[1.0.10.15] Late game (turn 300+) instability on larger than vanilla maps

[Gyrofalcon]

Is there any way to fast up the turn time? it take about 100-120 seconds each turn after i discovery say 8-10 civs out of 20 on giant earth

Without knowing your hardware specs, I would suggest:
-Play in Strategic view
-Turn down leader graphics
-Set all movement and combat to Quick

20 civs on giant map are very demanding on hardware. And if you throw in a mod or 20 or so, it will be noticed.

 

[Lohit]

Without knowing your hardware specs, I would suggest:
-Play in Strategic view
-Turn down leader graphics
-Set all movement and combat to Quick

20 civs on giant map are very demanding on hardware. And if you throw in a mod or 20 or so, it will be noticed.

Thank you for your response. One of the reasons I upgraded my PC was for this game. I have 17 8700K @ 3.70 GHz .. 16 GB RAM and NVIDIA GeForce GTX 1080 Ti

Also, I was running on medium settings but now I've put my settings to lowest. I'm playing with 21 civs including mine and 3 city states on Giant Earth (Gathering Storm)

I'll update here if I see any improvement / weird stuff.

Note - I'm loading the data base from the create game before loading my save. I'm at turn 95 on epic and so far so good.

 

[JG2400]

I just finished a 496-turn GS playthrough (w/o New Frontier) on Giant Earth(Full Map) w/ 12 Civs and 24 City-States. I was initially crashing to desktop around turn 260 when trying to Load the Game with an EXCEPTION_ACCESS_VIOLATION error, but I stumbled upon something which seems to have solved it: running "Load Database" under "Create Game" before trying to load the save. I don't know if this is something we were supposed to be doing the whole time, but seems to have fixed my issue.

Once I figured this out, I have managed to win the game w/ a Diplomatic Victory on turn 496. The inter-turn is still slow (there is definitely at least a minute of waiting), and loading the game takes 20-30 minutes by turn 450, but I was able to finish regardless. I disabled coastal lowland on game setup and did not start the Space Race, though I did reveal most of the map through regular exploration (IDK if this makes a difference), so these are not tested.

Another interesting note is that some models are really messed up post-260 (like some units have shields for heads), but this issue was only visual afaik. I believe that this may have been caused by one of the other mods I used being updated mid-playthrough (this could have also been why I crashed around turn 260, incidentally). The game seems to run fine otherwise.

I have no clue if any of this is helpful, but I just wanted to mention the "Load Database" thing in case it helps.

To follow up on this, I made it to turn 561. I only had to stop because "Religions Expanded" had an overhaul which broke the save. Will be interested in how far other people can go!

 

[archimedes0-0]

Has the crash when the sea level rises been fixed?

 

[Gerkizd]

Those that say load database stops crashes, try reveal all and see if it crashes

Tested on my machine and Giant map still crash

 

[Gedemon]

"load database" is just launching a small map in WorldBuilder mode, add a few data (used by some of YnAMP setup options) into a shared Lua table and go back to the main menu, that's all.

I don't see how it could prevent crashes.

 

[Gyrofalcon]

"load database" is just launching a small map in WorldBuilder mode, add a few data (used by some of YnAMP setup options) into a shared Lua table and go back to the main menu, that's all.

I don't see how it could prevent crashes.

...but still I see a radical decrease in number of game crashes... I don't understand....

 

[Gedemon]

Well, I don't know for random crashes that occurs late game (maybe "preloading" the game engine has a kind of effect on stability ???) but I can say that it never prevented the sea rising, satellite reveal or other hard crashes during my tests.

 

[Lohit]

Not sure if anybody is experiencing sudden increase in AI turn time. I played with latest on GS ruleset and it's still better but new one is taking more than 2 mins per turn after I meet 10 out of 20 civs on giant map without barbs and with lowest CPU settings.
I have top notch PC though, 17 8700K @ 3.70 GHz .. 16 GB RAM and NVIDIA GeForce GTX 1080 Ti

 

[Gyrofalcon]

Not sure if anybody is experiencing sudden increase in AI turn time. I played with latest on GS ruleset and it's still better but new one is taking more than 2 mins per turn after I meet 10 out of 20 civs on giant map without barbs and with lowest CPU settings.
I have top notch PC though, 17 8700K @ 3.70 GHz .. 16 GB RAM and NVIDIA GeForce GTX 1080 Ti

Is it the same in Strategic view?

 

[Gyrofalcon]

Short update on my frezzing and/ or CTD:
I found this article https://nvidia.custhelp.com/app/answers/detail/a_id/5046/kw/451.85

So I updated from 451.48 to 451.67 to hotfix 451.85.
And in my illusion, it is more stable. Both in normal graphical and strategic view. Have not tested with DX12 yet.
But of course, it crashed (7z with logs, exe.dmp and dxdiag):

 

[dshole]

I'm still testing, but I'm not sure the load of the database is helping. I thought our Dev friends were looking at this forum?

 

[Lohit]

Is it the same in Strategic view?

Let me try the strategic view shortly and will respond back.

 

[Bkd_civ]

Unhandled Exception
Code: EXCEPTION_ACCESS_VIOLATION
Error reading address 0x0

I'm still having this hard crash after reinstalling steam, civ (2 times), running on dx12 or 11, playing on giant earth + agricultural revolution. Loading the database first doesn't solve it, it seems that something between AI turns is corrupting the save game at a certain and determined point

My .DMP + logs folder zipped:

 

[Blonx]

We do! Logs and dumps appreciated!

Someone pointed on Steam IIRC that for the largest size crashing on load, it seems to be a allocation issue

extract from Renderer.log:

[2020-04-25 22:59:20]    PSO creation took 0.825176 seconds
[2020-04-25 22:59:20]    Growing UpdateSubresource invocations
[2020-04-25 22:59:20]    IncFrame:1
[2020-04-25 22:59:26]    AppHost::WndProcEx(): Window change detected, resizing buffers...
[2020-04-25 22:59:26]    IncFrame:2
[2020-04-25 22:59:26]    IncFrame:3
[2020-04-25 22:59:26]    IncFrame:4
[2020-04-25 22:59:26]    IncFrame:5
[2020-04-25 22:59:26]    IncFrame:6
[2020-04-25 22:59:26]    IncFrame:7
[2020-04-25 22:59:26]    IncFrame:8
[2020-04-25 22:59:26]    IncFrame:9
[2020-04-25 22:59:26]    IncFrame:10
[2020-04-25 22:59:26]    IncFrame:11
[2020-04-25 22:59:26]    IncFrame:12
[2020-04-25 22:59:26]    IncFrame:13
[2020-04-25 22:59:26]    IncFrame:14
[2020-04-25 22:59:26]    IncFrame:15
[2020-04-25 22:59:29]    Growing UpdateSubresource invocations
[2020-04-25 22:59:36]    Growing UpdateSubresource invocations
[2020-04-25 23:06:03]    Error: ResourceList.AddResource(), attempting to add more than 32768 items.
[2020-04-25 23:06:03]    Error: ResourceList.AddResource(), attempting to add more than 32768 items.
[2020-04-25 23:06:03]    Error: ResourceList.AddResource(), attempting to add more than 32768 items.
[2020-04-25 23:06:03]    Error: ResourceList.AddResource(), attempting to add more than 32768 items.

---------------------------------------------------------------------------------------------


I am sure you guys (firaxis) have already looked into it but I figure it couldn't hurt to offer a suggestion that might have slipped by.

From these logs, I noticed the number "32768" as being 2^15. This being value of a [short int] in c++ (http://www.cplusplus.com/reference/climits/)
...

I am going to just write some sudo code of how I imagine this works

------------------------------------

ResourceList<Resource> = new List<Resource>();

Resource
{
    ResourceType Type;
    list<ResourceObject> Instances;
}

ResourceObject
{
    Location loc;
    short int Quantity;
}

----------------------------------

My guess is that somewhere in the engine, to save space someone hard capped the max size of a list to 2^15. or perhaps the "count" variable or "size" variable is of type short int.

The reason this wouldnt present issues on most map sizes is that the list on "Resource" - "Instances" would be ridiculous to exceed 32,000. Infact, it would require more than 3 instances of the resource on EVERY tile. (120*80 = 9600, 32000/9600 aprox=3)



I could be completely off. Im going off complete guess work and 1 single item in the error log.
-------------------------------------


Considering this post was 3 months ago- Firaxis, any updates or progress on this?

 

[Lohit]

Just responding back. Yup tried strategic view as well. No change. GS is still ok but new Front is taking avg 2 mins per turn on my high end PC with minimum graphics setting.

 

[Bkd_civ]

Is anybody able to play on Giant Earth without crashes until the end?
It's always crashing for me and don't allow me to progress pass ~100 turns. I disable sea rise, satelite reveal, disasters, barbarians, and starting on renaissance...

Also, in my database.log i get these errors:


[3234741.692] [Gameplay] ERROR: UNIQUE constraint failed: Requirements.RequirementId
[3234741.692] [Gameplay]: While executing - 'insert into Requirements('RequirementId', 'RequirementType') values (?, ?);'
[3234741.692] [Gameplay]: In XMLSerializer while inserting row into table insert into Requirements('RequirementId', 'RequirementType') with values (REQUIRES_CITY_HAS_PALACE, REQUIREMENT_CITY_HAS_BUILDING, ).
[3234741.692] [Gameplay]: In XMLSerializer while updating table Requirements from file Expansion1_Governments.xml.
[3234741.692] [Gameplay] ERROR: UNIQUE constraint failed: Requirements.RequirementId
[3234760.115] [Gameplay]: Validating Foreign Key Constraints...
[3234760.142] [Gameplay]: Passed Validation.
[3234763.918] [ColorManager] ERROR: UNIQUE constraint failed: PlayerColors.Type
[3234763.918] [ColorManager] ERROR: UNIQUE constraint failed: PlayerColors.Type
[3234763.918] [ColorManager]: ../../../DLC/Nubia_Amanitore/Data/Nubia_PlayerColors.xml
[3234763.918] [ColorManager] ERROR: UNIQUE constraint failed: PlayerColors.Type
[3234763.918] [ColorManager] ERROR: UNIQUE constraint failed: PlayerColors.Type
[3234763.918] [ColorManager]: ../../../DLC/Australia/Data/Australia_Colors.xml

 

[Lohit]

Not sure what it is now. GS ruleset, almost was at the end of science victory, this is happening.
Unhandled Exception
Code: EXCEPTION_ACCESS_VIOLATION
Error reading address 0xfce59822
Call Stack
-----------------------------------------------------
??? ! ??? ???, line 0
??? ! ??? ???, line 0
??? ! ??? ???, line 0
??? ! ??? ???, line 0
??? ! ??? ???, line 0
??? ! ??? ???, line 0
??? ! ??? ???, line 0
??? ! ??? ???, line 0
??? ! ??? ???, line 0
??? ! ??? ???, line 0
??? ! ??? ???, line 0
??? ! ??? ???, line 0
??? ! ??? ???, line 0
??? ! ??? ???, line 0
??? ! ??? ???, line 0
??? ! ??? ???, line 0
??? ! ??? ???, line 0
??? ! ??? ???, line 0
??? ! ??? ???, line 0

 

[Lohit]

I think I'm done with this game. I only like playing on realistic giant earth but its not supportive so it's ok. Civ 7 it is whenever its out

 

[Blonx]

I think I'm done with this game. I only like playing on realistic giant earth but its not supportive so it's ok. Civ 7 it is whenever its out

Hoping you might look at notifcations, I know I dont.


There is a mod that can be used in tandom with this one and it seems stable. https://steamcommunity.com/sharedfiles/filedetails/?id=1826889771
called: "Sammaye's YnAMP Earth128x80 (Modified)"

Its not as good as the bigger one in this series, but its still nice.