AU402 - Gargantua

[vmxa1]

Oh I will play it out, but I am not having much fun. I did learn a few things about playing on water.

[Jawa Jocky]

Originally posted by Theseus
no one has any idea that AU402 is probably the most resource intensive game on the planet today.

Do you think IBM would let us borrow deep blue (or some other supercomputer)?

Between turns at 2000AD would only take 4 minutes

[Theseus]

Atari should use this from a marketing perspective.

Actually, Atari should use AU altogether from a marketing perspective!!

[vulture]

Originally posted by vmxa1
Of all the things that I can not understand, the time lag for the settler to get from the squat to a city view is the strangest. It can take many minutes and I have no idea what it needs to be calculating for city in the midst of a bunch of other cities of mine.
Yes turn will get to an hour late in the game if you have not eliminated a bunch of civ and are less than a 2gh machine.

Almost all that time is spent calculating trade routes and updating the games internal representation of the trade network, as I understand it. It may seem trivially obvious to us what the connections are, but we have an awful lot of processing power in our heads dedicated to nothing but jobs like this. For a computer it can be fiendish, and any 'simplified' algorithms tend not to work because you never have to look very far to find a problem that breaks the algorithm, and one will come up at least a few times per game.

The same applies for obvious reasons to the lag noticed by Theseus selling navigation and astronomy. Building and destroying harbours and airports is also costly in time for the same reason. There isn't much that can be done to reduce this without abolishing the trade system entirely (I'm assuming, not unreasonably, that the algorithm implemented in civ 3 is close to optimal, since such problems are very well studied these days and the fastest solutions are probably widely known).

[cumi]

We should find the algorithm for finding of Hamilton-Circles to improve the game. We will get a Nobel-prise for it, too. (But not from the intelligence agencies )

[nbarclay]

Vulture, half the battle in choosing an algorithm is picking a definition of the problem that lends itself to an efficient algorithm. It may well be that Firaxis is using an excellent algorithm, but if so, it is an excellent algorithm to solve an extremely poor definition of the problem (assuming that trade route checks are what causes so much slowness when cities are founded, captured, or disbanded).

In Civ 3, there are enormous numbers of tile-equivalences for trade purposes. All connected cities are equivalent, so if there is a trade route to one, there is a trade route to all. Similarly, the vast majority of the ocean is equivalent for trading purposes, the exception being cases of blockades. That should lend itself to algorithms that can resolve at least the vast majority of trade situations in a small fraction of a second even on a 250x250 map.

[vmxa1]

Yes that is my presumtion, it is about checking routes.
My point is that it is not done well if that is the case.
A quick check could end the checking. Am I now connected to my civ, yes. Do any of the cites close to me and connected have a Harbor or an Airport, yes. End of checking for now.
The same issue comes when I raze or capture one of their cities. They lose a city that was in the midst of a lots of other cities. All have RR. How much checking is need to be done here? Not much, but it takes a lot of time? Now if this city had the only harbor or airport, that is another story.