Every day i work on making my game much more optimized by bringing down part counts and making stuff look better, but as stuff looks better and as i go on im worried about my part count. Here is my part count.
Is this too much?
5 Likes
Personally, I think that’s fine. Keep working on bringing it down but don’t worry too much about that. I think at that amount (as long as you don’t have a hecka lag), you should be good to go. If the part count keeps increasing to 15-20k+, it could cause a problem for mobile users and potentially PC users, varying between devices and how well their PC performs. A suggestion would be to add a loading screen. It won’t help take down the part count, but it’ll give the server time to load all the parts.
tl;dr - sounds good, keep going. loading screen is useful for loading parts.
This is just my personal opinion, I might be wrong but this is what I’d consider if I were in your position. 
7 Likes
13,000 parts is a lot, but if you space them around the map in a reasonable manner (especially if it’s a larger map) then you should be fine for the most part. And by spacing out parts, I mean making sure no cluster of, for example, 10,000 parts can be seen in one view of the camera.
The part that worries me are the Unions, as you have nearly 3,000 of them. It really depends on where clusters of parts are on your map and how big your map is, as well as how script-heavy your game is.
If your place is just a showcase with very minimal scripts (again keeping in mind the points above) then you should be fine, but if your game is very script-heavy then you may start to see performance issues.
4 Likes
I have a loading screen already
1 Like
My game is not a showcase more of a roleplay game of some sorts. If you want to check out my game then ill link it. Im going to link the test game where im testing my new update that is where the part count you see in the post is.
https://www.roblox.com/games/821905618/Meeskops-Neighborhood-V5-1
Even better. Does it actually wait for the game to load or does it just tween? (shuddering at the thought)
For some references about game loading:
https://developer.roblox.com/en-us/api-reference/function/DataModel/IsLoaded
https://developer.roblox.com/en-us/api-reference/event/DataModel/Loaded
1 Like
Well it waits a good 10 seconds its more of a intro then a loading screen. If it waited till the game was loaded then xbox would have issues with loading so long as i have a day 1 edition xbox and it takes a while to load.
Just to clear the air of any misdirection, @Polyheximal and @JukeRblx, you (and many other people) can’t accurately discern what is or isn’t a lot of primitives without seeing how they’re used because they aren’t entirely reflective of performance. Instead of guessing, ask to see the game, or at least the memory use.
Which is what I did:
682mbs is high but it isn’t terrible. That can still run without a problem on all but the very lowest end devices (like the iPhone 5, etc)
For reference, this is the memory use of Power Simulator:
Less memory use and this has 45 thousand primitives (parts and meshparts).
How did I manage that? By using instancing to my advantage with all of my meshes and parts. The largest discrepancy between OP’s place and Power Sim is that GraphicsTerrain, TerrainVoxels, and TerrainPhysics add up to around 200mb extra memory use, but that could be contributed to the fact that the OP’s place is just generally massive.
6 Likes
Thanks for clarification. Assuming you’re comparing the “PlaceMemory”, there is a ~44 difference between Power Simulator and the OP’s game and therefore, what would be an ideal memory usage?
2 Likes
Well ideally, as low as you can, but eventually you’ll be investing more time in such minor optimizations that it’ll waste time that should be invested into game development.
That being said, I know that as of 2018, 600-700mb is around the higher side of memory usage, but that number threshold gets higher as Roblox becomes more optimized (which if I remember correctly, a few larger game optimization updates are coming down the pipeline?). So again, it’s complicated and very game-specific.
When we QA tested event games, we did so on a full server of testers while staff played on the lowest end devices (RIP fried iPhone 5) because that’s more reflective of a typical game experience (since player characters add to memory use).
4 Likes
While you’re generally correct, this rule has a lot of caveats that just aren’t reflective of a usual game, and I’d advise to change your wording to reflect what I said above instead of adding to the confusion (since there are a whole lot of sweeping general statements and misinformation that aren’t realistic or applicable).
3 Likes
13,000 is quite a lot. Could I have an idea on what sort of game you’re doing. Because if it’s a simulator that’ll require a lot of scripts than that number could be reduced.
Ill send you the test game where that part count is… no its not a simulator
https://www.roblox.com/games/821905618/Meeskops-Neighborhood-V5-1
Read the entire thread before responding. Not only did I specify above why 13k parts isn’t a lot, but also that it as a metric isn’t useful. The OP also supplied a link above as well, which you would have seen if you didn’t just read the first post and reply right away.
3 Likes
Personally, I think this isn’t that many. If you are worried about it you can always put a high-part warning in the description, and out a (READ FIRST) in the title.
1 Like
Nah, 13,000 points isn’t that much, my game which is rather simple has about 40k parts, and 50k instances.
1 Like