[Observation v3.9] Memory Usage in v3.9

[knightly]

Hi James,

You are really a diligent developer -- can fix so many bugs in a short time! Many thanks for your hard work!

I have not truly test-played v3.9 yet, but I have done a memory survey as I did in the past, using Wipi's save game. Wipi's save game is well-developed with all types of goods, so it is an ideal candidate for testing how much extra memory is needed with your new implementation. As usual the game starts in Dec06 :

ST STD v102 :  (copied from my previous report)
    Dec06    84MB
    Dec07    93MB

ST EXP :  (new data)

v3.8
    Dec06    102MB
    Dec07    105MB

v3.9
    Dec06    126MB
    Dec07    129MB

Hope the data above are useful to you.

BTW, is v3.9 currently using smart pointers as you described in the other thread?


Edit : Figures updated.

[jamespetts]

Ahh, thank you very much for your report - it is always useful to know how these things work out :-) I decided against using smart pointers in the end, as I realised that I did not need them, and not having them improves performance. How does performance in 3.9 appear to compare to that in 3.8?

[knightly]

You are most welcome, James.

As for game performance, I really need to seriously test-play v3.9 first before giving any comments.

So, you didn't use smart pointers. Can you push your recent changes to Github so that I can take a look? Many thanks in advance!

[jamespetts]

Already done!

[sojo]

Sorry for my question here, but I think it's ok.

How many memory needs Simutrans normal?
And how big is the difference between ST normal and ST experimental by CPU-time?

[jamespetts]

I don't really have a definitive answer yet, as I haven't had enough test responses. All that I can say is that it works fine on my system (Pentium 4 3.0Ghz, 2.0Gb of RAM, Windows XP, 32-bit), albeit it's a little slower sometimes than Simutrans-Standard. By all means, try it out, and post your system specs, and tell me whether your performance is acceptable :-)