Is it a good move to disable windows' file-caching capabilities (via I/O completion ports) for resource loading in a game environment? Most games will do their own resource management and caching anyways, so the OS will probably cache it a second time. Of course, you don't get the caching across the process' lifetime, but that's usually not a big downside. On the other hand, it probably makes file loading times more predictable - so you can actually give accurate load-time predictions.
If your goal is loading-time prediction, I would say leave the OS file-caching in place. It's probably general, but well-written, so I would say that you shouldn't disable it unless:
Otherwise, it's probably not worth the work; and in the worst case (average case?), you may even decrease efficiency. |
|||
|
|
|
I'd say no. Windows will be using spare memory for this caching - it's basically free. Unless you've got a fancy system to somehow expand your process memory to cache more data when Windows is under low load, and then contract it again when Windows decides it wants more memory, just let the computer do what it's good at it. I don't think increasing load time in order to give better load-time predictions is a good tradeoff :) |
|||
|
|
