CMO RAM usage

[Nikel]

In a recent long session testing Transient's last scenario, Operation Ardent Shield 2030, CMO used up to 9 GB of RAM in the end.

Did not save a screenshot of the task manager.

Just wondering if this is normal, because before the 64-bits update, CMO could not use so much RAM and so the scenario would crash or hang in the testing? What seems strange considering the AU count is less than 2000.

[thewood1]

"Just wondering if this is normal, because before the 64-bits update, CMO could not use so much RAM and so the scenario would crash or hang in the testing? What seems strange considering the AU count is less than 2000."

Thats not true. CMNAO and early CMO (32-bit) had RAM issues all over the place in large scenarios. There were threads all around that issue back then. They went away once CMO went to 64-bit.

Not saying you aren't seeing it, but let's not change history.

[Nikel]

I am not changing anything, just asking, 9 GB is normal in a long session of gaming CMO?

I do not think this scenario in particular is so large, but could be wrong.

[thewood1]

Just ran The same scenario. Steam version at double flame and a number of player missions.

Before CMO is started its at 5.2GB:

Screenshot 2024-11-28 164028.jpg (244.02 KiB) Viewed 1075 times

On CMO start its 6.1 GB:

Screenshot 2024-11-28 164132.jpg (242.44 KiB) Viewed 1075 times

On scenario start its 6.8GB

Screenshot 2024-11-28 164254.jpg (253.98 KiB) Viewed 1075 times

At one game day in its at 8.3 GB:

One day to go.jpg (243.05 KiB) Viewed 1075 times

The max I saw was 9.5 GB by the end of the game. The math it used 4.3 GB at max usage. That's about what I get in my benchmark scenario at 1800 units and a ton od AA missiles flying around. Not sure how you are getting 9 GB.

[Nikel]

Well, it took me hours of real time to play it till the end.

Saved the game several times, but it was a continuous session. And the game behaved perfectly.

Yes, I should have taken a screenshot.

[Nikel]

Here is a screenshot with TBLackey's Operation Sallyport 2029.

Memory leak.png (12.8 KiB) Viewed 751 times

[thewood1]

How about posting a save so no one has to track it down.

[Nikel]

Saves of this scenario I have many, here is the last.

Operation Sallyport 2029 12n.zip

(2.2 MiB) Downloaded 5 times

[TBLackey]

Noticed the issue too when you PM'd me that save file. Had to wait a few minutes until my machine calmed down, before I could take a look!

idk what the problem is, but in the specific instance of your save file, it is noteworthy.

[Nikel]

Indeed it takes some time to load it, with a pair of "not responding" moments.

Both yours and Transient's scenarios require a lot of hours of gameplay, there is accumulation of RAM usage and so you reach the 9 GB, perhaps are all the calculations that are stored or a memory leak?

I only use borders and coastline layer, so cannot be that CMO is loading maps when playing.

[thewood1]

I would hesitate to just dump Operation Ardent Shield 2030 into this. My PC had no issues running that scenario to the end. RAM usage was right around what I would have expected.

The difference could be system-specific or incident specific around air battles my play through avoided. Having that save of the other scenario is key and should make it easier to reproduce.

[blu3s]

We have fixed some issues regarding memory allocation and improve our GC.

The changes are on build 1502+, so please give it a try once a new beta is released (or the steam branch updated with a version 1502+)

Thanks

[Nikel]

Thanks blue3s

GC, is this?

https://en.wikipedia.org/wiki/Garbage_c ... r_science)

[blu3s]

Correct