croxis

When I watched the glassbox gdc video the agent system sounded similar to entity-component design. If it is then the agents themselves are little more than a collection of variables and then a system loops through the agents its tracking each tick. Each system lends itself to threading fairly well. This means the traffic system could be on a different thread from the health system, with a little bit of boiler plate to make sure an agent isn't being accessed by two systems at the same time. What would be more difficult in this setup is splitting one system over two threads.

And 4 gigs of ram, minimum. I think the community is underappriciating that aspect. Curret games are often ports or multi platform meaning they are designed to run with low ram requirements (PS3 and 360 have only half a gig of ram) so the pc minimum requirements are usually around 2 gigs. Sim City requires at least 4. The problem with ram is that once it starts filling up the OS will start paging things to the hard drive. Eventually it will try and page the game itself if it eats up too much ram, and that is when the computer "locks up" with a thrashing hard drive. This is afar worse user expeirence than a choppy framerate from a slow cpu or gpu.

Ars's review is disapointed. I'm still waiting on rock paper shotgun but their past coverage has been cautious but optimistic.

Multithreading the phyics, sound, graphics, and simulation is a mostly trivial task. Multithreading the simulation itself is anything but. First multithreading makes code much more difficult to debug and adds another layer where bugs can live in. Multithreaded games are often compiled to be single threaded during testing. It would not surprise me if glassbox has multithreaded architecture but EA chose not to open that box of worms for its grand introduction. Multithreading the simulation might not make it faster either. Memory can only be accessed by one thread at a time. Things called locks and sephamores are used to protect memory in multithreaded stuff, one thread will lock out other threads that are trying to work on the same bit. Those other threads will stop what they are doing waiting for the lock to let go. This isn't much of an issue in programs like a web browser where the different tabs don't really need to talk to each other much, but in a game the performance hit from locks are much bigger.

[Removed reference to removed text] Increasing tile size isn't as simple a matter of adding #include "4kmcity.h" to the source, compile, and tile size are bigger with no problems. There are simulation, graphic, network, ui, and performance issues as well as balance issues and Q/A that have to be considered and planned for. I personally think many of these problems were allready addressed and take the devs at face value, that for the resolution of the simulation and how new the engine is the peformance penalties are just too much. It wil take improvements in technology, consumer "standard" computer, and iterations and optimization of the engine to get larger city tiles. Also, quick check of wikipedia is that the plans for the pentagon were not done in one weekend, but concurrent with construction where construction even outpased the plans. And yes EA is going to milk me for what they are worth. They are a pubiclly owned company and are beholden to their shareholders to max out their short term profits. This has almost always been the case, it is just less obvious in the consumer goods market. There are always exceptions (Nintento is a great example), but I am not convinced it has ever been the rule outside of local buisness.

You must use origin to play sim city.

oivaizmir?

Sure, I would love larger city tiles. I don't thing anyone is arguing against having larger tiles. For all the gnashing of teeth and boycotting the game "because EA" has the change you (the generic you, not a specific person) been looking for happened? No? So maybe another strategy is needed. I think people could be put into one of three groups on this issue: 1) People who don't care (and are often a big group, even the majority). They don't care so they don't post. 2) People who don't like the small tiles and go wailing on forums and youtube (the most vocal so they seem like the majority) 3) People who don't like the small tiles, probably grumble at first, but continue to investigate it, think about it, and see how it actually plays in beta and demos. At first it felt like the forums was mostly in group2, but after the demos and people given some time I think more and more are in group 3.

We have to be careful here. A game designed for dlc does not mean things were removed from the vanilla game. I recall reading a reddit thread where a number of developers talked about how dlc works. IIRC the main, vanilla game is given a set budget that pays for x developer hours across coding, sound, art, etc. Sometimes things go really well and there is time, and therefor money, to add in additional features. Usually things don't go according to plan and features have to be dropped in order to get core elements done in time and in budget. Expansions and dlc (usually) have their own budgets that pays for y developer hours. This is seperate from the main game budget. Often features and content which would of otherwise been left on the cutting room floor get their own budget and see the light of day as a bundled expansion or as dlc. This is a far cry from the conspiracy theory that developers pull out content just to make more money. I'm not saying that it doesn't happen, but I doubt it is widespead or I would see developers posting about it on reddit. Have you ever played Civilization 2 1.0? Or Outpost 1? Heck, even Super Mario World had slowdowns in some levels! Buggy games are nothing new.

I checked the wine app db and Origin is flagged as bronze/garbage.

no its not. The scales are different. 2km in this new simcity isnt the same as 2km in SC4. You can easily see that. AND sc4 offerered larger AND contiguous maps. And before you resspond that the new Sim City is new and dynamic in that the job only starts with filling the map- the same would apply to a larger map, requiring even more time and attention than the puny map provided. I can't just not mention the simulation engine though. It's the entire reason for the smaller map size. SC4 is only larger in 2dimensions. But it's flat. Sc2013 isn't flat. Is a 50" monitor displaying 640x480 resolution really bigger than a 24" monitor displaying 1600x1200 resolution? I've watched the betas. Those maps were filled with ease. Services were included. Sure a few things were locked. That may add a few hours. And before we go back on nuances, in the end sc4 gave the player more time per region than this thing. And ill even have the argument one hand behind my back. Im not even asking Sc2013 have the map space. Im just asking that it not be stupidly small. If your so content with this size as youll argue as its sufficient to no end then dont play larger maps if they offer then, as what is there now is more than adequate for you. They have mentioned in recent interviews/tweets/reddit that the final game balance will be a lot longer to fill up (road costs were increased as one example). The reason why the beta was so quick to fill up is they wanted to get as much data as possible during the beta. The other option would be to keep the realistic scale and no time limit, but fewer people would make it to that stage.

I don't think there is a lot of simulation going on, mostly telling cities what resources there are from other cities and the world, and moving said resources around. 1) I think EA wants an even playing field no matter what system you are running. Think the leaderboard. 2) It wouldn't surprise me that EA mandated it. However the devs did say they also want the next sim city to be very connected with the world to reflect the more modern times. To be honest I think Maxis did a really good job making the online system meaningful despite the limitations. 3) It would be hard from a graphical and gameplay perspective. First you would at least quadrupal the number of things to render. Eight times if you count the corners. Then the simulation. If a sim enters from a neighboring city and you have 8 road connections, how should the simulation decide which road the sim enters on? Sure one sim doesn't matter, but if there is a lot it may not be in a realistic pattern.

Random muse: What makes a feature an essential feature in SimCity? Think about the code that computed land value in sim city classic (which also used grid-based simulation even though the grid wasn't visibily drawn). It was a formula that took in parks, pollution, crime, etc and calculated land value. I'm sure quite a bit of time was spent balancing the forumula so it made sense and was fun to play the game with. SimCity2k came around. They could take the code from sim city classic, which was already playtested and balanced, and just tweek it. They could then use the extra time and resources to add new features. This brings us to SC4 which had over a decade of playbalance, code, and testing. EA/Maxis didn't want to use the grid simulation for SC2013. That means all of that playtesting, code, and knowledge is thrown out the window. Everything has to be redone from scratch. That is the cost of doing a simulation system that isn't on a grid. You want curvy roads? It comes at a price. This is the price. I really do think the simulation can scale larger before it breaks. I think the issue is that EA wants an even playing field due to the cooperative/competative nature (ie the leaderboard). Because of that they limited the size so the lower end machine and the higher end machine runs the same. This is just my hunch.

EA, for better or worse, didn't want players on lower systems to have unfair disadvantage due to the cooperative/competative connected nature of the game. Now that I am thinking about it I am not sure how to scale an agent based system using LOD simulation without risk of exploits along the literal edge cases. X3 does this and the results are often amusing or frusturating. It is very easy for LOD simulation to go wrong. Personally I blame Apple and the iPhone. They've created the expectation for consumers of always having a smooth user expeirence even at the cost of modern features (such as how long it took iOS to get true multitasking) Nique, graphic planet renderes are easy. Heck even I wrote one (don't get me started on atmospheric scattering shaders....) Running a simulation at that scale is much, much harder.

The small city tile wont bother me IF the gameplay is more than zone-and-forget. My hope is that it will be difficult for the city to enter homeostatis and a great deal of time is needed to curate and balance the city, and the other cities in the region also impacting each other in noticeable ways. Without that iterative gameplay style the small city sizes will be very troublesome.

As mentioned, part of the reason SC4 had so much was that it shared the same underlying simulation design as its predecessors going all the way back to the origional. The grid wasn't just a graphical constraint, the simulation used the grid (called cellular automata). Not only was code probably reused, but there was a wealth of playtesting and prototype knowledge. This freed up developers to work on new features because the existing features and tech was well understood. So SimCity 2013 began development and they wanted to have freeform building. No more grid. That means that the grid-based simulation design would no longer be useful so they needed another design. That is 20 years of code, prototyping, and playtesting knowledge out the window. That means everything has to be coded AND tested AND balanced from scratch. From a developer standpoint it really is a reboot. As mentioned it comes down to compromising expectations. I'm going to put the always online drm aside to just look at game play features (I know some are tied to the drm, like modding and disasters) and put forward this thought for pondering: We now know the "cost" of features to make a new grid free SimCity. Is that cost worth it or would you personally rather have more features, but still using grid based building and simulation?

While the city tile is small, I am more concerned on how quickly the simulation can enter homeostatis (where the city can maintain itself) and how stable it is. My biggest complaint about CitiesXL was that a city became self sustaining quickly. It was a zone and forget game and it would take quit a bit of work to destabilize the system. The past sim cities had a similar problem but no where near as bad. Once the city was stable I could put the speed on fast and go to school. When I got home nothing bad happened and I was rich. I'll be fine with the small tile size if the cities require constant care to keep them from falling apart. Otherwise if the game is zone and forget, which seems like to be the majority play style in this thread, we will all get bored quick.

Not really a good chunk of the Simcity fanbase would be very unhappy if they were forced to buy a new PC to play the new game. It just makes it fair to everyone. But this is 2013, everyone has a good computer according to EA's logic, because everyone has a constant Inet connection =D Also its not glasbox thats requires your PC to be good, its the 3D rendering. The simulation goes fluently. And if you zoom out you cant even see the sims so I'm not buying that excuse. Its EA's doing to just simplyfy the game and making it fast paced to get more players atracted to it, period. Incorrect. The developers have said that the limitation is ram. Moving back to sprite graphics would not fix that (see: Dwarf Fortress)

I think the devs have always said they want to have modding eventually (remember it took a few months for the mod tools for sc4 to come out), the issue is how to manage it with the online system.

And where, pray tell me, did I say that? I've read my post again for three times now, and you have me totally lost. I repeat my question: What are you referring to? "Any shift of a game's overall graphics appearance on the axis from "Cartoonish" to "Realistic" (no game will ever be entirely realistic) is based on conceptual and design-related decisions and not on resources. One example I mentioned in that thread is Borderlands 2 - pretty high system requirements, and the graphics have a clear and intentional exaggerated and cartoonish look to them." As far as the scale and gameplay realism issues, for me a game is like reading fiction or watching a movie. Suspend my belief. The sims is a really great example. I know how much it costs to work on a house and buying building materials. The in game prices and wages are hardly reallistic. But it is internally coherent and I can enjoy it. There is some old phrase that I can't remember exactly, that people wont believe true stories because they think it is unrealistic. Simulating real life has that same effect on a game. I've read numerous dev blogs from different designers and it seems to be a common story. They try to be realistic and it ends up making their game either unbelivable, unfun, unplayable, or some combo of the three.

Maybe I'm exhausted, but your first section is making no sense to me at all. I think you might of misinterprited what I said. I did not mean to say that every and all performance issues are because of hardware. I even referenced Homeworld 1. Run it on a GForce 660 and it will still choke with a few thousand polys because of the nature of the engine. Same thing with sc4. Its (probably) a single thread and the draw system just wasn't made to work at that resolution. The point I was making is that the performance of SimCity2013 will hit a bottle neck. According to EAMAXIS it is because anything much larger than 200,000 agents results in too much Ram being used for most people's systems. I am confident they playtested and prototyped like crazy to figure out how many agents and containers should be simulated per sq km, and that limit seems to be about 50,000 agents per sq km. Next question, which no one has the answer to except for EA, is the nature of this specific limit at this moment in time. Is it like the render system of SC4 and HW1 which is that the software is already maxed out? Or does glasbox still have room to scale and consumer hardware is the limit? Now was agent based simulation an appropriate choice for features vs cost? That is an opinion. Raph Koster (theoretical game dev) wrote a bit on this. As a player plays a game they develop a model of how the game works in their head. In a FPS if you wrote an AI with 12 emotional states, but only made a happy, sad, and angry texture that you shoehorn the 12 emotions into, the players internal model of the game is that the AI only has 3 emotions. So don't spend the time coding 12 emotions if the player only "experiences" 3 of them. The opposite is true too. A player can develop a very complex model of how the game works, when really the game just does some statistical handwaving. SC1-4 has had both problems. On one hand they used statistical hand waving to create the illusion of complexity but the actual simulation was simple. But there was also some deeper simulation that they player didn't experience directly or visually, so as far as the player was concered it didn't exist. This time around maxis is going right at the target. The player will see everything that is simulated so the player's psyche has a correct model of how the game works and no time is wasted on features that the player wont experience. I am personally a fan and I feel the cost of city size is worth it. But I also know that is my opinion and the part of games I enjoy most.

Well, stylized art in and of itself doesn't have lower system requirements (Borderland 2 for example), but when using lower requirements a stylized art direction looks better.

As I posted in another thread, realistic graphics don't age well. I compared world of warcraft, a game with highly stylized graphics, to flight simulator 2004 which came out the same year, which was praised for its photorealism. Vanilla WOW. Definitely stating to show its age. Look at the resolution of the rock texture for example. Even today it still looks decent despite the polygon count and texture sizes. Flight simulator 2004. This was herald as "photo realistic." I just flat out disagree with T Wreck's statement that photorealism works better at lower resolution than stylized. There is too much evidence to the contrary.

You need to get this whole 3d monolithic engine idea out of your head. It is just incorrect. I referenced the graphics engine of homeworld to illustrate what a software-bound bottleneck looks like, where no matter what advance hardware you put the game on you just can't get modern polygon counts or textures. Each software has a bottleneck somewhere. For most games the bottleneck is the graphics card. For dwarf fortress the bottle neck is the speed of one core of a cpu. For minecraft the bottle neck is either cpu, ram, or if hosting a server its the network speed. In interviews the developers have said the 3d graphics engine of Sim City is not the bottle neck of the game. They could add even more visuals if they wanted. The bottle neck is the simulation engine -- which they call glassbox -- because it requires a great deal of ram and cpu power to simulate a city at that fidelity. THEY COULD USE TOP DOWN 2D GRAPHICS FOR GLASSBOX AND IT WOULD HAVE THE SAME CPU AND RAM REQUIREMENTS. Also -- minecraft can *generate* a map with a large area, but it does not simulate it all at once. As any serious server operator knows, the game only loads about 100 - 150 chunks around a player at any given moment, which can be adjusted with bukkit. On my poor 1 gig of ram vps I can only handle about 8 spread out players before running out of ram. Bad press and bad word of mouth. What will be your average player's reaction when they play on a region that is a little too large for their system? "Oh this game suzors its buggy poo!" and the proceed to tell everyone they know online. A decade ago with SC4 I would wager that when this happened it happened on forums, and the more thoughtful of us would respond with what you said, that their computer sucks. Now it happens through facebook and skype (of all the things to get popular, skype?!) where there is less of a chance that there is a rational mind to set the younggens strait. According to the steam hardware survey only 36% of steam systems have more than 4 gigs of ram. 53% are at 4, and 9% have 2 or less. 14% are on windows 7 32 bit. Assume that the steam hardware reports are biased to the hard core, then those stats are even worse. I would love it if we were in the 64 bit era. Stupid hardware companies holding us back. The fact that 10% of the market is being excluded is actually surprising to me. According to past interviews they were talking about maxing out the active agent count at 200,000. A lot more than what PA is going for. I wouldn't sneeze at the ai for these either as at least some of these agents will be using fairly robust path finding when simulating cars. Assuming a 1:1 relationship between agent count and ram (which we both know is false, but its the best approximation I got) then if it takes 4 gigs of ram to simulate a 2x2km area, then it will take 16 gigs of ram to simulate 4x4 (double the side of a square you quadrupedal the area). Do some people have that? Sure. Is it worthwhile to spend developer hours on it? Probably not.

Logical fallacy. Just because A is true means B is true, does not mean if B is true A is true. Yes boarderlands 2 has steep requirements for its real cartoony look. Does not mean cartoony looks require advanced requirements. I think what UC and I are saying is that it is easier to make stylized graphics look good on weaker specs, while going for photorealism is much more difficult and it doesn't age as well. The two examples I use for this are World of Warcraft and the non organic CG from Babylon 5. It has only been in the past couple of years that they have really started showing their age graphically. Much of this is attributed to the art direction of bold colors and using simple shapes to their advantages. Granted Wows graphics have been given a few facelifts of the years, but if you look at the low requirements there is a lot of bang for buck. Compare that to what was herald as a king of photorealism from the same year Wow was released, Flight simulator 2004 Or Doom3 (check out the resolution of the subway car textures and the floor. Eiack!)