Did the latest Better AI make it onto 2.8?

[Arkatakor]

As the title states I am wondering if the latest version of Better AI made it onto 2.8 (release)?

As of two weeks ago, I downloaded 2.8 and noticed some very slow turns mid game, seemingly slower then those of the 2.8 beta (though I cannot confirm this). I am assuming the bottleneck is BetterAI hence the query...

 

[os79]

As the title states I am wondering if the latest version of Better AI made it onto 2.8 (release)?

As of two weeks ago, I downloaded 2.8 and noticed some very slow turns mid game, seemingly slower then those of the 2.8 beta (though I cannot confirm this). I am assuming the bottleneck is BetterAI hence the query...

The 2.8 version uses RevolutionDCM 2.6 as its base. Thus the BBAI version in that version, IIRC, is .81M.

 

[Arkatakor]

The 2.8 version uses RevolutionDCM 2.6 as its base. Thus the BBAI version in that version, IIRC, is .81M.

lol could you translate that into English please?

I forgot to clarify that I was running the revolutions mod which seems to be the direction you are pointing me in. Perhaps that could be the culprit. I don't know if disabling it will substantially increase my speed or not. Should that be the case I would be more then happy to.

 

[os79]

lol could you translate that into English please?

I forgot to clarify that I was running the revolutions mod which seems to be the direction you are pointing me in. Perhaps that could be the culprit. I don't know if disabling it will substantially increase my speed or not. Should that be the case I would be more then happy to.

Personally, I already played some RevDCM without Revolutions on. It runs at well-paced speed. Did the same with RoM without revolution on. It slows down terribly.

So,k in the final analysis, it is possible that the fact Zappara adds in more techs, buildings, units and etc to original game, that the game slows due to extra calculation necessary. This is the price you pay for playing the great mod that has its goal as historical realism, or at least close to it as possible.

 

[Arkatakor]

Personally, I already played some RevDCM without Revolutions on. It runs at well-paced speed. Did the same with RoM without revolution on. It slows down terribly.

So,k in the final analysis, it is possible that the fact Zappara adds in more techs, buildings, units and etc to original game, that the game slows due to extra calculation necessary. This is the price you pay for playing the great mod that has its goal as historical realism, or at least close to it as possible.

Whats RevDCM?

There has not been substantial addition of of techs (if any at all) since 2.8 beta and 2.8 release. My point is that i have noticed a slowdown in the release as opposed to the beta. I even tried deleting a few civs to no avail. speed was the same.

There have been numerous discussions on slow turn speeds already and the general consensus is that AI takes up the bulk of the CPU power thus the culprit is BetterAI for the most part, NOT RoM per se. Thats why I am specifically asking if the latest BetterAI made it onto 2.8 release.

 

[zappara]

Better BtS AI 0.81M is included in RoM v2.8 (just hover mouse over civ flag and you'll see this info).

 

[civ_king]

Personally, I already played some RevDCM without Revolutions on. It runs at well-paced speed. Did the same with RoM without revolution on. It slows down terribly.

So,k in the final analysis, it is possible that the fact Zappara adds in more techs, buildings, units and etc to original game, that the game slows due to extra calculation necessary. This is the price you pay for playing the great mod that has its goal as historical realism, or at least close to it as possible.

maybe I should learn C++ so when RoM becomes mostly finished or pieces confirmed they are staying I could write it into DLL to speed up the game

 

[Arkatakor]

Better BtS AI 0.81M is included in RoM v2.8 (just hover mouse over civ flag and you'll see this info).

Then maybe turning off the revolutions mod could speed things up potentially?

 

[os79]

Whats RevDCM?

RevolutionDCM

As ofr your other points. I have no answer to your observations other than saying that it is RoM, not other stripped down mods.

You should ask Zappara exactly what was change between 2.8beta and official one other than RevolutionDCM 2.6 as its base.

The RevolutionDCM 2.6 have BBAI .81M, BUG 4.1, and BULL 1.0. Go to RevolutionDCM thread in Modpacks forum and you should learn more about it. Or go to Revolutions forum in Projects forum.

 

[os79]

Then maybe turning off the revolutions mod could speed things up potentially?

You can experiment. Personally, I doubt it. But report back with the results please .

 

[Afforess]

As for game speed, I'm declaring myself the proverbial expert here.

The game is not slowed down by Revolutions. The RevDCM mod, alone, may actually run 2-5% faster than Vanilla BTS due to some optimizations.

The bottleneck comes not from all the calculations so much as from all the added content. For every building you add, the game has to do yet one more calculation for every loop that checks buildings. One building added could result in 100's of new calculations. Sure, the game can do these in all of a space of 1/10 of a second, but it adds up rather quickly.

I've had ideas to cut the time down before. One such thought would be to limit the AI from considering or doing calculations for buildings/units that are over one era away. This already happens with technologys, ATM. The AI doesn't even know that technologies outside of one higher era than the current one even exist, to speed calculations. I believe this could be ported to other calculations too...

I think one of the new added time killers in 2.8 is the increase in the number of storms. Storms force the game to do mad calculations on every tile. If that wasn't bad enough, the AI then do calculations to try to find "safe" routes through the storms. Since the routes end up being very convoluted, and change every turn, this causes lots and lots of calculations. As much as I like the storm feature, I think we should do some speed tests with RoM without storms and with storms to see the full difference.

 

[os79]

Let's see if that check for buildings for current era only by AI like you described above improve speed somewhat first. If it is not doable by you, then I can experiment with storms off. I just need instructions how to turn it off.

 

[Afforess]

Let's see if that check for buildings for current era only by AI like you described above improve speed somewhat first. If it is not doable by you, then I can experiment with storms off. I just need instructions how to turn it off.

I'll see if I can speed it up some for my next beta. ATM I'm adding new code for buildings for some modmod's, but once that's done, I'm going to focus on speed optimizations and bug stomping.

 

[dexter_inside]

You can experiment. Personally, I doubt it. But report back with the results please .

It doesn't.

The AI really feels better since Afforess got the inquisitor spam fixed

And another thing - somebody (I don't remember who atm.) was concerned that defensive pacts were not working. I thought so too, until Gandhi popped up:

 

[os79]

I just commented out Storm feature from FeatureInfo.xml. I will experiment now. If the game improve significantly, then we will need to let Zappara know and see if he have ideas how to adjust it. But first, let me play the game and report back.

 

[Afforess]

The AI really feels better since Afforess got the inquisitor spam fixed

Zappara fixed that, not me.

This is Zappara's mod after all.

 

[os79]

I stripped Storms from RoM (don't ask me how it involves XML work that only brave and crazy are willing to tread ). The game improves in speed vastly. Actually, it improves by having a steady and very slow increase as normal games are supposed, after considering that I was playing with 18 Civs map with Revolutions and barbarians off. That is a big improvement.

Now off to report my findings to Zappara and let him decide how to use storms feature in future. For now, I just strip it out and have Seafaring modmod do all the job of restricting my ships to coast (before astronomy) until a such time when Storm feature improves.

 

[Afforess]

I stripped Storms from RoM (don't ask me how it involves XML work that only brave and crazy are willing to tread ). The game improves in speed vastly. Actually, it improves by having a steady and very slow increase as normal games are supposed, after considering that I was playing with 18 Civs map with Revolutions and barbarians off. That is a big improvement.

Now off to report my findings to Zappara and let him decide how to use storms feature in future. For now, I just strip it out and have Seafaring modmod do all the job of restricting my ships to coast (before astronomy) until a such time when Storm feature improves.

Hmm, so my suspicions were correct. If they are a really big factor in calculations (Calculations Time > 5 Seconds) they should just be removed. That feature isn't that interesting, and turn times > Storms.

 

[Arkatakor]

As for game speed, I'm declaring myself the proverbial expert here.

The game is not slowed down by Revolutions. The RevDCM mod, alone, may actually run 2-5% faster than Vanilla BTS due to some optimizations.

The bottleneck comes not from all the calculations so much as from all the added content. For every building you add, the game has to do yet one more calculation for every loop that checks buildings. One building added could result in 100's of new calculations. Sure, the game can do these in all of a space of 1/10 of a second, but it adds up rather quickly.

I've had ideas to cut the time down before. One such thought would be to limit the AI from considering or doing calculations for buildings/units that are over one era away. This already happens with technologys, ATM. The AI doesn't even know that technologies outside of one higher era than the current one even exist, to speed calculations. I believe this could be ported to other calculations too...

I think one of the new added time killers in 2.8 is the increase in the number of storms. Storms force the game to do mad calculations on every tile. If that wasn't bad enough, the AI then do calculations to try to find "safe" routes through the storms. Since the routes end up being very convoluted, and change every turn, this causes lots and lots of calculations. As much as I like the storm feature, I think we should do some speed tests with RoM without storms and with storms to see the full difference.

Thanks again for the input, Afforess. It seems according to at least one other person here, that the storms are a substantial bottleneck. As much as they are a nice feature to have, it is one that I am more then happy to do without if it can significantly reduce turn times, be even as little as 10-20%. Your theory also makes sense in the light that there were not so many storms before 2.8 which could explain why we now have longer end of turn times.

EDIT: I was unable to find FeatureInfo.xml (did a recursive search from the root RoM directory) - also, is that the only file I need to modify to disable storms?

 

[RobO]

I stripped Storms from RoM (don't ask me how it involves XML work that only brave and crazy are willing to tread ). The game improves in speed vastly.

Well, I'm going to ask anyway. I really hate the slowdown. I'm guessing that simply setting the probability of getting storms to zero won't do the trick - but it might help.

Is there a reasonably efficient quick fix?