[Freeciv-Dev] Re: (PR#6721) A Quincuncial topology/ speed

["Marcelo Burda" <mburda@xxxxxxxxx>]

<URL: http://rt.freeciv.org/Ticket/Display.html?id=6721 >

Le mer 31/12/2003 à 16:33, Per I. Mathisen a écrit :
> <URL: http://rt.freeciv.org/Ticket/Display.html?id=6721 >

> No. You need to remove whatever code that is producing different autogames
> for gen 1/2/3/4 from your patch, and either drop it or put it in a
> different patch.
> 
> Sorry, but you are changing very crucial code here and we need to know
> what effect your changes have on game speed.
> 
>   - Per
from 4 test maked over a non patched cvs version of the 24 Dec

NDEBUG defined
patched)
real 42m33.267s
user 3m52.410s
sys 0m0.699s
-
real 3m59.244s
user 3m47.739s
sys 0m0.531s

original)
real 39m34.630s
user 3m9.200s
sys 0m0.547s

real 3m8.773s
user 3m4.714s
sys 0m0.487s

user time +23%

if NDEBUG is undefined


1p)
real 4m36.989s
user 4m26.714s
sys 0m0.856s
real 4m41.718s
user 4m29.646s
sys 0m1.044s

1o)
real 3m50.489s
user 3m39.336s
sys 0m0.820s
real 3m58.061s
user 3m42.204s

user time +21%

this code are not optimised for this particular test.
the run time up 40 sec on a play of 2400 sec!! timeout=1
this is a +1.6% over a speeded game!!

we begin the test game from a saved game to begin as equal as posible


Marcelo