4 ms·
We lack the internal knowledge to know whether an offline mode would have been a good experience for most customers. I have no idea whether Maxis+EA tested this
by spartango 14y ago
We lack the internal knowledge to know whether an offline mode would have been a good experience for most customers. I have no idea whether Maxis+EA tested this, and how well it worked.
That said, you can imagine a situation where doing these computations locally did compromise the experience for a certain spec of machines. Maxis has already explained that city size limits were imposed because the simulation didn't run sufficiently well on the low-mid spec machines above the allowed size.
As the owner of a higher spec machine, I find it somewhat silly that this couldn't just be a preference item (as AA and resolution are), but I can understand the decision to cut down the computational demands to keep gameplay smooth.
Games that have launched with framerate issues recently have met a huge amount of flak--see what's going on in the Planetside 2 community. You can argue that what's happening to PS2 is far worse: they released a poorly optimized game, and those issues prevented people from playing for months.
- MichaelGG 14y agoYou're saying that the client PCs are so weak, yet EA can offload it into the cloud for cheap? How exactly does that work? My 3-year-old ultralight laptop has as much CPU as an EC2 M1 large. I find it hard to believe the game is so complicated a laptop will struggle, but EA can efficiently load it onto servers? That sounds pretty hard to believe. Even if the target player is using a crappy Netbook, you're talking about a dedicated compute server for every 4-20 players. That's insane.
- ariwilson 14y agoSimCity system requirements: Intel players will need, minimally, a 2.0 GHz Core 2 Duo, while our AMD players will need at least a an Athlon 64 X2 Dual-Core 4000+. Those sound pretty capable of a significant amount of simulation themselves...
- johnward 14y agoYep they ran Simcity 4 fine.
- spartango 14y agoThis is an over-simplification of the performance demands of the game. Keep in mind that there are many levels of computation involved in the simulation, from rendering loads to inter-agent complexity, to city management (i.e. taxes). It all adds up, and the graphics needs alone are the limiter in most games. But beyond the "feather that broke the camel's back" rationale for offloading compute, keep in mind that there are issues around shared data that would vastly improve performance of region-level computations in the cloud. If the region state is maintained in the cloud, any computations on that state will be far more quick when done in a centralized way. Further, there are fewer challenges around handling client desynchronization, which is a classic problem in large scale strategy games. Reconciling one or more desync'd client's state can get very difficult. Under a centralized system, the master copy reigns. Obviously there are downsides, but it's not like it's a crazy idea to offload some of the compute to the cloud.
- MichaelGG 14y agoReally? You're suggesting they use a software renderer instead of the GPU? Even if 70% of the CPU was wasted on a modern machine, that's like a quarter core per user that they'd need to rent from Amazon. I don't think so. As far as the multiplayer, I don't think anyone has a problem needing servers for multiplayer. All I'm saying is that the suggestion that computations need to be done online to save local CPU, even on a singleplayer game, is obviously ridiculous.
- jiggy2011 14y agoI doubt they are using EC2, probably buying servers in bulk from IBM or whoever. When you centralise you also have other advantages, such as being able to cache the results of previous computations. So if you can structure the computations in such a way as to be fully or partially reusable you can memoize and scale much better.
- dagw 14y agoIf you google for people attempting to reverse engineer the protocol you'll see that they've found that EA is indeed using EC2. It is however unclear if this is a permanent state of affairs or if they're simply using EC2 in an attempt to handle the initial burst of traffic