Why We Won't Have Larger Map Sizes

[Architechnid 186]

Well, I for one don't see why the 'Region' needs to be off limits, particularly now that it's offline.  Let us make the region ourselves.  Let us place freeways ourselves, even within the city.  Developers of a sandbox game should be less worried about making a game, and more worried about making the tools.  There is no reason the modding community shouldn't be given every tool the developers have to make the world as we see fit.  Even if city size is limited due to performance issues, (and I'm not fully convinced that is the case, particularly as time goes on and computuers get more powerful), then the conceptual issues the current game has would be mitigated at least.  This model worked for previous iterations, and it would have doubly worked for this one.

[ebalosus 19]

 

 

Theoretically quadrupling the size of the simulation would also quadruple the CPU usage, (not scientific) which would put my CPU at 144% of its limit. 

 

Uh Wut?!

 

That's not how programming works. I actually posed this question to friends whom are embedded software engineers, and they said that no, quadrupling the size would not quadruple the resource requirements.

 

Would an increase in simulation size increase the required resources? Yes, but not to that level. Given how many corners the simulation itself cuts, I wouldn't be surprised if quadrupling the simulation size would only require 1.5 times the current amount of resources.

[Darth Apple 54]

Who knows? It's quite possible that now that they've milked the whole online, multiplayer concept enough now, that they will open things up in the coming months. Time shall tell, but I do hope that they will listen to the community on this one. 

[humus 0]

 

 

 

Theoretically quadrupling the size of the simulation would also quadruple the CPU usage, (not scientific) which would put my CPU at 144% of its limit. 

 

Uh Wut?!

 

That's not how programming works. I actually posed this question to friends whom are embedded software engineers, and they said that no, quadrupling the size would not quadruple the resource requirements.

 

Would an increase in simulation size increase the required resources? Yes, but not to that level. Given how many corners the simulation itself cuts, I wouldn't be surprised if quadrupling the simulation size would only require 1.5 times the current amount of resources.

 

 

why not make the "CLOUD"  help simulation process ?

ah right, the "cloud" dont work that way   

 

Either way, I think GlassBox as city simulation is broken if it failed to scale.

with that kind of limitation, a fate has been sealed

[flowmotion 24]

 

 

 

Uh Wut?!

 

That's not how programming works. I actually posed this question to friends whom are embedded software engineers, and they said that no, quadrupling the size would not quadruple the resource requirements.

 

Would an increase in simulation size increase the required resources? Yes, but not to that level. Given how many corners the simulation itself cuts, I wouldn't be surprised if quadrupling the simulation size would only require 1.5 times the current amount of resources.

 

 

Without knowing the internals of how the simulation works, it's really useless to speculate IMO. It could scale sub-linearly, or it could scale super-linearly, we have no idea.

[The_Balm 4]

 

 

 

Uh Wut?!

 

That's not how programming works. I actually posed this question to friends whom are embedded software engineers, and they said that no, quadrupling the size would not quadruple the resource requirements.

 

Would an increase in simulation size increase the required resources? Yes, but not to that level. Given how many corners the simulation itself cuts, I wouldn't be surprised if quadrupling the simulation size would only require 1.5 times the current amount of resources.

 

 

Without knowing the internals of how the simulation works, it's really useless to speculate IMO. It could scale sub-linearly, or it could scale super-linearly, we have no idea.

 

Agreed, we also don't know how much of the CPU's resources are being used for overhead that wouldn't need scaling. 

[calm_blue_sea 3]

Although performance is certainly a limiting factor for many users, I think the traffic simulation places an ultimate limit on the size of the city tiles that can support viable cities. As we all know, Sims look for new jobs and new homes every day. Rather than dispersing, all the Sims follow the same path, resulting in a 'conga line' of vehicles. I can run a city with a large populace on my computer without any appreciable slow down - however, traffic becomes a problem. Even though I plan my cities (in an often unintuitive fashion) to fit the simulation, I am certain that traffic would grind to a halt if the cities were any larger. As the population increases with increased city size the length of the 'conga line' will increase to the point where the Sims are unable to find a residence or a workplace during a traffic cycle. If this happens, buildings will be abandoned and the city will die. I suspect when the game goes offline modders will be able to increase the tile size. If you have a powerful computer you may find that your frame rates don't drop much with larger cities, however that becomes a rather moot point if none of the Sims in your city are able to move.

[USA_Pride 49]

 

 

 

 

Theoretically quadrupling the size of the simulation would also quadruple the CPU usage, (not scientific) which would put my CPU at 144% of its limit. 

 

Uh Wut?!

 

That's not how programming works. I actually posed this question to friends whom are embedded software engineers, and they said that no, quadrupling the size would not quadruple the resource requirements.

 

Would an increase in simulation size increase the required resources? Yes, but not to that level. Given how many corners the simulation itself cuts, I wouldn't be surprised if quadrupling the simulation size would only require 1.5 times the current amount of resources.

 

 

why not make the "CLOUD"  help simulation process ?

ah right, the "cloud" dont work that way   

 

Either way, I think GlassBox as city simulation is broken if it failed to scale.

with that kind of limitation, a fate has been sealed

 

 

Exactly.  The "CLOUD" never did a thing for us other than hold our cities, something that our massive hard drives could do with ease and at a faster speed.

 

Glassbox is broken.  If it's the reason city sizes are why they are, then it's a failure.

 

 

Here's another possible explanation for small city sizes:  The bandwidth and disk space required to host cities in the "CLOUD".  Perhaps localizing the storage of the cities is the first step to larger city sizes for the offline mode only?

[jaredh 195]

There is a lot of talk in this thread about software scaling.

 

Software typically scales in three ways: 

 

1)  O(n) - linear growth.  Your traditional doubling/tripling/1.5 times etc

2)  O(log n) - logarithmic growth.  This is what you want...

3)  O(n²) - exponential growth (VERY BAD)

 

There are other scaling values (see http://www.javacodegeeks.com/2011/04/simple-big-o-notation-post.html?ModPagespeed=noscript) but they usually don't  describe a program overall)

 

An entire program does not fit one scaling method...one piece of the code might scale O(n) and the next piece scales at O(log n).  But some loops will generally dominate a program and thus their scaling will tend to impart the overall scaling.  That's why cubic scaling happens, but usually isn't observed at the overall program level.

 

Scaling probably was not the main problem here.  The problem was the language it was coded in (Java) and it's inherent slowness.  Java is just awful (when compared to other options) when the scaling gets worse than linear.  It works great if you are not thrashing the CPU (like a web page doesn't), but if you start thrashing Java, it has scaling problems, and it always has.  It used to be FAR worse than it is today, but they could not get rid of it entirely.  Java slowness can be overcome with big servers...but when you go from an Xeon 16 core with 200gb of memory and an insane backplane to a Core 2 Due with 2gb of memory...well Java chokes there with the same workload.

[The_Balm 4]

There is a lot of talk in this thread about software scaling.

 

Software typically scales in three ways: 

 

1)  O(n) - linear growth.  Your traditional doubling/tripling/1.5 times etc

2)  O(log n) - logarithmic growth.  This is what you want...

3)  O(n²) - exponential growth (VERY BAD)

 

There are other scaling values (see http://www.javacodegeeks.com/2011/04/simple-big-o-notation-post.html?ModPagespeed=noscript) but they usually don't  describe a program overall)

 

An entire program does not fit one scaling method...one piece of the code might scale O(n) and the next piece scales at O(log n).  But some loops will generally dominate a program and thus their scaling will tend to impart the overall scaling.  That's why cubic scaling happens, but usually isn't observed at the overall program level.

 

Scaling probably was not the main problem here.  The problem was the language it was coded in (Java) and it's inherent slowness.  Java is just awful (when compared to other options) when the scaling gets worse than linear.  It works great if you are not thrashing the CPU (like a web page doesn't), but if you start thrashing Java, it has scaling problems, and it always has.  It used to be FAR worse than it is today, but they could not get rid of it entirely.  Java slowness can be overcome with big servers...but when you go from an Xeon 16 core with 200gb of memory and an insane backplane to a Core 2 Due with 2gb of memory...well Java chokes there with the same workload.

 

I really don't think Java is a problem. It doesn't seam to really do much in Java.

[GrayAnderson 6]

 

The same applies as to what was said then...the game's design is fundamentally flawed. They simulated the wrong aspects and never made an engine capable of scaling.

Poor game design, buffing computers with new powah won't allow for larger maps as the software simply won't scale.

 

I agree.

 

I appreciate Catman's analysis, but it's their job to either code around it, or make compromises to have a more fun game.  The current city size is pathetic and laughable.  The game should be called "SimTown", not "SimCity".  If the game is only 32-bit, then make it 64-bit, or code it to use RAM more efficiently.  If the CPUs are busy making all these calculations, either make the calculations more efficient, or compromise and make less of them so that city size can be increased.

 

The GlassBox engine wasn't all that it said it would be.  It's cool to be able to see some of the game mechanics in the different data views, but the majority of the game is spent with these data views hidden.  What we do see from the very moment we start a new city is a miniscule city size.

 

In SimCity 2013, you spend your entire time compromising your city design to fit onto the microscopic canvas we've been given.  Your dreams are muted and scaled from day one.

 

Too bad they didn't include the gingerbread houses, playgrounds, and candy stores.

 

(For anyone missing the reference, there was in fact a "SimTown" made in the early 90s by Maxis, aimed at kids)

 

All joking aside, I tend to agree that something is "off" about Glassbox.  The "search for jobs every day" principle never made sense to me in terms of realism...but now that I think more about what's being said here, it does present potential issues in terms of city size.

 

Edit: To save a post, the other thought here is that "I want to play the region, not a city within it".  That was my gripe about SC4 as well...I don't like having to back out and re-enter cities as part of developing a region, worry about screwing up a region because I changed one city, etc.

[SAGGG 47]

Ooh, I like USA_Pride's logic...

[Boniuss 0]

@OP - Resource monitor from a low-/midrange laptop doesn't really say anything. My 3rd gen stationary runs this in power-saving mode and my old dustbin from 2005 runs it at the same resource usage as you.

 

The reason we won't get bigger maps is because somewhere in the early stages of development, someone came up with the brilliant idea of tying the DRM to the backend and market it as cloud computing. That's the one and only reason we won't get bigger maps (atleast not officially).

 

They simply went halfway with creating an MMO and left out all the important bits and pieces. Afterall; How come EVE/WoW/PS2/GW2/etcetcetc can handle billions upon billions of entities, simlulations and calculations in real-time, and SimCity can't?

  Edited January 17, 2014 by CaptCity  

[ebalosus 19]

There is a lot of talk in this thread about software scaling.

 

Software typically scales in three ways: 

 

1)  O(n) - linear growth.  Your traditional doubling/tripling/1.5 times etc

2)  O(log n) - logarithmic growth.  This is what you want...

3)  O(n²) - exponential growth (VERY BAD)

 

There are other scaling values (see http://www.javacodegeeks.com/2011/04/simple-big-o-notation-post.html?ModPagespeed=noscript) but they usually don't  describe a program overall)

 

An entire program does not fit one scaling method...one piece of the code might scale O(n) and the next piece scales at O(log n).  But some loops will generally dominate a program and thus their scaling will tend to impart the overall scaling.  That's why cubic scaling happens, but usually isn't observed at the overall program level.

 

Scaling probably was not the main problem here.  The problem was the language it was coded in (Java) and it's inherent slowness.  Java is just awful (when compared to other options) when the scaling gets worse than linear.  It works great if you are not thrashing the CPU (like a web page doesn't), but if you start thrashing Java, it has scaling problems, and it always has.  It used to be FAR worse than it is today, but they could not get rid of it entirely.  Java slowness can be overcome with big servers...but when you go from an Xeon 16 core with 200gb of memory and an insane backplane to a Core 2 Due with 2gb of memory...well Java chokes there with the same workload.

 

I agree with this except for the part about Java. I'm not disagreeing with you about Java (I used to be a Java programmer myself), but about what you said regarding it and Simcity.

 

According to the blogpost about offline mode, the parts of the game that were programmed in Java were the parts that interacted with the online infrastructure, and that they would  have to rewrite all those parts into C++ for the game to run in offline mode...which implies that what you said regarding Java and scalability may infact be solved by them rewriting a lot of the code into C++

[The_Balm 4]

The reason we won't get bigger maps is because somewhere in the early stages of development, someone came up with the brilliant idea of tying the DRM to the backend and market it as cloud computing. That's the one and only reason we won't get bigger maps (atleast not officially).

 

I don't think that's why the maps are small and that doesn't even make sense. And they're working on offline support, so... Their servers aren't the problem.

[Noobi-Wan 22]

I have a feeling that we will find out soon enough after offline comes as to the practical implications of having a larger city plot. This is the most in demand addition to the game so I imagine there is going to be a lot of pride attached to being the one who cracks it.

I think the bigger problems will come from having to remap water tables, resource deposits and the other overlays to not just cover the 2x2 plot but the entire region. I don't know what sort of impact this would have on performance but I cannot imagine there are no system resources allocated to it.

This would mean having access to the tools used to create the maps in the first place which would include the road laying at a region level etc.

Regardless of whether the larger maps debate is ever resolved there have to be other ways of fixing the problems that we have with the small maps.

In some ways I would like the option to put in a few smaller cities of a custom shape to fill out the spaces in the region or to have the smaller suburbs spring up around the "large" city centres.

[hlaalu 0]

I want to chime in here. The reason they didn't give us larger maps is simple: they wanted to "socialize" the game. By limiting the map sizes, they force us to rely on others to help out with resources, special buildings, etc. And it works. I'm constantly finding myself trying to get other people to play with me because it really is a PITA switching from city to city, trying to remember what is where. Players become their best marketing tool.

 

IF they really, truly wanted to, they could have made larger maps an OPTION for those of us with high end machines.

 

I'm really rooting for all the smart modders out  there to figure out a way to make larger maps possible when offline mode turns up.

[CaptCity 4,868]

Let's remember that modding is not the same as 'cracking' or altering game files which would be against user agreements. As per site rules, discussion of these types of topics is not permitted.

[Eriseley 4]

I'm not sure what I think. The decision trees and behavior of agents seem incredibly simple, yet on the other hand the fudging of population numbers compared to actual agents seems a clear decision to prevent an explosion of agents to process.

 

I do think a lot of the GlassBox engine would kinda fail with larger maps, from water and energy not reaching sections of city to mass transit vehicles never hitting stops and that kind of thing.

[Sleeping Politician 32]

Part of the issue is that Maxis will not optimize the GB engine to work with larger maps on low end machines.

[Telles0808 0]

Well, if the first topic is right, this should not be possible: http://www.reddit.com/r/SimCity/comments/1d77t7/mod_pack_build_outside_city_box_simroller/

 

If it's not possible, how it happened?

 

Bigger towns (not cities, this is SimTown, remember?) are a simple question of data storage in my mind.

 

Also, they have many workarounds if they wanted add this, like being possible to buy expansion borders to your town, where you could go visit and build.

 

I really hate how near my industry is from my commercial and residential area, specially because of pollution and other things.

 

Plus, this is looking like an aquarium of ants, and not a town.