Multiple Maps and Mapscripts

[Koshling]

Nice! How does the non-graphics memory usage scale though? Is it by map size, number of cities, or a combination of the two?

Num units, num cities, an element of map size. However the scaling constant is a lot lower than the graphics element so there's much more headroom.

 

[KingArthur666]

Preliminary result of viewport usage in terms of Civ memory usage. Please note that this is one data point from the port to C2C, and I'm still debugging so the code is not yet finalized. However...

....I loaded an industrial age GEM game I happened to have lying around, and after game load the memory usage without viewports enabled was 1.4G. With viewports enabled (using a 50 X 50 viewport) it's 900M. Since the base before any map laoding is around 600M this means the map-related contribution dropped from around 800M to about 300M. Furthermore it should be fairly stable at that sort of level for ANY map size.

in'! Can't wait til this is in so I can start playing with larger maps.

 

[Cyrusfan]

Can the viewport size be varied by circumstance? What I mean is, 60x60 or whatever is good for giving orders to units, depending on how far you want to move them, but when I'm watching AI moves/attacks on my units, or giving cities build orders, it seems to me that something around 7x7 to 11x11 would be enough to show me what I need to see while saving some time drawing.

 

[Koshling]

Can the viewport size be varied by circumstance? What I mean is, 60x60 or whatever is good for giving orders to units, depending on how far you want to move them, but when I'm watching AI moves/attacks on my units, or giving cities build orders, it seems to me that something around 7x7 to 11x11 would be enough to show me what I need to see while saving some time drawing.

No. So far we have found no way to tell the game engine that the map size (as far as it is concerned, since it thinks the viewport IS the map) is changing within a game, so the constraint is that viewport size is fixed within any one load.

 

[Cyrusfan]

Hmm. Is there a way you could lie to the game engine during AI turns about what it should be drawing outside the immediate neighborhood of the action? For example, if you momentarily convinced it that such beyond view territory was still hidden by the initial fog of war, might that cut down on the times involved?

 

[Koshling]

Hmm. Is there a way you could lie to the game engine during AI turns about what it should be drawing outside the immediate neighborhood of the action? For example, if you momentarily convinced it that such beyond view territory was still hidden by the initial fog of war, might that cut down on the times involved?

That's what the existing option to use dynamic entities does. It works well in the early game by in the late game, when everything is revealed, it doesn't help much.

 

[Koshling]

I'm running (significantly) late for a number of reasons:

• Work has been busy this week, so I've had less C2C time
• C2C is a significantly more complex environment than the basic BtS of the parallel maps mod, and interactions with other changes are making the port harder than expected
• Now I'm back in C2C, I have access to a large library of matrure game saves as extended tests, which I didn't have in the parallel maps mod. Those extended tests are showing up further issues that need fixing

I'll need another week.

I think you should release V24 without it, since it would be default disabled anyway, and it'll be available on the SVN in a week or so for those that want to enable it and try it.

 

[ls612]

I'm running (significantly) late for a number of reasons:

• Work has been busy this week, so I've had less C2C time
• C2C is a significantly more complex environment than the basic BtS of the parallel maps mod, and interactions with other changes are making the port harder than expected
• Now I'm back in C2C, I have access to a large library of matrure game saves as extended tests, which I didn't have in the parallel maps mod. Those extended tests are showing up further issues that need fixing

I'll need another week.

I think you should release V24 without it, since it would be default disabled anyway, and it'll be available on the SVN in a week or so for those that want to enable it and try it.

That's OK, we all know how much work you do for C2C. The very fact that this is even possible is amazing, so I'm OK with the wait.

 

[strategyonly]

I'm running (significantly) late for a number of reasons:

• Work has been busy this week, so I've had less C2C time
• C2C is a significantly more complex environment than the basic BtS of the parallel maps mod, and interactions with other changes are making the port harder than expected
• Now I'm back in C2C, I have access to a large library of matrure game saves as extended tests, which I didn't have in the parallel maps mod. Those extended tests are showing up further issues that need fixing

I'll need another week.

I think you should release V24 without it, since it would be default disabled anyway, and it'll be available on the SVN in a week or so for those that want to enable it and try it.

OK thx for letting us all know what is going on, this realy really helps out, and again many many thx.

 

[Thunderbrd]

I say we wait for v24. I have a lot coming in soon too and I'm racing to try to get it done in time (but done right too...)

 

[strategyonly]

I say we wait for v24. I have a lot coming in soon too and I'm racing to try to get it done in time (but done right too...)

No offense, but then wont that mean it needs ALOT of testing to be done, if so, maybe some now, some later after v24, just upload some of it as non-working stuff, so you dont lose it, then after v24 change it to a working copy that way "we'll" have alot of time to test stuff out, hows that sounds, good/bad?

Its YOUR choice either way, its YOUR stuff!!

 

[Thunderbrd]

I take no offense. I do expect some playtesting to reveal issues needing some adjustments. If you are wanting to get v24 moving, then I have no objection to its release and will simply strive to make sure that v25 has a polished version of what I'm modding here. Its not as if it matters much to the SVN version anyhow, so yeah, it could be good to just use the v25 development run to smooth everything out on my projects here.

 

[ls612]

@Koshling:

How much work will it be to port the actual Multi-Map stuff to C2C once LyTning has it polished? I know that it will require a good bit of interface and terrain work, but would it be feasible to see multiple maps during the pre-V25 development? How is LyTning's work on that front going?

 

[Koshling]

@Koshling:

How much work will it be to port the actual Multi-Map stuff to C2C once LyTning has it polished? I know that it will require a good bit of interface and terrain work, but would it be feasible to see multiple maps during the pre-V25 development? How is LyTning's work on that front going?

Easy I think. The hard part is the graphics stuff and that all happens in the viewports. The rest, although complex, is entirely within the DLL and therefore within our control. The problem with the graphics stuff is that you continually have to second-guess what the engine is doing since we don't have access to its source to see. For example the problem I am currently stuck on is that after loading a mature game (and not every game, but happens quite a lot) the game engine calls asking for details of a plot, passing NULL for the plot pointer. Now it must have gotten that pointer from us origianlly in response to SOME request asking for a plot (presumably on outside of the viewport which is why we return NULL), but trying to pin down why is proving very hard - its not any call it makes any place close to the point at which it makes the fatal call with the NULL pointer. Since it isn't theoretically even aware of plots outside of the viewport the issue is tarkign dow how it got hold of it. Up to that point it has made just over 2 million calls to the DLL, so (given its not one made temporarily locally) it's a bit of needle in haystack problem. Anyway, trying to manipulate the engine is always like this, since we have zero idea of its inner workings, and therefore building up a model to predict/explain what it will do is not easy.

 

[strategyonly]

Thought i would this since alot are asking about it lately?

 

[Koshling]

Thought i would this since alot are asking about it lately?

I have not heard from Lytning for a month or so, and there has been no activity on the Parallel maps SVN since I did the viewport work, so I don't think Lytning has had time to do any work on this yet.

 

[ls612]

I have not heard from Lytning for a month or so, and there has been no activity on the Parallel maps SVN since I did the viewport work, so I don't think Lytning has had time to do any work on this yet.

Didn't MrAzure's friends say that they were developing Multi-Maps independently? Are they even still around, or did that whole thing all fizzle out. At any rate, I have also noticed much enthusiasm over Multi-Maps, but I don't think that we should release them until we have some more AI and performance improvements.

 

[Koshling]

Didn't MrAzure's friends say that they were developing Multi-Maps independently? Are they even still around, or did that whole thing all fizzle out. At any rate, I have also noticed much enthusiasm over Multi-Maps, but I don't think that we should release them until we have some more AI and performance improvements.

They did say that. They also said they would be releasing it the day after V25 went live, so I guess we'll know soon enough.

 

[ls612]

Now that we have Viewports in and mostly working and it appears that MrAzure's team will not be making their own version, I wanted to ask when we will start making our version. Aside from the issues with floating improvements and missing unit flags, the Viewports feature is mostly complete. I personally think that we should wait until AIAndy gets back from vacation, as he will be the one that will most likely be making the UI. This would mean waiting until after V26 is released.

@Koshling

What needs to be done in the DLL to enable the actual multi-maps? As LyTning got back to you about his progress on that matter? Does getting started on it in V26 sound achievable?

 

[Koshling]

Now that we have Viewports in and mostly working and it appears that MrAzure's team will not be making their own version, I wanted to ask when we will start making our version. Aside from the issues with floating improvements and missing unit flags, the Viewports feature is mostly complete. I personally think that we should wait until AIAndy gets back from vacation, as he will be the one that will most likely be making the UI. This would mean waiting until after V26 is released.

@Koshling

What needs to be done in the DLL to enable the actual multi-maps? As LyTning got back to you about his progress on that matter? Does getting started on it in V26 sound achievable?

Quite a lot (as much again at least as viewports was). I have heard nothing from Lytning.

I also (personally) intend to be bogged down in AI for the next month or two, so I prefer not to be the primary mover in this (apart from where it intersects wtith viewports, so the (map) graphical side I'm happy to keep updated).