public inbox for gcc-bugs@sourceware.org help / color / mirror / Atom feed
From: "bonzini at gnu dot org" <gcc-bugzilla@gcc.gnu.org> To: gcc-bugs@gcc.gnu.org Subject: [Bug tree-optimization/26854] Inordinate compile times on large routines Date: Sat, 04 Jul 2015 09:59:00 -0000 [thread overview] Message-ID: <bug-26854-4-mGKLe8me3E@http.gcc.gnu.org/bugzilla/> (raw) In-Reply-To: <bug-26854-4@http.gcc.gnu.org/bugzilla/> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=26854 --- Comment #130 from Paolo Bonzini <bonzini at gnu dot org> --- A late update... all.i: with GCC 4.8.3 on a Xeon E5 v3 time is taken mostly by alias stmt walking alias stmt walking : 272.52 (65%) (-O2) alias stmt walking : 116.06 (67%) (-O1) Requred memory is 700 MB. With GCC 5.1, but on a Core i7 Ivy Bridge laptop so times are not comparable, time is also taken mostly by alias stmt walking: alias stmt walking : 604.43 (54%) usr (-O1) and memory usage is also around 700 MB. Brad was using -fschedule-insns too, and it's pretty expensive: scheduling : 430.61 (38%) (-O1 -fschedule-insns, 5.1) scheduling : 122.68 (41%) (-O1 -fschedule-insns, 4.8.3) It also brings the top memory usage to 1 GB. --------------------- compile.i: with GCC 4.8.3 time is taken mostly by scheduling and some tree passes (-O1): alias stmt walking : 206.77 (36%) tree CFG cleanup : 42.66 ( 7%) usr 0.02 ( 0%) sys 42.65 ( 7%) wall 1108 kB ( 0%) ggc dominator optimization : 39.98 ( 7%) usr 0.04 ( 1%) sys 39.97 ( 7%) wall 23123 kB ( 3%) ggc Required memory is around 1 GB. I haven't tested 5.1 but the alias stmt walking seems to be a common feature of the Gambit testcases. In both cases, memory usage is at least under control. However, total compile time has regressed since the previous report in comment 115. As expected walk_aliased_vdefs has a high % of time spent, but the time is spent in bitmap operations rather than the callbacks! This is because the callback is the trivial mark_modified function. The guilty walk_aliased_vdefs invocation is parm_ref_data_preserved_p, invoked from ipa_load_from_parm_agg and in turn from unmodified_parm_or_parm_agg_item. Memoization via parms_ainfo seems like a plan, I'm opening a separate bug.
next prev parent reply other threads:[~2015-07-04 9:59 UTC|newest] Thread overview: 96+ messages / expand[flat|nested] mbox.gz Atom feed top [not found] <bug-26854-4@http.gcc.gnu.org/bugzilla/> 2011-01-18 14:42 ` rguenth at gcc dot gnu.org 2011-01-18 15:08 ` hubicka at gcc dot gnu.org 2011-01-18 15:15 ` dberlin at gcc dot gnu.org 2011-01-18 15:51 ` hubicka at ucw dot cz 2011-01-18 15:53 ` dberlin at gcc dot gnu.org 2011-01-24 23:03 ` jsm28 at gcc dot gnu.org 2011-01-24 23:47 ` ian at airs dot com 2011-02-02 17:54 ` dnovillo at gcc dot gnu.org 2015-07-04 9:59 ` bonzini at gnu dot org [this message] 2021-12-17 7:40 ` pinskia at gcc dot gnu.org 2021-12-17 20:21 ` lucier at math dot purdue.edu 2021-12-17 20:24 ` lucier at math dot purdue.edu 2022-01-03 10:48 ` rguenth at gcc dot gnu.org 2023-02-06 12:14 ` rguenth at gcc dot gnu.org 2023-02-07 13:39 ` cvs-commit at gcc dot gnu.org 2023-02-07 22:23 ` lucier at math dot purdue.edu 2023-02-08 21:53 ` lucier at math dot purdue.edu 2023-02-09 7:24 ` cvs-commit at gcc dot gnu.org 2023-02-09 7:54 ` rguenth at gcc dot gnu.org 2023-02-09 8:47 ` rguenth at gcc dot gnu.org 2023-02-13 14:55 ` rguenth at gcc dot gnu.org 2023-02-15 14:05 ` cvs-commit at gcc dot gnu.org 2006-03-24 20:25 [Bug c/26854] New: " lucier at math dot purdue dot edu 2006-03-25 16:21 ` [Bug tree-optimization/26854] " rguenth at gcc dot gnu dot org 2006-03-25 22:22 ` lucier at math dot purdue dot edu 2006-04-19 6:43 ` law at redhat dot com 2006-04-19 15:32 ` law at redhat dot com 2006-04-19 22:34 ` law at gcc dot gnu dot org 2006-04-20 3:18 ` lucier at math dot purdue dot edu 2006-04-20 3:28 ` law at redhat dot com 2006-04-20 3:39 ` lucier at math dot purdue dot edu 2006-04-20 16:13 ` law at gcc dot gnu dot org 2006-04-20 16:17 ` law at redhat dot com 2006-04-20 16:21 ` dberlin at gcc dot gnu dot org 2006-04-26 18:59 ` amacleod at redhat dot com 2006-04-27 2:29 ` amacleod at redhat dot com 2006-04-27 2:30 ` amacleod at redhat dot com 2006-04-27 20:22 ` amacleod at gcc dot gnu dot org 2006-11-30 4:36 ` lucier at math dot purdue dot edu 2006-11-30 4:54 ` dberlin at dberlin dot org 2006-12-07 17:33 ` lucier at math dot purdue dot edu 2006-12-07 17:54 ` dberlin at dberlin dot org 2006-12-07 17:54 ` dberlin at dberlin dot org 2006-12-07 21:51 ` lucier at math dot purdue dot edu 2006-12-08 1:24 ` lucier at math dot purdue dot edu 2006-12-11 6:28 ` lucier at math dot purdue dot edu 2007-01-10 18:49 ` lucier at math dot purdue dot edu 2007-01-10 19:48 ` amacleod at redhat dot com 2007-11-14 9:56 ` steven at gcc dot gnu dot org 2007-11-14 10:07 ` rguenth at gcc dot gnu dot org 2007-11-14 12:04 ` steven at gcc dot gnu dot org 2007-11-14 12:40 ` lucier at math dot purdue dot edu 2007-11-14 13:14 ` rguenth at gcc dot gnu dot org 2007-11-14 13:38 ` lucier at math dot purdue dot edu 2007-11-14 14:08 ` rguenth at gcc dot gnu dot org 2007-11-14 16:57 ` dberlin at dberlin dot org 2007-11-14 19:05 ` lucier at math dot purdue dot edu 2007-11-14 19:06 ` lucier at math dot purdue dot edu 2007-12-19 21:49 ` lucier at math dot purdue dot edu 2007-12-19 22:13 ` steven at gcc dot gnu dot org 2007-12-19 23:31 ` lucier at math dot purdue dot edu 2007-12-20 0:02 ` steven at gcc dot gnu dot org 2007-12-20 2:29 ` lucier at math dot purdue dot edu 2007-12-20 3:07 ` zadeck at naturalbridge dot com 2007-12-20 3:52 ` lucier at math dot purdue dot edu 2007-12-20 14:49 ` zadeck at naturalbridge dot com 2007-12-20 15:08 ` stevenb dot gcc at gmail dot com 2007-12-20 15:31 ` zadeck at naturalbridge dot com 2007-12-20 16:06 ` zadeck at naturalbridge dot com 2007-12-20 16:11 ` lucier at math dot purdue dot edu 2007-12-20 17:28 ` zadeck at naturalbridge dot com 2007-12-20 18:56 ` lucier at math dot purdue dot edu 2008-01-17 21:41 ` zadeck at naturalbridge dot com 2008-01-17 21:55 ` rguenth at gcc dot gnu dot org 2008-01-17 22:07 ` zadeck at naturalbridge dot com 2008-01-17 22:20 ` lucier at math dot purdue dot edu 2008-01-17 22:54 ` lucier at math dot purdue dot edu 2008-01-17 23:58 ` zadeck at naturalbridge dot com 2008-01-18 1:46 ` lucier at math dot purdue dot edu 2008-01-18 2:18 ` zadeck at naturalbridge dot com 2008-01-19 0:51 ` zadeck at gcc dot gnu dot org 2008-01-20 2:21 ` zadeck at gcc dot gnu dot org 2008-01-22 13:59 ` zadeck at gcc dot gnu dot org 2008-01-23 15:45 ` lucier at math dot purdue dot edu 2008-05-15 2:49 ` lucier at math dot purdue dot edu 2008-05-15 2:51 ` lucier at math dot purdue dot edu 2008-05-15 2:52 ` lucier at math dot purdue dot edu 2008-05-15 5:59 ` steven at gcc dot gnu dot org 2008-05-19 2:00 ` vmakarov at redhat dot com 2008-05-19 2:04 ` vmakarov at gcc dot gnu dot org 2008-05-19 2:09 ` vmakarov at redhat dot com 2008-05-19 17:55 ` lucier at math dot purdue dot edu 2008-07-10 17:37 ` lucier at math dot purdue dot edu 2008-07-10 17:45 ` lucier at math dot purdue dot edu 2008-07-10 19:38 ` rguenth at gcc dot gnu dot org 2008-07-10 19:40 ` zadeck at naturalbridge dot com 2008-09-10 13:40 ` lucier at math dot purdue dot edu
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=bug-26854-4-mGKLe8me3E@http.gcc.gnu.org/bugzilla/ \ --to=gcc-bugzilla@gcc.gnu.org \ --cc=gcc-bugs@gcc.gnu.org \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).