EU2 source code released

[Taylor]

Problem is not with armies really suffering attrition. If it is checked every day, we have to check all armies anyway and increase time devoted to this check 30 times it takes currently. It could be a problem for general performance of the game.

Yes but still... There are only, like 200 armies in the world at max. To check those each day would result in a very limited number of calculations (in computer terms). It will take 30 times the current time, but if the current time is very short, then the extra time will also be very short.

 

[cool-toxic]

You don't play with mods do you ? If you do you should know that lot more armies will be present.

 

[MichaelM]

One thing I hate is lack of army attrition when you have a state that is money poor. I think there should be a % attrition per month for lack of payment to troops.
so as an example , if you had 50000 men and you ran out of money, then 5% of 50000 will disappear until treasury is replenished.

I prefer this method than the treasury interest rate system

I think this could work. I've also thought of making low maintenance increase attrition, like Victoria's system except doing it gradually. Anyone else?

 

[Taylor]

You don't play with mods do you ? If you do you should know that lot more armies will be present.

I usually play agceep on a vanilla map (no watk/mymap). There could be more armies but my point is that as long as there aren't like 1 million, it should be possible.

edit: okay, on the vanilla-agceep map I guess that there are about 500 in a normal save-game. But that's still the same order of magnitude.

 

[BurningEGO]

That only at the start of the game... The AI usually wanders around with lots of armies, tiny or big alike. And at mid game they will have more then at the start.

Instead of a "burning manufactory" option, you should have acess to "scorch land" just like in EU3. Extra atrition, less tax, population growth for a certain time period on the province. Perhaps some extra income for the one that pillages it as well (and maybe some slower movement in the province too).

Also i dont know if this was mentioned before, but a player should be able to see how many troops are besieging/assaulting a fort that the player controls but does not own.

 

[MichaelM]

I think Taylor is correct on the feasibility. The attrition check is only a small part of the monthly update; there are many other things which are province-based, and thus usually take longer than army-based checks. Even if there are 1500 armies, that's still less than the number of provinces.

 

[Taylor]

That only at the start of the game... The AI usually wanders around with lots of armies, tiny or big alike. And at mid game they will have more then at the start.

I intentionally checked a save game somewhere in the middle before I came up with the 500 number.
Even if the number of armies were somewhere in the thousands (which I think it's not), it's still a low number for a computer to calculate.
Like MichealM says, there's so much more to be updated at the beginning of a month. Each nation's income (province tax, production, cot income etc etc), cot value, maintenance, population growth, diplomatic stance of every nation to every nation it knows (accounting for thousands of calculations) etc., and last but certainly not least: AI decisions (which take most of the processing power I think, but correct me if I'm wrong).

 

[MichaelM]

Yes, I'd say AI is probably the hog. I haven't seen all of the code, but I know there are tons of things to calculate.

 

[BurningEGO]

Just hope you guys dont exagerate like paradox did with crusader kings - even with a duo core the game lags badly at every month.

Of course there are a lot of things to be calculated in there, like characters, but still.

P.S: EU2 already lags sometimes - in certain mods that enhance the AI, the game starts moving snail-like mid game.

 

[Taylor]

Yes, I hope the modding team will be able to make the ai more efficient in terms of cpu-usage.

 

[unmerged(4344)]

Problem is not with armies really suffering attrition. If it is checked every day, we have to check all armies anyway and increase time devoted to this check 30 times it takes currently. It could be a problem for general performance of the game.

Currently we have a system with predictable movement attrition, and that sucks because it induces micromanagement that is tedious, uninteresting, and radically ahistoric.

But computing attrition daily would take 30x the CPU horsepower.

Solution: compute movement attrition, on average, once per month, but don't make it predictable when. So, after movement attrition is computed, set the next "movement attrition day" to be a random number of days later, from 1 to 59 days. (Or maybe from 10 to 50; this would make taking movement attrition twice in many typical moves impossible.)

Now, perhaps we do have more compute power to devote to attrition, but just not the full 30x needed to do it daily.

General solution: if we are willing to do N times as many attrition checks, drop movement attrition from 1% to (1/N)%, and set the next attrition day to be from 1 to (60/N-1) days later. I.e., if we can spend 3x as many checks affordably, each one does 0.33% attrition and the next day is from 1 to 19 days later.

 

[Monkii-sama]

Just dropping in wondering if attrition really can be of any high priority. Do you really bother micromanaging troops in late- and mid-game? Usually you only do that with big armies with cannons. At least I do it that way.

 

[MichaelM]

Currently we have a system with predictable movement attrition, and that sucks because it induces micromanagement that is tedious, uninteresting, and radically ahistoric.

But computing attrition daily would take 30x the CPU horsepower.

Solution: compute movement attrition, on average, once per month, but don't make it predictable when. So, after movement attrition is computed, set the next "movement attrition day" to be a random number of days later, from 1 to 59 days. (Or maybe from 10 to 50; this would make taking movement attrition twice in many typical moves impossible.)

Now, perhaps we do have more compute power to devote to attrition, but just not the full 30x needed to do it daily.

General solution: if we are willing to do N times as many attrition checks, drop movement attrition from 1% to (1/N)%, and set the next attrition day to be from 1 to (60/N-1) days later. I.e., if we can spend 3x as many checks affordably, each one does 0.33% attrition and the next day is from 1 to 19 days later.

I suppose doing it every 15 days ought to work, wouldn't it? Or just often enough that it's impossible to avoid it entirely while moving through a high-attrition province.

As mentioned above, attrition shouldn't be a huge CPU hog, but we still want to be careful with adding things to the daily update.

 

[unmerged(40707)]

Just dropping in wondering if attrition really can be of any high priority. Do you really bother micromanaging troops in late- and mid-game? Usually you only do that with big armies with cannons. At least I do it that way.

Yes but you can't afford loosing any soldier in early game, especially for a small country with limlited funds.

I agree with MichaelM and Wreck's idea. It could be a solution.

 

[Toio]

Historically attrition happened every day to every army in our game period, the solution would be 1% attrition per month , per army regards if mobile or static,
unsure if the engine can handle this

also attrition in different weather (snow etc)i assume will not be altered and remain as it is currently

 

[Monkii-sama]

Yes but you can't afford loosing any soldier in early game, especially for a small country with limlited funds.

I agree with MichaelM and Wreck's idea. It could be a solution.

Oh yes, in early game I micromanage every army there is but later on the time needed to manage the troops are better spent on something else.

It's nice to have an advantage in the start but it would be better if you didn't have to consider date at all when moving troops around. Just how long your troops have been in a province/supply limit and what wheather there is.

Edit: In HoI attrittion is counted daily, is there really such a huge difference in CPU usage compared to EU2?

 

[cool-toxic]

EU2 got loads more of days than HOI...

 

[unmerged(4344)]

Michael, it can't be any fixed day of the month, unless the period is 1 day. Even if you did it every 2 days, players could achieve small gains via MM, and (at least to me) the entire point of the change is to make it so I don't have to MM movement.

Toio, although I agree with you it seems like armies had a lot of attrition, I do think that they had more when moving -- easier to desert, then, for one thing. But in any case, the point of this rule is not only about historical accuracy. Rather, by making movement costly it induces players to keep armies still, and thereby cuts off a particularly unpleasant kind of MM. If there was no cost to movement, a player would do best by keeping all of his armies in motion all the time, so that they were essentially covering two provinces at all times. We don't want to make the optimum behavior for a player a MM-intense nightmare.

 

[MichaelM]

Michael, it can't be any fixed day of the month, unless the period is 1 day. Even if you did it every 2 days, players could achieve small gains via MM,...

You mean timing attacks on high-attrition provinces, and that sort of thing?

and (at least to me) the entire point of the change is to make it so I don't have to MM movement.

Yes, I was looking at it that way too.

Toio, although I agree with you it seems like armies had a lot of attrition, I do think that they had more when moving -- easier to desert, then, for one thing. But in any case, the point of this rule is not only about historical accuracy. Rather, by making movement costly it induces players to keep armies still, and thereby cuts off a particularly unpleasant kind of MM. If there was no cost to movement, a player would do best by keeping all of his armies in motion all the time, so that they were essentially covering two provinces at all times. We don't want to make the optimum behavior for a player a MM-intense nightmare.

I definitely agree. Can you suggest any more changes in this line (if you haven't already)?

 

[BurningEGO]

I think atrition is fine as it is...

In Multiplayer, most people always gets more looses from atrition then from real battle (unless on a defensive war, even so there are exceptions). I already said once somewhere in these AGCEEP forums, but i remember loosing about 600-700k around 1550-1600 to atrition alone.

... And dont come and say i am a bad atrition manager. (of course in single player looses will be far smaller, but again, that is an AI problem, not about the engine).

Why bother on making atrition a "daily problem"? I am fairly sure there are more important things then atrition to be worked on...