RE: Thought process of a colonizer

["Stanley Sutton" <sutton at t-surf dot com>]

Tat's a difficult question to answer, XConq runs on a lot of machines
currently, and processor speed per se is not to useful.  I work mostly
on a SunBlade 1000, using either Sun's compiler, or gcc.  The compilers
are comprable for debug code, but for optimized compiles, the Sun
compiler easily produces code that runs in 4 or 5% of the time as the
gcc code.  So on the same machine with the same code I can get a
considerable perfomance improvement by using a different compiler.
 
But I got a 50 fold improvement in some code I'm currently developing at
work by changing one of the algorthims I was using in part of the
problem.
I guess I would always want to use the best algorthm I could, and
provide options for alternates as much as possible for people with slow
machines.  After all, even the fastest machine can slow down on a large
enough world with the maximum number of players.  :-)
 

	-----Original Message----- 
	From: Erik Jessen [mailto:ejessen@adelphia.net] 
	Sent: Sat 22-Jun-02 10:35 
	To: 
	Cc: xconq7@sources.redhat.com 
	Subject: Re: Thought process of a colonizer
	
	

	Changing the thread slightly here: what do we consider a
"reasonable"
	computer
	for this game? (i.e. the target hardware)
	
	400MHz?
	1G?
	1.5?
	2G?
	
	I ask, simply because the game may do things in awkward coding
ways, to save
	some cycles, when with faster hardware, there's more
maintainable/flexible
	ways to do it.
	
	Erik
	
	----- Original Message -----
	From: "Hans Ronne" <hronne@pp.sbbs.se>
	To: "Lincoln Peters" <peters2000@mindspring.com>
	Cc: <xconq7@sources.redhat.com>
	Sent: Saturday, June 22, 2002 2:33 AM
	Subject: Re: Thought process of a colonizer
	
	
	> >So, does it make sense to anyone?  Any questions or comments?
	>
	> I think most of what you wrote makes sense. If you want to do
some xconq
	> hacking, you can always look at good_cell_to_colonize and
check if any of
	> your ideas can be added by modifying the current code.
	>
	> There is a problem with letting all colonizers consider all
visible cells
	> every turn, though, and that is the CPU load. Even on a fast
computer,
	this
	> would slow down things to a crawl. That's why the colonizers
now look only
	> at the cell they are standing in. Possible, one could extend
this to a
	> small patch around the colonizer, but anything more than a
handful of
	cells
	> is going to spell trouble, particularly if all the stuff you
propose also
	> is added to the algorithm.
	>
	> Hans
	>
	> Hans Ronne
	>
	> hronne@pp.sbbs.se
	>
	>
	>