate a lot of malloc traffic. The question is why the parser is being called so much. Can you put a printf() or something at http://www.sqlite.org/src/artifact/1e86210d3976?ln=397 and figure out what is being parsed so excessively? > > One interesting thing I noticed is that in 3.7.5, the average <1kb malloc > request size is 181.95 bytes. In 3.7.6 that average request size drops > to 17.74 bytes. > > Both examples are running the same dataset. > > version mallocs mallocs by yy_reduce > 3.7.5 838,789 5,545 > 3.7.6 70,003,878 68,870,137 > > Major yy_reduce case consumers > > 3.7.5 > Count: 947 = 'case_112 > yy_reduce' > Count: 696 = 'case_189 > yy_reduce' > Count: 537 = 'case_173 > yy_reduce' > Count: 229 = 'case_243 > yy_reduce > case_173 > yy_reduce' > Count: 214 = 'case_37 > yy_reduce > case_173 > yy_reduce' > > > 3.7.6 > Count: 5,606,345 = 'case_112 > yy_reduce' > Count: 5,267,458 = 'case_243 > yy_reduce > case_112 > yy_reduce' > Count: 4,922,428 = 'case_37 > yy_reduce > case_112 > yy_reduce' > Count: 4,623,402 = 'case_189 > yy_reduce > case_112 > yy_reduce' > Count: 4,278,372 = 'case_8 > yy_reduce > case_112 > yy_reduce' > _______________________________________________ > sqlite-users mailing list > sqlite-users@sqlite.org > http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users > -- D. Richard Hipp d...@sqlite.org _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
