On Sun, Jan 26, 2003 at 05:02:28PM +0200, Jarkko Hietaniemi wrote: > Based on my quick&dirty testing (the eventpos.pl and the input file xaa > from the bug id #20218), on Tru64, with cc -O4, this claws us back almost > to the 5.6.1 speeds (execution speed 5.6.1 = 100, maint branch after this > patch = 117). I'll do some further profiling to see where's the rest going. > (Unfortunately it seems that reverting the second-most-slowing-down > change #14920 that did cause a 40% drop (IIRC) back in between 5.7.2 > and 5.7.3 doesn't give us back much, if any, of the speed.) Well, the main bug, now fixed, was that under -d and use integer, DB::DB was being called for each and every statement. I guess that #14920, Ilya's "autoloaded DESTROY bugfix", was just making DB::DB run a bit slower. Since we're no longer calling DB::DB at all, it's probably irrelevant (to this issue, anyway). Surely any remaining differences are just general 5.6 -> 5.8 slowdown, which we're all familiar with? -- This email is confidential, and now that you have read it you are legally obliged to shoot yourself. Or shoot a lawyer, if you prefer. If you have received this email in error, place it in its original wrapping and return for a full refund. By opening this email, you accept that Elvis lives.Thread Previous | Thread Next