Benchmarking : register_think, register_touch VS register_forward(FM_Think/FM_Touch)
Dears Friends.
I have a question about run time efficiency with engine methods and fakemeta methods. With engine, it's possible to register an entity think or touch by inputing the ClassName as a filter, and trig easily a delegate function. Ok. With Fakemeta, registering purpose that we must catch ourself the class name by doing theses features : => Allocating predefined string variable, dedicated to read the class name with pev function. => When it's done, we must compare collected value with classname we want to compare, with equali... => ... and calling function used when the name is matched. I'm afraid that adding a lot of classname comparaison will take more cpu time than the engine methods. Isn't it ? I think this is a good investigation no ? What do you think about this ? Yes I LOVE FAKEMETA and i want no longer use engine methods, but for theses cases, engine is always more efficiency ? "Voily voilou" ;o) |
Re: Benchmarking : register_think, register_touch VS register_forward(FM_Think/FM_Tou
Actually,the only difference is that engine compares the names itself,with fakemeta you can choose what to do,filter by entity id,etc.And remember to use statics to help performance a bit.
|
Re: Benchmarking : register_think, register_touch VS register_forward(FM_Think/FM_Tou
I would reckon that if you look at the engine module's source code it's doing roughly the same thing.
|
| All times are GMT -4. The time now is 16:07. |
Powered by vBulletin®
Copyright ©2000 - 2024, vBulletin Solutions, Inc.