glob2-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[glob2-devel] smart time-dividing computation of resource gradients idea


From: Nuage
Subject: [glob2-devel] smart time-dividing computation of resource gradients idea (new)
Date: Fri, 25 Nov 2005 21:57:36 +0100
User-agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.12) Gecko/20051102

* resource-index based:

One idea I had is to spread the resource update other the time more carefully.
We know that some resources are much more critical than other, and used much
more often (corn). Oppositely, some don't move at all (stone). Obviously, we can
order importance of updates:
1) corn
2) wood
3) algae
4) stone
5) papyrus
Currently, they are updated one after the other. We could do something more
clever, and make less updates. Overall the quality would be the same or higher,
but with less computation. For instance:
 1 - corn
 2 - x
 3 - wood
 4 - x
 5 - algae
 6 - x
 7 - corn
 8 - x
 9 - wood
10 - x
11 - stone
12 - x
13- papyrus
14 - x
Here corn and wood are updated all 6 or 8 steps. The quality is 30% lower, but
we save 50% processing time. Well, maybe this is not that great.

* swim-index based:

In the same idea, remember that we have each resource gradient for both swimming
and non-swimming units. Most probably, a team will have most of his unit to swim
or not to swim. So instead or updating:
1- swimming
2- non-swimming
We could detect if the team has more swimming or non-swimming units time to
time. Then use this info as a dynamical optimisation base:
If most units of the team can swim:
1- swimming
2- non-swimming
3- swimming
4- x
If most units of the team can't swim:
1- swimming
2- non-swimming
3- x
4- non-swimming
This is 25% speed improvement, cost is 0% lower quality for most units, and 50%
less quality for few units.

What do you think of this one ? This is 2.5% CPU on a quite simple trick.




reply via email to

[Prev in Thread] Current Thread [Next in Thread]