From: ts Date: 2002-12-29T20:21:20+09:00 Subject: Re: [1.8] speed >>>>> "n" == nobu nokada writes: n> I guess this issue and [ruby-talk:59662] concern with following n> change. GC only knows "big buffers" were allocated, but n> doesn't know whether they're refered by live objects, so the n> live memories amount would be estimated too small than the n> allocated. Well, I have also changed Data_Make_Struct to be sure that it call also malloc() rather than ruby_xmalloc() and it seems to have a problem with the GC. It is called more often 1.8.0 11.32 0.13 0.06 379092 0.00 0.00 rb_gc_mark 7.55 0.23 0.04 22 1.82 4.26 gc_sweep 3.77 0.31 0.02 204403 0.00 0.00 rb_gc_mark_children 1.89 0.40 0.01 177487 0.00 0.00 mark_source_filename 1.6.8 2.86 0.22 0.01 8681 0.00 0.00 rb_gc_mark 2.86 0.34 0.01 2 5.00 5.00 gc_sweep 0.00 0.35 0.00 15069 0.00 0.00 mark_source_filename Guy Decoux