|
From: | Gunnar Farneback |
Subject: | Re: [gnugo-devel] engine/influence.c (and DFA) |
Date: | Tue, 03 Sep 2002 20:07:45 +0200 |
User-agent: | EMH/1.14.1 SEMI/1.14.3 (Ushinoya) FLIM/1.14.2 (Yagi-Nishiguchi) APEL/10.3 Emacs/20.7 (sparc-sun-solaris2.7) (with unibyte mode) |
Dave wrote: > I was just having a look at influence.c, since that is where the > profile shows the most time being spent. On what platform and what exactly did you run? I just did a profile on an AMD Athlon running neurogo.tst and then the influence code doesn't take much time. There the pattern matching is in the top. % cumulative self self total time seconds seconds calls s/call s/call name 16.05 37.36 37.36 142878688 0.00 0.00 scan_for_patterns 6.27 51.96 14.60 104763600 0.00 0.00 check_pattern_light 5.08 63.79 11.83 179092 0.00 0.00 compute_primary_domains 3.85 72.76 8.97 64742873 0.00 0.00 fastlib 3.05 79.87 7.11 30039637 0.00 0.00 do_play_move 2.94 86.72 6.85 97175 0.00 0.00 do_push_owl 2.39 92.29 5.57 31788626 0.00 0.00 hashtable_search 2.29 97.61 5.32 18859166 0.00 0.00 order_moves 2.21 102.75 5.14 117157416 0.00 0.00 neighbor_of_string 2.14 107.73 4.98 17859836 0.00 0.00 do_dfa_matchpat 2.00 112.39 4.66 17704809 0.00 0.00 assimilate_string 1.89 116.79 4.40 38474122 0.00 0.00 incremental_order_moves 1.80 120.99 4.20 30037001 0.00 0.00 undo_trymove 1.64 124.81 3.82 69735447 0.00 0.00 approxlib 1.46 128.20 3.39 157757 0.00 0.00 accumulate_influence [...] What we see here is probably that the neurogo test suite is rather owl intensive. For a more representative profiling one should probably replay a couple of entire games against varied opponents (self play just doesn't suffice). The most straightforward way to do that would be to have a GTP file simulating replay of a selection of games. /Gunnar
[Prev in Thread] | Current Thread | [Next in Thread] |