Testing server performance

[rafaeru]

Hello, im working on my ot from longer time and i have certain question. Namely do you know any ways to check which lua scripts are the least efficient ?

For expample, from experience I noticed that sending a lot of magic effects / animated texts for all online players causes felt lags.
What are your ways to find out what is causing lags?

I know i can turn all scripts off and turn on them one by one but problem is in simulating e.g. 200 players on which this script will affect.
I would like to rewrite the least efficient scripts into source to achieve the best server performance.
What do you think about this?

 

[Evil Puncker]

I would like to rewrite the least efficient scripts into source to achieve the best server performance.

that would not help at all in most cases..., and yes, scripts that deal with iterating with all players and timers will mostly be the ones that will cause "lag", and btw you gave us no info at all, like server version (not tibia version) and the hardware (and OS) that it will be hosted


and here are some useful links:

[TFS 1.2+] Performance tuning

Introduction. A few questions asked on GitHub recently have convinced me that it might be necessary to write a tutorial about some quirks and new performance characteristics of the upcoming 1.2 release of TFS. I won't go very deep into the internals as I know that most people reading this...

otland.net

Cpu High %

hello everyone, i have been watching my otx recently and strangely it is consuming the cpu way too much, with 246 players online it consumed 90% of the cpu media, so i removed all the bodies that were not disappearing, i reduced the time they stay in the chao, I expected to decrease consumption...

otland.net

C++ - [1.3 ]this attack speed formula can cause bad performace?

How i talked, this new formula of attack speed can cause bad performance (if for example have a lot of players online )? When player have more fist skill, he hit more fast, so i changed it uint32_t getAttackSpeed() const { return vocation->getAttackSpeed(); } to it...

otland.net

and also an optimized tfs in case you don't wanna to worry:

SaiyansKing/optimized_forgottenserver

Contribute to SaiyansKing/optimized_forgottenserver development by creating an account on GitHub.

github.com

 

[rafaeru]

My question is totally theoretical how you deal with it because I pick out something for myself.

 

[Night Wolf]

you can benchmark them with os.clock..

e.g:


local benchmark =os.clock()

-- your script here

print(os.clock() - benchmark)

In your console it will output the time that the code took to run. Higher values mean the script is too demanding.

You can also make this more generalized in the cpp to whenever you're calling a script in Lua, you benchmark it. Storing benchmark results, taking the mean and outputing later the event called and it's mean time to run may be a good idea. Just remember that sometimes an event is called but returned before running entirely, so the meantime wouldn't be a real value.

 

[josephross]

you can benchmark them with os.clock..

e.g:

local benchmark =os.clock()

-- your script here

print(os.clock() - benchmark)

In your console it will output the time that the code took to run. Higher values mean the script is too demanding.

You can also make this more generalized in the cpp to whenever you're calling a script in Lua, you benchmark it. Storing benchmark results, taking the mean and outputing later the event called and it's mean time to run may be a good idea. Just remember that sometimes an event is called but returned before running entirely, so the meantime wouldn't be a real value.

Thanks, I give it a go to figure it out.