Single Player bugs and crashes v36 plus (SVN) - After the 24th of October 2015

[Arakhor]

A minute between turns? Rather you than me!

Then again, you are rather asking for it by playing a Giant map. Even a Huge map used to lag tremendously in the late Renaissance for me, which is why I decided to try out a Large map and thus discovered my latest finding.

 

[Toffer90]

A minute between turns? Rather you than me!

Explainable by his relatively slow CPU: 2.4 GHz; as he said himself it is not a 64bit game so turn times only rely on how high a clock any single core can reach. I'm glad I still have my 4.5 GHz i5 2500k.
10 000 units per civ sounds a bit excessive though, if that is a correct number then something is seriously bugged and we would really have to look into that. about 20-50 units per city on average per civ should IMO be adequate through all eras.

 

[Dreifels]

A minute between turns? Rather you than me!

Then again, you are rather asking for it by playing a Giant map. Even a Huge map used to lag tremendously

Large: 104 x 64
Huge: 128 x 80
Giant: 160 x 100
Gigantic: 200 x 120
Enormous: 240 x 160
Immense: 300 x 200

there is no huge map around for C2C

map size is not the problem, that only requires more VRAM, and the game can handle more than 1,000 MB VRAM well.

RAM und CPU time is the point, RAM because the 3GB border (2.956 exact) and CPU time because the amount of cigars you can smoke between the turns.
(I remember, in January I had a game with more than 8 Min waiting time between the turns)

 

[Dancing Hoskuld]

The only way I see to reduce calculation time is to reduce the total amount or units, which sometimes can be more than 10,000 each civ (as the last game had).

This means, create a way that a unit can get much more stronger and upgrade and at same time reduce the amount of units, so that (i.e.) now 10 good attacker are fine to get a city while 10 defender are good to defend it, and not, as in my last game, a city had more than 250 defender (all low level units) and around there had been in each field also more than 250, so that I neede more than 1,000 military units to attack well.

The AI uses more mass of low level units than smaller amount but well upgraded. Maybe that you can manage by maintenance costs (food or gold) which should be same for low level and high level unit, so high that the AI can't build that mass and decides to upgrade instead to create a new unit.

Limit amount of units per tile does not help. This only forces that any field gets maximum of allowed units and units no longer can move and the civ in case freezes de facto (that I had when I set the limit for test, it gives terrible result.)

In my opinion this is what Size Matters could/should be addressing and why I feel it should only allow units to be merged and never split. By not being able to split the merged unit you don't need to keep track of as much info about the individual units.

 

[Thunderbrd]

In my opinion this is what Size Matters could/should be addressing and why I feel it should only allow units to be merged and never split. By not being able to split the merged unit you don't need to keep track of as much info about the individual units.

A player may use the system how the player wishes. The AI tends to prefer, except in some rare cases, merging. Some strategies for splitting can be a WAY to go about things that at some point I'd like to program AI's to utilize based on more 'personality' factors.

But there's no reason to disallow splitting. At all. The idea of the mod is to make it possible to manage your forces how you see fit and to do so with a concept of realism infused into the fabric of that system. If you wanted to split up your forces, there should be no reason you can't, but the reality is, in MOST cases, there is strength in greater numbers.

 

[Dreifels]

Explainable by his relatively slow CPU: 2.4 GHz; as he said himself it is not a 64bit game so turn times only rely on how high a clock any single core can reach. I'm glad I still have my 4.5 GHz i5 2500k.
10 000 units per civ sounds a bit excessive though, if that is a correct number then something is seriously bugged and we would really have to look into that. about 20-50 units per city on average per civ should IMO be adequate through all eras.

these figures are from my January game. I opened the worldbuilder and checked what another civ had, and see any (!) field of his land full of units and I counted more than 40 fields with more than 250 units each field. All healer and other non-military stuff and two or three low level mil units. That was so much and so pressed probably to maximum that the civ could not move any unit. That was the way to defense: mass of units, not quality.

And this was not only one civ, but several cvis, but not all. (I posted it that time here)

In between you made several changes, so I cannot say if such still happens with the current svn. (the more, as my new game just crashhed)

 

[Dreifels]

If you wanted to split up your forces, there should be no reason you can't, but the reality is, in MOST cases, there is strength in greater numbers.

agree, and this would not help. If ther AI's strategy is mass of low level and civilian instead less but high quality, then it is because its calculation say it gives better result and is cheaper (in food or hammer or gold).

Only if the AI's calculation says that this is not the right way, then it works right. And such you can force by make a price for a unit, a one time pirice for building (hammer) and an other price for maintenance (food or gold) and the AI is forced to upgrade to higher level because that is cheaper than creating and maintenance for more low level units. And the maintenance price should be same for all units, not depending from level, as one man eats the same, independent if he is a well trained fighter or a worker or whatever.

Explainable by his relatively slow CPU: 2.4 GHz; as he said himself it is not a 64bit game so turn times only rely on how high a clock any single core can reach. I'm glad I still have my 4.5 GHz i5 2500k.

this you cannot expect from all users, because most having high end machines want wo play high end games.
You need to see what systems are around in mass, and I think my system is reflecting the mass and 2,4 GHz for an I7 is fine, as with a 64 bit game it is strong enough because Vram and SSD.

btw, on my old i5 3.5 GHz it is not faster.

 

[strategyonly]

at TB

go back one page i have 2 CTD's just incase u missed them , , thx . .

and also lets move this talking about AI/turn times to the discussion thread, thx . .

 

[Dreifels]

Assuming you mean the plot improvement palisade you can, it just takes a different tech. If I remember correctly it was done this way to represent the idea that a garrison is needed to provide the man power to upgrade a palisade. This is why fortifying a unit in a fortification will upgrade it.

currently wooden palisade => fort
I think there should be a step between:
wooden palisade => walled palisade => fort
while walled palisate (or other name) is able to build / upgrade at same time when a city can build its wall, as a job for workers, not automatically.

 

[Dreifels]

at TB
go back one page i have 2 CTD's just incase u missed them , , thx . ..

with or w/o mine?

and also lets move this talking about AI/turn times to the discussion thread, thx . .

yes, it becomes confusingly. Is there a way to move the postings realted speed to the discussion thread?

 

[Toffer90]

this you cannot expect from all users, because most having high end machines want wo play high end games.
You need to see what systems are around in mass, and I think my system is reflecting the mass and 2,4 GHz for an I7 is fine, as with a 64 bit game it is strong enough because Vram and SSD.

True enough, I can't expect my old machine to represent the norm. ^^ I was merely pointing out that an old game may perform far better on an old CPU with a higher clock than a new CPU.
Old games rarely utilize the better architecture found in newer CPU's very well.

 

[JosEPh_II]

I have a brand new computer JUST for C2C, its a Nvidia 980, 6th Gen Intel® Core™ i7 processor; 16GB memory; 2TB hard drive and 256GB solid state drive
Special features: Blu-ray playback; 4GB dedicated graphics; built-in wireless networking; Bluetooth; HDMI output, and i have a special button just for TURBO-Boost.

And my turn times are over 10 seconds from start, on a LARGE map, and only 7 civs. . your map, then when i get 1/2 way through PreH its goes to 15 seconds, and then when i hit 1/2 through the Ancient Era, it goes to 30 seconds . .

Sorry guys but to assert that 10 sec EoT in a 32bit engine game with a New state of the Art Comp is "Too Long" is in and of itself rather ridiculous.

To play with a 5 sec EoT with a SSD is rather amazing But you should Not bench mark the game by these standards.

I have an i7 2500K 3.2GHz cpu with 8GB 1066Mhz ram (I also use the 4GB Patch which gives me an extra 500MB more than the 32bit 3GB switch). I have an Nvidia GT 550 ti vid card 1GB DDR3. My turn time in Preh and ancient avg About 7-15 sec. which I find is acceptable on a Large PM map with 7AI and I do not use (generally) Any REV or Combat Mod Options this included H&S. These ALL add extra processing time. By Classical I'm up to 20-30 sec. By Ren Era I'm up to 2-10 min each EoT dependent upon number of Battles taking place by that time.

Toffer wrote:Then I'm convinced that we should remove something, the question is what. Perhaps disease, pollution and crime diffusion will have the biggest impact. What do you think TB?

Don't forget the (gawd awful)Tourism and Education either.

Crime and Disease diffuse to adj city tiles. So 8 more tiles for every city Must have 2 separate loop thru(s). 1 to "diffuse" crime to the 8 individual adj tiles and a 2nd to then process and return a portion back to the Main city tile from each of the 8 adj tiles. So for every city in a game 16 processes for crime and another 16 for Disease. Now as the game gets more cities the processes involved are multiplied by the number of cities in the game. There are of course variances due to a city losing one or more of the adj city tiles thru culture border loss. If you do not use the Minimum City Tile Option which the guarantees that all 8 tiles stay in your main city tiles control.

Pollution, Tourism, and Education are different stories.

To remove the diffusion alone would save much processing. But because we've allowed it to be unchecked for so long taking it out will cause Much rework of the core concepts of crime and disease, etc. (Basically taking the Mod back to v 23/24 when Hydro 1st introduced crime. And before the property system was put in by AIAndy.)

T-Brd, I would bet, would Not be in favor of removing the property system.

So do you rip out parts that have now become core to the mod's game play so that your new fancy computer can stay at 5sec or less per turn?

Your Mod you supply that answer. Or stop whining over 10 sec EoT's. Which will it be?

Just sayin'.

JosEPh

 

[Thunderbrd]

Hi

I'm getting a CTD during end turn.
I think its right before my next turn starts when it happens.

I'm now runnig 9186 but it was happening to a few builds ago.

Well... gentlemen I can only address one issue at a time and this one being the next on the plate is one I'm tempted already to simply declare outside of my ability to solve. The crash is taking place in the exe with no hint whatsoever as to how. This doesn't strike me as a graphic issue which often gives a clue. I'm hundreds of lines deep into blindly walking through what's happening in the exe (which is just a disassembly of code line hits without any explanation of the coding in those lines whatsoever) and I have no idea how many hundreds more its going to take to figure out what the LAST possible call in the dll may have been so I can possibly, maybe, but unlikely, find a HINT as to what may be taking place IN the exe so I can get an idea on how to formulate a theory as to the nature of the crash.

Perhaps from my text you can tell how frustrating this one is. So far no guesses as to the cause and it may take days of further analysis to find that I have no basis upon which to make any further guesses aside from perhaps making educated guesses based on recent code changes, almost all of which have been to resolve other problems.

This COULD be the death knell of this NPC splitting project entirely however there will be no way to prove (since the change would break saves) that retracting that project in its entirety would avoid whatever was causing this particular crash.

I would not leap, therefore, to that assumption. I would instead see if there's any correlation between this bug and the next reported crash points and maybe one of them will give a necessity for change that may inadvertently solve this one (which would not be the first time that's happened during the course of this ever so frustrating project.)

@SO: Removing things from the mod will not help speed it up until you remove more than we can possibly fathom removing. Furthermore, it's nice to see people ing about speed when I just sped things up significantly (like maybe by 20%) with the last commit. But let's face it, it was known that the beginning of the game would take on a LITTLE more time cost per turn while it would have no impact on the later game because the NPCs would be mostly eliminated by then.

At this point though, speed is not my concern. My concern is absolutely simply viability of the mod in the status its in. I'm getting tired of debugging crash after crash because I'm fighting against the exe I cannot see nor make adjustments to that would be so much more easily capable of resolving the most basic problems and make everything run the way it was intended to. That and a host of glitches thanks to poor python programming that I don't have the skill to fix, meanwhile we have our best python programmer losing heart as badly as I am.

My conclusion is growing to open up to the concept that it may be necessary to back everything up to before this project began and leave barbarians as the only npc team handling all things. This kind of waste of effort, however, is exactly the sort of thing that a guy starts feeling like walking away from a project after.

So perhaps, if I can get a LITTLE help from another programmer who can take a look at some of these issues piling up, take a look at some of the methods I've used and maybe let me know if I'm making any really bad base assumptions somewhere, that could be very helpful. I'm getting better but I'm still no professional unfortunately.

Furthermore, if we ARE talking about moving towards removing properties, I will leave the team at the next hint of further discussion on that. Properties IS what defines C2C as being above and beyond Vanilla and is crucial to further development plans. Without them I might as well give up now.


Look... at some point I'll be looking at profiling turns and getting a better idea which AI routines are causing the greatest delay (it's been my experience all noteworthy delays that can be improved on are generally there.) Or wherever else it's taking place. But taking random guesses as to what is currently causing delays and slowdowns and taking aim at projects just to speed things up is absolutely pointless. It will fail to fix the problems and it will succeed at fouling up longer term development plans and it will knock us back years in development. Years. Not months. Not days. Years. Years we'll never get back people.

 

[Dreifels]

I also use the 4GB Patch which gives me an extra 500MB more than the 32bit 3GB switch

you are confusing me. what "4GB" Patch you have and what is the effective useable ram with it?

The only patch I know is that of http://www.ntcore.com/4gb_patch.php and this is a 3GB patch as it allows to use up to 2.956 GB, all other above is kernel.

That was the same like the former 2GB-break for XP which had done the same.
A patch that give 500 MB only for XP I never seen.

 

[Toffer90]

T-Brd, I would bet, would Not be in favor of removing the property system.

I'm strongly opposed to this notion myself.

@TB: Take your time, ignore the nagging, I'm sure we'll get the mod on the right track soon enough.

 

[strategyonly]

So these current CTD's cant be fixed?? Because right now i cant even get pass 1/3 of the beginning of the Ancient Era . .

 

[Toffer90]

So these current CTD's cant be fixed?? Because right now i cant even get pass 1/3 of the beginning of the Ancient Era . .

It's IMO way too early to jump to the conclusion that it can't be fixed.

See if playing without Neanderthal cities helps.

Edit: I must have misread SO's post.

 

[Dancing Hoskuld]

Furthermore, if we ARE talking about moving towards removing properties, I will leave the team at the next hint of further discussion on that. Properties IS what defines C2C as being above and beyond Vanilla and is crucial to further development plans. Without them I might as well give up now.

We have added a major change to the way things happen in C2C and the solution to turn times is to go back and take something that is working fine out? Seems a bit strange to me. Properties are fine as far as I can see; even the few that spread are working OK, so I am with TB on this one. Besides I need 3-5 game level properties and 10-20 player level properties for some of the things I want to do.

The problems we were having with turn times have gone as far as my last game goes. The turn times have been good for me and while I have an SSD I discovered yesterday that I don't use it for Civ IV

As far as i can tell the problems, besides the graphical ones, are all about the number of units in the game. In the early game this is the number of animals but in later games it is the number of normal units. As has been said elsewhere we need to figure out a better way of doing the combat units and as far as the animals are concerned we need a better spawn set up.

The old way of one set of animals which were also barbarians worked but had a few areas which were not the way we wanted them in game. With the three animal nations we are just finding our feet. The two problems I have with the animal option I have tried is that choosing it or not gives extreme results ie there is either 10 animals per plot or only 1 animal per 100 plots. Neither is a good result and we need to look at ways to fix it. I have only been trying the option which has the animals fighting each other or not.

 

[Toffer90]

As far as i can tell the problems, besides the graphical ones, are all about the number of units in the game. In the early game this is the number of animals but in later games it is the number of normal units. As has been said elsewhere we need to figure out a better way of doing the combat units and as far as the animals are concerned we need a better spawn set up.

The old way of one set of animals which were also barbarians worked but had a few areas which were not the way we wanted them in game. With the three animal nations we are just finding our feet. The two problems I have with the animal option I have tried is that choosing it or not gives extreme results ie there is either 10 animals per plot or only 1 animal per 100 plots. Neither is a good result and we need to look at ways to fix it. I have only been trying the option which has the animals fighting each other or not.

I agree, lets find our feet before we do anything rash at this point.

Have anyone any idea what may cause the strange city bug shown in this picture?? Why is it that only 1 of the NPC's (barbs) can have cities without bugs when they are set up more or less the same way in xml.
I'm not implying here that animals should have cities in any capacity, they were included in the picture because they are technically in the same boat.

 

[strategyonly]

It's IMO way too early to jump to the conclusion that it can't be fixed.

See if playing without Neanderthal cities helps.

Dont know what ur talking about here, all i asked if my CTD's can be fixed??