[issue16061] performance regression in string replace for 3.3

[Antoine Pitrou]

Antoine Pitrou added the comment:

64-bit windows results:

3.3          patched

925 (-90%)   97     1 'a' 'b' 'c'
881 (-54%)   405    2 'a' 'b' 'c'
623 (-51%)   308    3 'a' 'b' 'c'
482 (-48%)   252    4 'a' 'b' 'c'
396 (-44%)   223    5 'a' 'b' 'c'
344 (-40%)   208    6 'a' 'b' 'c'
306 (-38%)   190    7 'a' 'b' 'c'
276 (-37%)   173    8 'a' 'b' 'c'
254 (-36%)   162    9 'a' 'b' 'c'
241 (-35%)   156    10 'a' 'b' 'c'
158 (-24%)   120    20 'a' 'b' 'c'
107 (-12%)   94     50 'a' 'b' 'c'
87 (+7%)     93     100 'a' 'b' 'c'
70 (+3%)     72     1000 'a' 'b' 'c'
63 (+8%)     68     10000 'a' 'b' 'c'

1332 (-92%)  106    1 '\u010a' '\u010b' '\u010c'
1137 (-60%)  459    2 '\u010a' '\u010b' '\u010c'
836 (-58%)   347    3 '\u010a' '\u010b' '\u010c'
660 (-56%)   288    4 '\u010a' '\u010b' '\u010c'
567 (-55%)   256    5 '\u010a' '\u010b' '\u010c'
503 (-51%)   245    6 '\u010a' '\u010b' '\u010c'
455 (-47%)   242    7 '\u010a' '\u010b' '\u010c'
414 (-44%)   231    8 '\u010a' '\u010b' '\u010c'
387 (-41%)   227    9 '\u010a' '\u010b' '\u010c'
365 (-35%)   237    10 '\u010a' '\u010b' '\u010c'
256 (-21%)   201    20 '\u010a' '\u010b' '\u010c'
185 (-9%)    168    50 '\u010a' '\u010b' '\u010c'
186 (-19%)   150    100 '\u010a' '\u010b' '\u010c'
137 (-1%)    136    1000 '\u010a' '\u010b' '\u010c'
131 (+3%)    135    10000 '\u010a' '\u010b' '\u010c'

1346 (-88%)  160    1 '\U0001000a' '\U0001000b' '\U0001000c'
1247 (-62%)  469    2 '\U0001000a' '\U0001000b' '\U0001000c'
965 (-64%)   352    3 '\U0001000a' '\U0001000b' '\U0001000c'
845 (-64%)   303    4 '\U0001000a' '\U0001000b' '\U0001000c'
720 (-65%)   251    5 '\U0001000a' '\U0001000b' '\U0001000c'
655 (-65%)   230    6 '\U0001000a' '\U0001000b' '\U0001000c'
604 (-58%)   256    7 '\U0001000a' '\U0001000b' '\U0001000c'
570 (-62%)   214    8 '\U0001000a' '\U0001000b' '\U0001000c'
546 (-63%)   203    9 '\U0001000a' '\U0001000b' '\U0001000c'
515 (-63%)   190    10 '\U0001000a' '\U0001000b' '\U0001000c'
404 (-61%)   157    20 '\U0001000a' '\U0001000b' '\U0001000c'
339 (-62%)   130    50 '\U0001000a' '\U0001000b' '\U0001000c'
308 (-60%)   122    100 '\U0001000a' '\U0001000b' '\U0001000c'
284 (-54%)   130    1000 '\U0001000a' '\U0001000b' '\U0001000c'
281 (-60%)   113    10000 '\U0001000a' '\U0001000b' '\U0001000c'

----------

_______________________________________
Python tracker <report at bugs.python.org>
<http://bugs.python.org/issue16061>
_______________________________________