#Late Game Performance

1 messages · Page 1 of 1 (latest)

rancid rapids
#

My Factory is now over 1m Buildings. I have 100% Research done and now just spamming milestone factories to increase my Operator Score.

On this run I have about 275 hrs of game play on it and an Operator Score of 320 or so.

Context. I have done this exact situation on every level besides Hex. Still working on that.

It was at about 650k-700k building that the input lag was getting to be frustrating. I would click to drag a space belt and I would not get the first space I selected. I had to decrease my clicking speed and scrolling speed to allow for selections to be made.

Will this always be an issue? Or will the amount of building limit truly reach an unlimited amount?

My goal: Attempt to use every color or asteroid in some fashion to make a true infinite amount of shapes. But without hurting my eyes to do so.

turbid nest
#

Because of the performance
I don't expand the factory in the end game.
Instead I trying to create a TMAM, which is more challenging and fun than copy and paste the existing factory.

iron fractal
#

Unfortunate reality is there will always be a performance cap eventually but changes are in the works to push that cap much further down the road so you have many more buildings before then

deft herald
#

The devs recently posted that the real limit is 500k buildings before performace drops. This is known, and they are working on it.

south sable
#

See #✨updates

rancid rapids
cloud nova
deft herald
#

Unity calls specific functions in a particular order every single frame. A lot of the UI is scheduled for polling and evaluation at a particular phase of evaluating every single frame.

#

And a lot of this is baked into the Unity engine. I can’t tell Unity to just skip the “LateUpdate” phase for this frame, and go straight to the decommissioning phase. It doesn’t work like that.

cloud nova
#

The correct and hard path is to run an independant factory simulation and use update only to copy underlying factory progress into visuals

#

This provides much higher performance for stuff offscreen (running without reference to Unity), enables multiplayer (easier to make raw simulation lockstep) and lets you update or save the factory without tying up the UI

deft herald
#

otherwise other functions (especially those related to rendering and music) would be reading the important variables at inconsistent times.

#

it would be crazy to just have the simulation running in parallel, and just let everything else just query the sim whenever it feels like it.

cloud nova
#

Once you have the previous sim, you kick off a background task which calculates the next, next state.

#

In a later update, if the background calc is done and it's time you copy next next into next. You then move next into previous.

#

If you want to save, you just start saving the next snapshot in the background. No need to lock the UI, since the next snapshot is not going to change.

cloud nova
cloud nova
cloud nova
#

You don't want your tested game to break when some gamer with a 175 Hz screen plays

deft herald
#

you can tell unity to cap the frame rate

#

and this is likely done in the FixedUpdate loop

#

i don’t understand what you are getting at, because you are proposing a potential major architectural overhaul, just so the UI doesn’t lag when the sim lags.

cloud nova
#

Yeah, it is something you want to do early in design.

deft herald
#

if you are usuing unity, it is vastly simpler to work within its limitations than to try to circumvent the core features of the engine that you are paying many thousands of dollars to use

pure quiver
#

Tobs recently mentioned they have a working fix that gives something like 45 FPS with a million or two buildings. No test build for players yet though.