Update of patch #4933 (project freeciv):
Status: Ready For Test = Done
Open/Closed:Open = Closed
___
Reply to this item at:
Update of patch #4933 (project freeciv):
Assigned to:None = pepeto
Planned Release: = 2.5.0, 2.6.0
___
Reply to this item at:
Follow-up Comment #1, patch #4933 (project freeciv):
Testing with this applied seems to take nearly the same amount of wall-clock
time (differences varied by run, but never exceeded 1% in either direction
against the same autogame). map_pos_to_tile() sample percentage with this
applied seems to
Follow-up Comment #2, patch #4933 (project freeciv):
I have got better results when profiling. Without the patch:
10.48269.30 269.30 6034611504 0.00 0.00 normalize_map_pos
7.72467.57 198.27 1074190408 0.00 0.00 pf_normal_map_iterate
7.02647.97 180.40
Follow-up Comment #3, patch #4933 (project freeciv):
Interesting. Possibly bad luck on my part, or differences between gprof and
oprofile. If this patch produces a nearly 10% improvement in total runtime
for some initial conditions, perhaps it should also be applied to 2.5.
Follow-up Comment #4, patch #4933 (project freeciv):
I think the most of the gain is visible in late game (when the AI begins to
build fueled units). It may explain why results could be so different.
___
Reply to this item at:
URL:
http://gna.org/patch/?4933
Summary: reduce the number of calls to normalize_map_pos()
Project: Freeciv
Submitted by: pepeto
Submitted on: ven. 11 juil. 2014 22:47:32 CEST
Category: general
Priority: 5
