On Mon, Jan 16, 2012 at 5:31 PM, John Elrick <john.elr...@fenestra.com>wrote:
> > I don't know if the calls to _realloc will help also, but I've added > tracing for those also. My updated output: > Cumulative _mallocs by size > <= 1kb: 1,118,058,528 bytes (63,640,300 > count) > 1kb to 4kb: 3,836,490,296 bytes (6,197,060 count) > 4bk to 8kb: 47,800,672 bytes (11,571 count) > 8bk to 16kb: 597,016 bytes (46 count) > 16bk to 32kb: 224,272,704 bytes (10,136 count) > 32bk to 64kb: 64,008 bytes (1 count)Cumulative > _reallocs by size > <= 1kb: 1,117,392,608 bytes (63,637,756 > count) > 1kb to 4kb: 3,836,372,120 bytes (6,196,954 count) > 4bk to 8kb: 30,800 bytes (6 count) > 8bk to 16kb: 0 bytes (0 count) > This is very curious and perhaps a useful clue. SQLite does call realloc(), but not that often. And yet here we see realloc() being called almost as much as malloc(). I'll investigate further on this end. In the meantime, are you *certain* that the numbers above are correct? Please register as a malloc() any call to realloc(A,B) where A is NULL. With that change, are the numbers above still the same? -- D. Richard Hipp d...@sqlite.org _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
