http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53263
dominik.stras...@onespin-solutions.com changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|RESOLVED |NEW
Version|4.7.0 |4.8.1
Resolution|FIXED |---
--- Comment #16 from dominik.stras...@onespin-solutions.com ---
The situation is much improved in 4.8.1, but still not usable for me in bigger
configurations:
time ./a.out 10000
real 0m0.907s
user 0m0.904s
sys 0m0.003s
time ./a.out 20000
real 0m3.713s
user 0m3.708s
sys 0m0.001s
time ./a.out 40000
real 0m13.820s
user 0m13.812s
sys 0m0.001s
time ./a.out 80000
real 0m56.759s
user 0m56.729s
sys 0m0.001s
Still O(n^2).
(I modified the test program to use argv[1] as loop counter.
W/o _GLIBCXX_DEBUG:
time ./a.out 10000
real 0m0.008s
user 0m0.006s
sys 0m0.002s
./a.out 20000
real 0m0.013s
user 0m0.008s
sys 0m0.004s
./a.out 40000
real 0m0.013s
user 0m0.013s
sys 0m0.000s
./a.out 80000
real 0m0.026s
user 0m0.024s
sys 0m0.002s
Which shows O(n) behavior.
