View Issue Details
| ID | Project | Category | Date Submitted | Last Update | |
|---|---|---|---|---|---|
| 0017883 | AI War 1 / Classic | Suggestion - Game Mechanics | Nov 28, 2015 3:18 pm | Nov 28, 2015 3:21 pm | |
| Reporter | Kalzarius | Assigned To | |||
| Status | new | Resolution | open | ||
| Product Version | 8.023 | ||||
| Summary | 0017883: Statistical Simulation for Massive Fleets | ||||
| Description | I don't know how the game currently does it, but if you have--for example--16 homeworlds, you can have absolutely massive battles of 10,000+ human ships versus 100,000 AI ships. Unfortunately, this battle goes at about 0.25 frames per second, so it's fairly unbearable to watch. No matter which performance profile you use, this makes engineers just sit there, not automatically boosting factories and the like, and obviously the game takes an excessively long time for even a ten minute period. This is clearly a simulation issue, and I can only assume each unit is being individually (or in very small groups) simulated. I think, and especially for units that are on other planets, that you could easily get by utilizing a statistical model wherein 10 or 100 or 1000 units are simulated as 1. This granularity could be automatically activated when the realtime speed drops below 80%. In the worst case scenario, all units of a single ship type and mark could be simulated as 1-10 units. This would obviously change how the game plays to some degree, but I think as an option it would be a fair tradeoff for performance speed. Especially on planets that you are not viewing, you aren't going to know what ships got hit by what ships, so does it matter if each ship or each other ship got simulated to the fullest extent? Hopefully, by reducing the simulation load, support ships can function properly. I don't know how other people's machines run the attached saved game, but given the hundreds of thousands of AI ships, my 5-year-old rig slows to a crawl. | ||||
| Tags | No tags attached. | ||||
| Internal Weight | |||||
