What does the AI do while the player is thinking, or drinking?

[Taillebois]

There are lots of posts on problems, time to process turns, too many events to calculate each turn etc.

I am not a programmer so I can't offer a technical solution.


But, if the AI is idle when it is not actually working imediately after the "next turn" button has been pressed - is it possible to make it do some work between turns - e.g. instead of running through all 2,000 events for 7 or more factions all in one go - could it run through them during the "idle time" and attach a rough probability of them being achieved in the next turn and then ignore all those with less than X% chance of occurring?

If X became a player option then those who have powerful machines or desire the full game experience can have that, whilst some of us with lower standards can play a simpler but quicker game.

[veji1]

There are lots of posts on problems, time to process turns, too many events to calculate each turn etc.

I am not a programmer so I can't offer a technical solution.


But, if the AI is idle when it is not actually working imediately after the "next turn" button has been pressed - is it possible to make it do some work between turns - e.g. instead of running through all 2,000 events for 7 or more factions all in one go - could it run through them during the "idle time" and attach a rough probability of them being achieved in the next turn and then ignore all those with less than X% chance of occurring?

If X became a player option then those who have powerful machines or desire the full game experience can have that, whilst some of us with lower standards can play a simpler but quicker game.

Interesting question and this is all part of how Ageod games could be improved. the cut off between "long time when the player is thinking and playing and nothing happens for the AI in the meantime" and "the players presses next turn and does nothing, can't access the different ledgers, think about his army org or anything, basically has the entire game taken hostage while the AI processes for 60 seconds at best to 5/6/7/10 minutes at worst" is massive. What you suggest and other similar actions (already precalculating which leaders will be active or inactive, what will the weather be in different zones, even running the smaller factions in the background and leaving only the big countries to be run during turn processing, etc... all these questions are really valid.

The actual turn processing phase is needed to run what happens on the board during those 7 days, ie the actual operational game. but RGDs for the AI, events for the AI, many other things could maybe take place in the background while the player plays. I usually spend 10ish minutes on my turns, fiddling around, making decisions, looking at my units, etc. if the engine used that at 25% intensity to do some of the turn processing that usually takes 4/5ish minutes on my fairly slow laptop, turn processing time could be cult by 50% for me !

This is all theoretical, but this is another example where the game architecture has hit its limits : in BOA or even AACW, waiting 40 to 60 seconds for turn processing, bah.. not really a big deal. but several minutes becomes more annoying, I can only go to the loo or do the dishes so many times !

[Ace]

I have wondered my self this as well a lot. And I have come to the conclusion that during player turn CPU power is used for minor stuff, watching over there is no lag in scrolling, panels opening smootly, etc. This is where old architecture lets AGE engine down. The core of the engine was made at the time where there were no multicore processors. If it were made from scratch today, I envision it would be programed in such way that one core takes over minor things, while other calculate AI moves. There is a limit there can be done by adding features to existing code, and I fear this may be the case. To split processing into two separate processes calculated by two separate cores is a mamouth task that would require rewriting age engine from scratch. I fear that AGE doesn't have financial power for such a task as it would drain their resources...
But I hope I am wrong

[veji1]

I have wondered my self this as well a lot. And I have come to the conclusion that during player turn CPU power is used for minor stuff, watching over there is no lag in scrolling, panels opening smootly, etc. This is where old architecture lets AGE engine down. The core of the engine was made at the time where there were no multicore processors. If it were made from scratch today, I envision it would be programed in such way that one core takes over minor things, while other calculate AI moves. There is a limit there can be done by adding features to existing code, and I fear this may be the case. To split processing into two separate processes calculated by two separate cores is a mamouth task that would require rewriting age engine from scratch. I fear that AGE doesn't have financial power for such a task as it would drain their resources...
But I hope I am wrong

I think this is the core of the issue. the AGE engine was built 10 years ago (Edit : actually more than 10 years ago), tested and validated with BOA and AACW and then expanded and developped to the limits (or beyond) of its abilities with more complicated games but remains at heart the sameish and the one big investment that was made progressively and keeps being used to build goods (aka games). Ageod is a small team who doesn't have the expertise (after all today's engine is the fruit of their labor, they could understandly having a hard time knowing or seeing what to change and how to to do it) nor the money to invest in that expertise to just start again a new engine from close to scratch.

It's a shame because contrary to what they may think the company's main asset isn't the engine itself, is the talent of the team to make games that are immersive, have a historical feeling, are intense, etc. Contrary to the saying, what makes a great cook is the cook himself, not the stove he uses... But to buy a new stove you still need money.

[Pocus]

Hi,

Sound analysis indeed! Over the years, we only had time to improves on the current engine, developing new games with it, but it was beyond our reach to change the underlying architecture. Now, it has been announced at the 'Hall of Wargamer 2015' conferences, where all Slitherine gathers (plus a few journalists) that there is a new engine in development from 'Slitherine Canada' (even if that's not the official designation of the developers there ) that will be used by us in the following years!

So 2016 and beyond sounds like good years for us Ageod!

[veji1]

Hi,

Sound analysis indeed! Over the years, we only had time to improves on the current engine, developing new games with it, but it was beyond our reach to change the underlying architecture. Now, it has been announced at the 'Hall of Wargamer 2015' conferences, where all Slitherine gathers (plus a few journalists) that there is a new engine in development from 'Slitherine Canada' (even if that's not the official designation of the developers there ) that will be used by us in the following years!

So 2016 and beyond sounds like good years for us Ageod!

Hi Pocus, thanks for those news. I must say that I have 3 reactions when reading this :
1/ About time you got a new engine !
2/ Great news, there might be super duper games to come, running fast, we better AI, etc.. !! yay i wanna play games like that about the napoleonic times, 17/18th century warfare, ACW, etc..
3/ Wait a minute.... I just bought WON and basically what I am being told is you realise that the architecture just can't work anymore and it's time to move on to a new engine.... Well bite me, I would have rathered waited 1 year and bought a great WON with the new engine !

[Ace]

Don't jump on it. If you ever developed anything, you would be aware that any new stuff in the beginning has more problems than the old ones. So, I expect this to have some good impact from '17 onwards...

[veji1]

Don't jump on it. If you ever developed anything, you would be aware that any new stuff in the beginning has more problems than the old ones. So, I expect this to have some good impact from '17 onwards...

Meh... BOA was a gem from the get go and their second game AACW was pretty darn awesome straight away and just got better. My point was rather that they know the engine is at its limits and has been for a while (see PON...).