Complete.Org: Mailing Lists: Archives: freeciv-dev: April 2005:
[Freeciv-Dev] Re: (PR#12924) freeciv profiling
Home

[Freeciv-Dev] Re: (PR#12924) freeciv profiling

[Top] [All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index] [Thread Index]
To: jdorje@xxxxxxxxxxxxxxxxxxxxx
Subject: [Freeciv-Dev] Re: (PR#12924) freeciv profiling
From: "Brian Dunstan" <bdunstan149@xxxxxxxxx>
Date: Thu, 28 Apr 2005 10:22:44 -0700
Reply-to: bugs@xxxxxxxxxxx

<URL: http://bugs.freeciv.org/Ticket/Display.html?id=12924 >


--- Jason Short <jdorje@xxxxxxxxxxxxxxxxxxxxx> wrote:
> 
> <URL:
> http://bugs.freeciv.org/Ticket/Display.html?id=12924
> >
> 
> Attached is a new server profile.
> 
> I ran with the crazy-barbarians patch (PR#12919) and
> barbarians set to
> 16.  The game ran for about 7 hours.  1416 sea
> transporters, 5725 land
> units, and 1543 barbarian leaders were generated as
> part of barbarian
> summoning (almost all sea barbarians) before the
> game ended in 3550 BC.

There should be a way of limiting:

1. # of barbarian leaders
2. # of barbarian sea transports
3. # of land barbarians in a stack

> 
> The profile makes clear that the largest amount of
> time was the AI
> controlling of the barbarian units.  I think this is
> exacerbated because
> most units don't usually do anything so they just
> build up over time.  I
> don't know if the barbarian-disbanding works well to
> limit the number of
> barbarians.

Yeah that function is either broken, or, because it
never disbands more than 10% of barbarian units per
turn, maybe they are being generated so fast you still
get a buildup.

> 
> The longest time was spend in
> ai_data_movemap_recalculate (75% of
> runtime).  movemap_check_ferry and pf_next were the
> most-called
> subfunctions.
> 
> -jason
> 
> 
> 
> 

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 





[Prev in Thread] Current Thread [Next in Thread]
  • [Freeciv-Dev] Re: (PR#12924) freeciv profiling, Brian Dunstan <=