Map memory consumption, which rate can be considered as normal?

[Avahor]

Hi guys:
testing my current project I got some considerable differences on memory consumption, basically on a rpg that spawns creeps on single areas if a hero walks nearby and after some minutes if the area keeps inactive (non hero near) the game slays the spawned creeps and reset the area. the rate of unit creation is 700 - 800 per game.
leak checks apart, I got variations between 1100 mb of ram consumption for a 15 hours continues of game, and 900 mb for a 58 mins of game.
Could be these rates considered as normal. Do you have checks of your own maps memory consumption?.

 

[Zwiebelchen]

Sounds normal to me.
The real question is: was there a significant amount of lag after the 15 hour session and how long did it take the game to leave the score-screen?

 

[Avahor]

Sounds normal to me.
The real question is: was there a significant amount of lag after the 15 hour session and how long did it take the game to leave the score-screen?

yes, that's a check point too. I've checked time and it doesn't take more than 2-3 seconds to leave. In fact there are some leaks but they seem to be minimized.

It would be great if we can manage a standar meassure about most common hardware that actual players use. Maybe these 2-3 seconds for me could variate for another player. I mean things like the difference between European player's hardware and Asian or American player hardware in terms of RAM rate consumption (which could be hard to define ). I know that some chinesse players choose play warcraft cuz its low requirements.

 

[Zwiebelchen]

yes, that's a check point too. I've checked time and it doesn't take more than 2-3 seconds to leave. In fact there are some leaks but they seem to be minimized.

It would be great if we can manage a standar meassure about most common hardware that actual players use. Maybe these 2-3 seconds for me could variate for another player. I mean things like the difference between European player's hardware and Asian or American player hardware in terms of RAM rate consumption (which could be hard to define ). I know that some chinesse players choose play warcraft cuz its low requirements.

If it took only 2-3 seconds, this is a strong indicator that you don't leak a lot in your map. Heavily leaking maps will often take several minutes to close, even on modern machines.

If you are still unsure about potential leaks, you could always just drop a handle counter in your map.

This basicly tracks the number of allocated handles (objects) in your map. If there is a constant rise of handles in your map (without having more assets on the map at the same time), this is a sure sign that something is leaking.

This does not catch reference leaks, only handle leaks.

Also, don't take the displayed value too literally; counting the actual number of handles is pretty much impossible. This is just an estimation and a certain fluctuation is normal even for non-leaking maps. It's best to monitor this over a longer time and see if there is a certain trend to be seen here (for example, the number suddenly rising unusually fast).

 

[Avahor]

thanks mate, I've used handle counter and made these test already. I got some doubts about running memory consumption but i cleared them by ur references.