#Using frames whole potential
1 messages · Page 1 of 1 (latest)
a ton as in
Idrk just many
make the parts disappear after a while
why do you need that many instances?
Terrain gen
optimize it to create as litle parts as possible, and then pool them
I still need to create it
But that's not the point
.
How would I constantly create them without dropping fps
the point of pooling is that you don't need to constantly create parts
But the original time
U gonna create them
To fill loaded terrain
you just don't load terrain until the cache is full (load screen), you can limit the cache to 100 parts per frame or so which shouldn't cause any big fps drops
So you're saying just fixed limit per frame
Is that rlyyy best method tho
depends on your terrain 
^
you can also adaptively yield
voxel terrain + greedy meshing is usually relatively low partcount-wise
LODs
you can also send the terrdata in a buffer and have the clients handle the parts
this reduces server strain, though you would need some extra checks
I want it on server
Just
Stop going to diff thing's
I wanna see if there's better way then fixed amount then task.wait
why do you care about FPS drops on the server 
there's nothing you can do other than optimizing your code using some of the methods we've listed.
there are a few more advanced methods as well, but not really necessary unless you plan on having high resolution/large scale
Cuz physics and shit also die
I wish limiting it to 5ms per frame worked
Aaaa
it does.
how many parts are you spawning in the span of 1 second
Constantly for first 5ms
Of each frame
Then stop
And wait untill next
that doesn't answer my question
Luau runner
I mean it runs for 5ms
Then stops
So 5/16ms
Rest is empty
Turns out I may or may not have forgotten to anchor
oh I completely misread your script lol
this will still lag though, you'll need to set the cframe to math.huge so it's not rendered
Ye ofc for pool
I was just trying for terrain gen
Without pool as pure test
Of instance new amount without death of server
no, I mean it's going to lag because you have 60000 parts on the screen at the same time
it's not necessarily because of instance creation
Even on server?
Like
I spawn stuff visible to everyone
Ik replication might lag it
For pooling it's ofc gonna be way farther away
But for visual parts
mostly this, instance creation and parenting
Alright but one more question
Cuz without anchored
It was doing 600 parts print then crashing basically
Now I get 300 without crash cuz anchored
How did it manage to do more
Also did amount of physics part kill it, or bc it was all in single place making collisions nightmare