Single Player bugs and crashes v38 plus (SVN) - After the 20th of February 2018

[george585]

There is no need, as players are usually too fast in relation to calendar.
Flat rate of research in buildings was unchanged, but modifiers (% things you can see in screenshots) were lowered.

That's true, but then why the buildings had the values they had, if they could be halved so easily? I mean it seems to me as a temporary quick fix. After upgrading to the latest SVN, while I no longer get overflow, my research rate has "lost" like 200000 points.

You are two and half era too fast and you didn't even conquered third of map.

I suppose it's a matter of personal preference and playing style - how quickly you expand initially, etc. In my view, it does not have to be the case that all leading civs should always be in the same era.

Hopefully @Thunderbrd can fix it in code (and perhaps just in case check for other possible overflows that could occur in late-stage game).

 

[Thunderbrd]

Is the phenomenon of misleading combat odds a known issue? I was starting to get suspicious of the results I was seeing, so I ran a bunch of damaged units into an enemy city to test, and after losing a bunch of low-probability attacks as expected, I had a run of 25 consecutive attacks where the game claimed I had a >60% chance to win/survive, all of which I lost (including 5 that supposedly had a 97% chance of success) - that shouldn't happen more than one in 3 quadrillion times. On closer inspection the odds shown were for attacking one unit, but the unit that actually defended (and won) was of a different type. These were regular military unit types (e.g. heavy axeman) jumping in to defend, not anything you'd expect to have stealth. Is the odds display limited in how many potential defenders it will consider?

Haven't seen this happen in a while but it used to be fairly common. I'm not sure why it's happening in this case and would need a save displaying the problem and a good number of weeks to sort this out. Not an easy issue.

I am using SVN 10564.
There seems to be Research Overflow for my capital city:
It currently produces 235078.46 research points. When I hover my mouse over Dream Visualizer building, which among other things gives +20% research, the actual effects are calculated as: -425589.52 research points. When I actually complete the turn and the Dream Visualizer building is built, the research for the city is reported as 0, and total research is reduced by ~230000. The same things happens when I select a new trait or change civic that results in slightly more commerce / research, it overflows and goes to 0 for the capital city.
I am sure same thing would apply for other cities as well once the get close to this apparent threshold. I thought the limit for those things was around 4 billion (i.e. Int32).

It's an int currently so about 20 million. It's also given a few decimal points so that makes it only capable of holding about 200k. I've never actually done the actual math to figure out where it would happen exactly and it's possible it's decimalized twice at some point. Yields/commerces are not easy to account. We've never seen a single city break the overflow barrier before. The national limit is a bit higher and may be more protected against overflow, though perhaps not as much as it will need to be if a single city can reach this degree of output. Fixing this will be a major pain.

Looking at the screenshot, I'm surprised cities can get THAT populated now. Interesting.

 

[Thunderbrd]

Thank you for a quick fix. But shouldn't the tech costs be adjusted then? I mean if now the buildings provide half as little research, then it will take longer to research new techs unless their costs are adjusted as well?

Tech costs aren't something to adjust individually. How many turns was it, vs is it now, taking you to research a tech on average?

 

[raxo2222]

It's an int currently so about 20 million. It's also given a few decimal points so that makes it only capable of holding about 200k. I've never actually done the actual math to figure out where it would happen exactly and it's possible it's decimalized twice at some point. Yields/commerces are not easy to account. We've never seen a single city break the overflow barrier before. The national limit is a bit higher and may be more protected against overflow, though perhaps not as much as it will need to be if a single city can reach this degree of output. Fixing this will be a major pain.

Yeah, it seems like city income was decimalized twice, otherwise overflow would happen at 20 million.
Same thing happened when I was testing national culture in single city by bumping culture income to extreme values.

As if something was multiplied by 100 and then wasn't divided by 100 before being sent somewhere else.
Maybe there was attempt at reducing rounding errors in commerce/yield modifiers?

 

[raxo2222]

That's true, but then why the buildings had the values they had, if they could be halved so easily? I mean it seems to me as a temporary quick fix. After upgrading to the latest SVN, while I no longer get overflow, my research rate has "lost" like 200000 points.

I suppose it's a matter of personal preference and playing style - how quickly you expand initially, etc. In my view, it does not have to be the case that all leading civs should always be in the same era.

Well I guess they didn't except to overflow so easily.
Also they never were balanced at first place.

By the way tech leader should be in Atomic era according to calendar in your game.
So you could be in Industrial or Atomic era.

 

[george585]

Looking at the screenshot, I'm surprised cities can get THAT populated now. Interesting.

I began the game before you adjusted growth rates in recent builds. Right now the growth is slower.

How many turns was it, vs is it now, taking you to research a tech on average?

It was about 0.8-1.2 turns in the late game (for this game), now, after the modifiers were halved it's like 1.5-2 turns per tech.

Yeah, it seems like city income was decimalized twice, otherwise overflow would happen at 20 million.

It would be great if such a place could be found in code and fixed.

 

[george585]

@Thunderbrd

Btw, thank you for fixing the slow UI during user's turn (apparently due to civ scores calculations). It is now back to being playable!

 

[KaTiON_PT]

That was toffer.

 

[pepper2000]

Regarding the civics, I would point out that all of them, including the ones I added, are in the core now. I'm not necessarily going to say that means they are "somebody else's problem", but they are a ways down on my list.

 

[Thunderbrd]

I began the game before you adjusted growth rates in recent builds. Right now the growth is slower.

I was not shocked by growth as much as ultimate potential. Your city is allocating an extra +20(and more) food consumed PER population and the population is over 200 so that's a change of +4000 consumed food and you STILL aren't starving??? That gives some concept of just how out of balance food consumption was in the first place. I would not have thought it possible for enough food to be generated to keep a size 200 city from starvation at any stage of the game. Impressive. We have needed this latest adjustment to consumption rules more than I realized. Part of the reason you're exceeding the overflow cap in research in that city is due to the massive amounts of specialists you have and with some of the new traits as they were established, the amount of output per specialist could be increased tremendously, making such an overflow even more possible. I hope that the growth rate adjustments make it tougher to ever reach this kind of city size now enough to balance this a little better.

It was about 0.8-1.2 turns in the late game (for this game), now, after the modifiers were halved it's like 1.5-2 turns per tech.

Still way too fast unless you're on Normal or a faster gamespeed. Tech costs should surely not be dropped based on that.

It would be great if such a place could be found in code and fixed.

I'll look for it soon. It was probably necessary for some reason.

 

[raxo2222]

I was not shocked by growth as much as ultimate potential. Your city is allocating an extra +20(and more) food consumed PER population and the population is over 200 so that's a change of +4000 consumed food and you STILL aren't starving??? That gives some concept of just how out of balance food consumption was in the first place. I would not have thought it possible for enough food to be generated to keep a size 200 city from starvation at any stage of the game. Impressive. We have needed this latest adjustment to consumption rules more than I realized. Part of the reason you're exceeding the overflow cap in research in that city is due to the massive amounts of specialists you have and with some of the new traits as they were established, the amount of output per specialist could be increased tremendously, making such an overflow even more possible. I hope that the growth rate adjustments make it tougher to ever reach this kind of city size now enough to balance this a little better.

Well you can get MASSIVE amount of food/production/commerce from trade too.

 

[Thunderbrd]

Well you can get MASSIVE amount of food/production/commerce from trade too.

yeah, late game will take some evaluating to see if and where to retract some modifiers in the system. Trade is kindof a peace bonus since the strongest is with foreign trade usually, so it has a lot of good cause to be strong. But we may have to reduce the strength some. We'll see.

 

[george585]

Still way too fast unless you're on Normal or a faster gamespeed.

I actually prefer to play on faster speeds. Don;t won't my game to take a year (literally) - even this one was started in March!

Personally, I think there's nothing wrong with cities producing that much research or food or production, - especially in the later game. I don't think such things should be artificially constrained.

As for food consumption, if I recall correctly a couple of years ago I remember hardly reaching size 3 or 4 by the end of Prehistoric Era. Starting from about a year ago (wish I knew exact SVN) it changed drastically that I was surprised that I could reach size 10+ by the end of Prehistoric Era.

Hope you will be able to fix the overflow. After all there is a huge difference between 20,000,000 being upper limit vs 200,000. I would not mind reaching those levels again in the later game.

 

[Thunderbrd]

I actually prefer to play on faster speeds. Don;t won't my game to take a year (literally) - even this one was started in March!

What is your current speed setting?

Personally, I think there's nothing wrong with cities producing that much research or food or production, - especially in the later game. I don't think such things should be artificially constrained.

When it gets too elevated it starts making for a data constraint problem. We already have those and controlling the volumes a bit can help. Some efforts to address overflows with an 'infinite fix' will start adding a huge degree of complexity to a lot of regions of the code I'd really prefer to avoid having so much bug potential in.

As for food consumption, if I recall correctly a couple of years ago I remember hardly reaching size 3 or 4 by the end of Prehistoric Era. Starting from about a year ago (wish I knew exact SVN) it changed drastically that I was surprised that I could reach size 10+ by the end of Prehistoric Era.

Yeah, some civic reconfiguring really opened the door to this and we've been working to adapt to that by changing base growth requirement factors. Consumption is a bit different than how much food it takes to grow. Increasing consumption per population gets cumulatively tougher as it goes. I'm not saying there's anything wrong necessarily with the food output right now... just that it's shocking how high it actually gets. No wonder cities would get to a point where they were growing every round and capped out at that. With that much food income, you couldn't possibly establish food-to-grow values that would slow things down... not with the current formulas anyhow and not without making early growth nearly impossible to compensate. The increasing consumption should help to address this high level cap quite a bit. I just hadn't imagined how needed it actually was.

Hope you will be able to fix the overflow. After all there is a huge difference between 20,000,000 being upper limit vs 200,000. I would not mind reaching those levels again in the later game.

There are numerous things that can be done but there are still thresholds I don't want us to have to adapt further to. When we hit new barriers like these, particularly on a local level, it's not a bad idea to see why it's getting so high and also address it there before immediately moving to enable the overflow volume. That said, I would like to expand this local limit some.

 

[george585]

What is your current speed setting?

QuickHandicap = HANDICAP_NOBLE
GameSpeed = GAMESPEED_BLITZ
WorldSize = WORLDSIZE_SMALL

Some efforts to address overflows with an 'infinite fix' will start adding a huge degree of complexity to a lot of regions of the code I'd really prefer to avoid having so much bug potential in.

Yeah, I can imagine it would open up possibility for all manner of bugs. But I was wondering why those are ints anyways instead of floats? I guess that was a decision by Firaxis.

When we hit new barriers like these, particularly on a local level, it's not a bad idea to see why it's getting so high and also address it there before immediately moving to enable the overflow volume.

I think my preferences and playstyle may have a big influence on how the game develops, so don't judge by it too much how you need to tweak things based on my experience alone. In particular, I like to play with many Civs while not as much land per Civ; no Tech. diffusion; building all possible Religious buildings in all cities; heavily using animals; no Nat. or World Wonder limits. Factors like these play a big role in why some cities end up with producing a lot of research, or food, or production.

 

[tmv]

But I was wondering why those are ints anyways instead of floats? I guess that was a decision by Firaxis.

Working with floats can be nasty. You get all kind of rounding effects, one of the worst being that you cannot just compare with "==", you need to compare the difference of two values to a number slightly above 0, and then there is the problem that you - obviously - don't get all numbers within the range, you won't even get all integers in the range. With e.g. a 32bit float, you get much less than the 4 billion numbers you should have with 32bit, because many 32bit representations are NaN.

More information about this topic is available here: https://en.wikipedia.org/wiki/Floating-point_arithmetic (and links from there)

 

[Thunderbrd]

Yeah, I can imagine it would open up possibility for all manner of bugs. But I was wondering why those are ints anyways instead of floats? I guess that was a decision by Firaxis.

They are ints. We found we needed more granularity. They aren't floats though, they are multiplied by 100 then later divided by 100 and expressed as a float. Somewhere along the lines another *100 /100 process was necessary for some reason.

 

[Zanden]

I'm looking for advice about my current 'crashing' saved game.
The game doesn't lock up and actually crash as I end the turn, but instead simply closes the game silently.
I don't think actually posting the saved game would help, as the last time I did it ran fine for others.

Only modmod used was PPIO.
Tried without PPIO, tried fullscreen and windowed, tried viewports, tried deletion of all wild animals off the map, .ini disabled audio and did a full SVN cleanup to a default state.

Large Earth Map scenario just entering Ancient era(Standard map size on menu), Windows 7, 8 GB RAM(1.1 GB used by the game at the time), i7 @ 2.67 GHz and a GTX 650 TI card.
Would greatly appreciate any suggestions on what may fix the issue as I have no actual error logs to work with or provide.

 

[raxo2222]

I'm looking for advice about my current 'crashing' saved game.
The game doesn't lock up and actually crash as I end the turn, but instead simply closes the game silently.
I don't think actually posting the saved game would help, as the last time I did it ran fine for others.

Only modmod used was PPIO.
Tried without PPIO, tried fullscreen and windowed, tried viewports, tried deletion of all wild animals off the map, .ini disabled audio and did a full SVN cleanup to a default state.

Large Earth Map scenario just entering Ancient era(Standard map size on menu), Windows 7, 8 GB RAM(1.1 GB used by the game at the time), i7 @ 2.67 GHz and a GTX 650 TI card.
Would greatly appreciate any suggestions on what may fix the issue as I have no actual error logs to work with or provide.

Upload save as this time it may be reproducible bug.

Also update SVN to latest version.

 

[JosEPh_II]

I'm looking for advice about my current 'crashing' saved game.
The game doesn't lock up and actually crash as I end the turn, but instead simply closes the game silently.
I don't think actually posting the saved game would help, as the last time I did it ran fine for others.

Only modmod used was PPIO.
Tried without PPIO, tried fullscreen and windowed, tried viewports, tried deletion of all wild animals off the map, .ini disabled audio and did a full SVN cleanup to a default state.

Large Earth Map scenario just entering Ancient era(Standard map size on menu), Windows 7, 8 GB RAM(1.1 GB used by the game at the time), i7 @ 2.67 GHz and a GTX 650 TI card.
Would greatly appreciate any suggestions on what may fix the issue as I have no actual error logs to work with or provide.

In the Civ BtS folder were the BtS.exe resides you should find a file with a .dmp extension. This is the Minidump that was added by one of our best coder/modder for collecting data about CTDs. You must Compress the file and upload Please. Use .zip, .rar, or .7z, all are acceptable compressions that this site will allow to be uploaded for this type file.

The file below is an example of the Minidump file compressed with .7z and uploaded. You game during going away to desktop at End of Turn wait should produce a MiniDump.dmp file. The example was from SVN version 10517 that I uploaded after a crash to desktop. Or as you put it "but instead simply closes the game silently." This is a CTD. The cause could be graphical or other sources that would do this.