------- Comment #17 from steven at gcc dot gnu dot org  2010-01-28 23:03 -------
Introduced by http://gcc.gnu.org/ml/gcc-cvs/2009-09/msg00094.html.

Needs 1.3GB and 68s to compile on an 800MHz AMD64 x86_64 X
armv5tejl-unknown-linux-gnueabi compiler. The compiler spends 87% of its time
in the sched1 pass, but almost no time at all in the sched2 pass (less than
1%).

Top of flat gprof profile (I tried callgrind but it still had not finished
after one hour):

  %   cumulative   self              self     total
 time   seconds   seconds    calls   s/call   s/call  name
  8.02      3.62     3.62 69054802     0.00     0.00  detach_dep_link
  5.27      5.99     2.38 69054802     0.00     0.00  attach_dep_link
  4.99      8.24     2.25 51417075     0.00     0.00  sd_iterator_cond
  4.18     10.12     1.89 34819238     0.00     0.00  get_back_and_forw_lists
  3.53     11.71     1.59 97495355     0.00     0.00  sd_iterator_cond
  3.44     13.27     1.55    21666     0.00     0.00  fix_tick_ready
  3.20     14.71     1.45    21661     0.00     0.00  priority
  3.16     16.14     1.43 17117782     0.00     0.00  sd_resolve_dep
  3.13     17.55     1.41 43578690     0.00     0.00 
sched_get_condition_with_rev
  3.10     18.95     1.40 17139633     0.00     0.00  try_ready
  2.73     20.18     1.23   635304     0.00     0.00  add_dependence_list
  2.51     21.30     1.13 69054802     0.00     0.00  remove_from_deps_list
  2.48     22.43     1.12 261206816     0.00     0.00  VEC_int_base_index
  2.38     23.50     1.07 138109604     0.00     0.00  depl_on_debug_p
  2.26     24.52     1.02 25036374     0.00     0.00 
sched_insns_conditions_mutex_p
  2.13     25.48     0.96 34288021     0.00     0.00  dep_cost_1
  1.71     26.25     0.77                             _fini
  1.67     27.00     0.76 103473085     0.00     0.00  bitmap_find_bit
  1.52     27.69     0.69 177031345     0.00     0.00 
VEC_haifa_deps_insn_data_def_base_index
  1.51     28.37     0.68 78460714     0.00     0.00  bitmap_bit_p
  1.49     29.04     0.67 17409619     0.00     0.00  create_dep_node
  1.39     29.66     0.62 24364740     0.00     0.00  ask_dependency_caches
  1.24     30.23     0.56 207656387     0.00     0.00 
VEC_haifa_insn_data_def_base_index
  1.20     30.77     0.54   863847     0.00     0.00 
sd_find_dep_between_no_cache
  1.19     31.30     0.54 25052073     0.00     0.00  add_or_update_dep_1
  1.15     31.82     0.52 69054802     0.00     0.00  add_to_deps_list
  1.00     32.27     0.45 38573532     0.00     0.00  sd_next_list
  0.93     32.69     0.42 19177050     0.00     0.00  pool_alloc
  0.89     33.09     0.40    21661     0.00     0.00  schedule_insn
  0.88     33.48     0.40 34618877     0.00     0.00  alloc_INSN_LIST
  0.86     33.88     0.39 42461692     0.00     0.00  check_dep
  0.84     34.26     0.38 17409619     0.00     0.00  sd_add_dep
  0.80     34.62     0.36 17117782     0.00     0.00  contributes_to_priority_p
  0.78     34.97     0.35 79199812     0.00     0.00  sd_iterator_next
  0.69     35.27     0.31 43670647     0.00     0.00  any_condjump_p
  0.63     35.56     0.28 25064307     0.00     0.00  maybe_add_or_update_dep_1
  0.60     35.83     0.27 34288895     0.00     0.00  insn_cost
  0.55     36.08     0.25 25061789     0.00     0.00  haifa_note_dep
  0.55     36.33     0.25 16991202     0.00     0.00  set_dependency_caches
  0.51     36.56     0.23    90447     0.00     0.00  validate_value_data
  0.50     36.78     0.23 34465340     0.00     0.00  free_INSN_LIST_node
  0.49     37.01     0.22 34300880     0.00     0.00  sd_lists_empty_p

It looks like something causes dependency calculation to explode...


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=41399

Reply via email to