v22 bugs and crashes

I followed your instruction and it still crash

Here's an new minidump (build 2410) and the save

It's a new game and I was trying to move the stonethrower, but in later games it appends with different kind (melee, archery, ships, workers, GP).
In fact, since this bug appears (1 week or so) i don't remember having on my first turn after starting C2C a move that don't make it crash.
To avoid it I save at the end of a turn or I use the keyboard to make the moves

So to make it crash after loading this game I need to move the stone-thrower where? Since you say the workaround is to use the keyboard I assume I have to use the mouse? Just right-clicking on any tile with the stone thrower selected?
 
this is really getting wierd, i went back to ALL the savedgames and it said "I was defeated."HUH! Plus it said that the name was Cyrus i was using, which i used Nazi Himmler LH?:confused:

Luckily i manually saved a few game, but now i cant get anything to happen as a CTD:crazyeye:

I get the Rev (2nd pic) to stop by at the same turn it happens, to re-load the saved game fro that point, that way no Rev occurs.

EDIT: Even getting weirder. i let it go this time for the Rev, it defeated that Hunter, but then the game had a bubbling sound then stop then again, but the game froze and i had to Alt/Ctrl/Delete.

OK tried again to go back and get the autosaved game but it came back again (see pic 3) then pic 4

Going to try the manual one again??

My guess would be that the save format doesn't load across changes in the assets that chnage leader ids correctly. So if you've added LHs since you saved it probably bugs out.
 
Thanks for reply. I am using 64 bit Vista atm, i wrote it. My post was regarding possible memory handling by Vista and win7, so that win7 might somehow fix errors occuring in vista64

Well... Considering that Vista isn't really any good I think it's worth trying anyway, in fact if you are running 64-bit anyway I'd probably change to Win7 64 completely instead of holding on to Vista.

Cheers
 
So to make it crash after loading this game I need to move the stone-thrower where? Since you say the workaround is to use the keyboard I assume I have to use the mouse? Just right-clicking on any tile with the stone thrower selected?

Just one right click to select it then hold the right button to move it.

take any direction it should CDT
 
2.7 gig memory usage by civ4 - then crash

savegame attached for Koshling, tho i don't think it's related to this particular game, as i have exactly same errors for last few months.

Updated video drivers today.
I dunno, what other possible options are there for memory usage by programs besides 3gb switch (which doesn't help me since i use vista 64)?

Well, not really good news. I've pinned down where this is happening. Basically every turn about another 200M gets allocated. It gets allocated in this call:

Code:
gDLL->getEngineIFace()->DoTurn();

which is a callback to the main game engine, over which we have no control. That routine does something that allocates this 200M or so in code we don't have access to. Only about 1% of the per-turn allocation is happening inside our DLL. I'm going to spend the rest of today looking for workarounds or ways to mitigate this, but basically it may just be that we've hit some limit in Civ4 that can't be gotten around as long as we are tied to Civ4.

Your game exhibits it FAR worse than my games of comparable size though, so I suspect there is something unusual in your game going on, though I don't know what. I'll edit this post with further information if I get anywhere trying to find a way round it.
 
Well, not really good news. I've pinned down where this is happening. Basically every turn about another 200M gets allocated. It gets allocated in this call:

Code:
gDLL->getEngineIFace()->DoTurn();

which is a callback to the main game engine, over which we have no control. That routine does something that allocates this 200M or so in code we don't have access to. Only about 1% of the per-turn allocation is happening inside our DLL. I'm going to spend the rest of today looking for workarounds or ways to mitigate this, but basically it may just be that we've hit some limit in Civ4 that can't be gotten around as long as we are tied to Civ4.

Your game exhibits it FAR worse than my games of comparable size though, so I suspect there is something unusual in your game going on, though I don't know what. I'll edit this post with further information if I get anywhere trying to find a way round it.
What happens if that call is commented out? Or rather, what does not happen then?
 
What happens if that call is commented out? Or rather, what does not happen then?

I tried that. Whatever housekeeping it does 'all at once' seems to get done anyway between the next few calls to CvGame::update() so the same memory footprint results, just over a couple of seconds rather than all at once. I'm wondering if something bad is happening in the game engine when the total number of graphical entities exceeds 32K (or 64K) or somthing...
 
My guess would be that the save format doesn't load across changes in the assets that chnage leader ids correctly. So if you've added LHs since you saved it probably bugs out.

Nope didnt add nothing to the SVN just playing as the Nazi's is all, nothing else was done:rolleyes: Didnt change anything ingame while playing either.
 
Great work, Koshling! I cross my fingers for you :)

Another bug or imbalance - in my current game Heke (Maori) bribes my capital, with troops inside (troops stay loyal to me), while we have peace treaty (23 turns left), so i attack him back automatically. After that happens i can move troops out of city but to get back i need to take down defences ofc.

There is more - his spy points on me remain unspent in Espionage Screen before and after that they are same.

I don't feel that kind of action looks realistic.

Here is savegame right before that happens.

p.s. Is there any way i can turn off city bribing, or would i have to manually erase his spies in WB?
 
Great work, Koshling! I cross my fingers for you :)

One thing that is apparent from your save game is that the rate at which it is leaking memory is VERY high. ANY save would MAF within 5 or 6 turns with that rate of leakage. That means that whatever the condition is that triggers this situation must have occurred recently in your game (or else you'd have been MAFing very few turns long ago). Can you think of anything significant that has happened in this game recently? The kinds of thing I have in mind are things like:
  • circumnavigation
  • map trading to get visibility on more or less the entire world
  • first unit with flight
  • some new unit category became available
  • ...

Do you have saves of the same game from somewhat earlier (ideally just bedore you started to get MAFs a lot) ?
 
One thing that is apparent from your save game is that the rate at which it is leaking memory is VERY high. ANY save would MAF within 5 or 6 turns with that rate of leakage. That means that whatever the condition is that triggers this situation must have occurred recently in your game (or else you'd have been MAFing very few turns long ago). Can you think of anything significant that has happened in this game recently? The kinds of thing I have in mind are things like:
  • circumnavigation
  • map trading to get visibility on more or less the entire world
  • first unit with flight
  • some new unit category became available
  • ...

Do you have saves of the same game from somewhat earlier (ideally just bedore you started to get MAFs a lot) ?

From that list i did circumnavigate (after i traded for cartography and traded maps, which of course revealed most map), i produced air balloon, new unit category coming in mind is hot air balloon :)

I have *feeling* (from previous games with often MAFs) that this starts right after i discover most of the map by map trading.

I did indeed start MAFing long ago :) I just was restarting game often due to elevated patience from RoM games on gigantic with 15 min turn times :)

I have some earlier saves (not that often as i was mostly using quicksave and autosaves (autosaves are all overwritten ofc)), i understand what you mean, and i will try to find it. If i won't be able to do so with saves from current game, in my next game i will take close look at the moment such things start to happen.
 
From that list i did circumnavigate (after i traded for cartography and traded maps, which of course revealed most map), i produced air balloon, new unit category coming in mind is hot air balloon :)

I have *feeling* (from previous games with often MAFs) that this starts right after i discover most of the map by map trading.

I did indeed start MAFing long ago :) I just was restarting game often due to elevated patience from RoM games on gigantic with 15 min turn times :)

I have some earlier saves (not that often as i was mostly using quicksave and autosaves (autosaves are all overwritten ofc)), i understand what you mean, and i will try to find it. If i won't be able to do so with saves from current game, in my next game i will take close look at the moment such things start to happen.

ok, thanks. I'm making some progress I think, but still too early to really say for sure.
 
ok, thanks. I'm making some progress I think, but still too early to really say for sure.

Yes i was correct to trust my feelings :)

Attached is savegame the turn i discovered Cartography. Before that i had game ran for 20 or so turns (some auto, some manually), memory usage increased around 5-20 Mb per turn. I didnt scroll around map much tho.

After i trade for maps (havent traded in savegame yet, but you can), and next turn circumnavigated globe (when that happened and camera zoomed out to globe view i had system stall and stop respond for 30 or so seconds, i have low graphics, globe view buildings disabled, and that option to only enable wonders visible on map, not regular buildings).

After that i ran 2 more turns, during which civ used 250 mb of RAM. I scrolled around a bit.

Hope that helps.
 
@Koshling: Do you have any plans to do memory optimization in the near future?

My main point in asking this is that adding the Multiple Maps is going to mean a 10 to 15% increase in memory, according to the maker of that modcomp, and not all of us are blessed with 64 bit operating systems.:rolleyes: I think that turn times are pretty good atm, so memory might be a good place to start making improvements.
 
@Koshling: Do you have any plans to do memory optimization in the near future?

My main point in asking this is that adding the Multiple Maps is going to mean a 10 to 15% increase in memory, according to the maker of that modcomp, and not all of us are blessed with 64 bit operating systems.:rolleyes: I think that turn times are pretty good atm, so memory might be a good place to start making improvements.

I did a major pas at memory optimization a couple of months ago. At this point the Civ4 engine itself massively dominates memory usage, and there may be very little more we can do. I'm trying some fairly radical things currently to se if I can help Talin's issue, which might have some spinoff, but TBH I don't think there is anything more we can substantially do in regard to memory footprint.

I'm also concerned about multi-maps in this regard. However, we might simply have to accept that one of the following has to go:
  • Multimaps
  • Large maps
  • Graphical variance in terrain (which will actually be compunded by multimaps due to using disjoint terrain types)
  • Civ4 as the base engine
 
I did a major pas at memory optimization a couple of months ago. At this point the Civ4 engine itself massively dominates memory usage, and there may be very little more we can do. I'm trying some fairly radical things currently to se if I can help Talin's issue, which might have some spinoff, but TBH I don't think there is anything more we can substantially do in regard to memory footprint.

I'm also concerned about multi-maps in this regard. However, we might simply have to accept that one of the following has to go:
  • Multimaps
  • Large maps
  • Graphical variance in terrain (which will actually be compunded by multimaps due to using disjoint terrain types)
    [*]Civ4 as the base engine

This! No idea how that would interfere with copyright issues and resourses/time needed for replacement, but that would be most logical direction, imho.
 
I did a major pas at memory optimization a couple of months ago. At this point the Civ4 engine itself massively dominates memory usage, and there may be very little more we can do. I'm trying some fairly radical things currently to se if I can help Talin's issue, which might have some spinoff, but TBH I don't think there is anything more we can substantially do in regard to memory footprint.

I'm also concerned about multi-maps in this regard. However, we might simply have to accept that one of the following has to go:
  • Multimaps
  • Large maps
  • Graphical variance in terrain (which will actually be compunded by multimaps due to using disjoint terrain types)
  • Civ4 as the base engine

What does the Civ4 base engine do with the memory, besides the graphics?

If we had to get rid of one of those four, I'd choose the Terrain graphical variance. Personally, I never zoom in to the point where that would even be noticable, and it has no actual effect on gameplay.
 
What does the Civ4 base engine do with the memory, besides the graphics?

If we had to get rid of one of those four, I'd choose the Terrain graphical variance. Personally, I never zoom in to the point where that would even be noticable, and it has no actual effect on gameplay.

That's all really, but in Talin's game 2G of the 2.7 used is Civ4 engine graphics wastage. It just handles them inefficiently.

When I said terrian variance I didn't mean terrain texture detail (though reducing that might help a little, but it's marginal), I meant number of different terrain graphic types (which is currently the same thing as number of terrain types).

This! No idea how that would interfere with copyright issues and resourses/time needed for replacement, but that would be most logical direction, imho.

While I agree that its logical, it would consume several people for a year or so, even if we had the right skill set to do it (which we don't). Since the Civ5 engine is being released soon, it would be a logical half-way-house to migrate to, I would think, in this option (though who knows what foibles it has!).
 
That's all really, but in Talin's game 2G of the 2.7 used is Civ4 engine graphics wastage. It just handles them inefficiently.

When I said terrian variance I didn't mean terrain texture detail (though reducing that might help a little, but it's marginal), I meant number of different terrain graphic types (which is currently the same thing as number of terrain types).



While I agree that its logical, it would consume several people for a year or so, even if we had the right skill set to do it (which we don't). Since the Civ5 engine is being released soon, it would be a logical half-way-house to migrate to, I would think, in this option (though who knows what foibles it has!).

Do we have any idea what the prime graphics culprits are, in terms of buildings vs units vs terrain?
 
Back
Top Bottom