Should Klei (eventually) optimize dst performance?

[Teaose]

Late game dst servers tend to face lag issues, part of this is due to accumulation of things in the world, part of this is the amount of things the server needs to handle in the late game.

This causes poor performance for dst which came out over a decade ago (WOW!) and runs on a single cpu core.

Do klei forum peeps think klei should one day push an update that focuses on performance and optimization? Or would that be a waste of development resources that could be spent expanding the game?

 

[sarreos]

I think it is much needed, although probably hard to do without massive code refactoring. Maybe after the current arc and the rest of the skill trees are finished.

Edited July 8 by sarreos

[Mike23Ua]

Not sure how relevant it is but I play on Xbox Series S and my friends (& non friends) who try to join my server do not seem to be able to actually get in to my game.

I am assuming it is because I built all of THIS directly at the Florid Postern so it all needs to try to load as soon as they attempt to Join.

 

Can anyone confirm if their Cawnival also prevents players from being able to join?

[Teaose]

4 hours ago, sarreos said:

I think it is much needed, although probably hard to do without massive code refactoring. Maybe after the current arc and the rest of the skill trees are finished.

It could also be done slowly in parts instead of all at once. The updates hat made some items auto stack comes to mind

[Captain_Rage]

They already do.

[_zwb]

would be great if Don't Starve uses all my CPU cores, but that is just as likely as Hamlet getting a new update

[NSAiswatchingus]

There's a child somewhere playing DST on the Switch that has spent months gittin gud, researching the wikis, studying YouTube guides, and finally got to a place that they can start making a megabase. Only to find out, well into their project, that their device can't handle it nor any kind of late game world smoothly. Heck, PS4 too. 

It feels odd for a game to release on platforms that can't really handle its own endgame without severe drops in quality. I'd say optimization is a must, but at this point in DST's life cycle I don't think we'll see any MAJOR improvements.

[Captain_Rage]

3 hours ago, NSAiswatchingus said:

There's a child somewhere playing DST on the Switch that has spent months gittin gud, researching the wikis, studying YouTube guides, and finally got to a place that they can start making a megabase. Only to find out, well into their project, that their device can't handle it nor any kind of late game world smoothly. Heck, PS4 too. 

It feels odd for a game to release on platforms that can't really handle its own endgame without severe drops in quality. I'd say optimization is a must, but at this point in DST's life cycle I don't think we'll see any MAJOR improvements.

It's like handing out food and supplies for a lifetime when you know that the life expectancy of the audience is about a week on average.

[Walrusst]

Webber's would be probably the best time to improvement to reduce AI costs/cull unnecessary behavior checks on a global level since a major cause of lag spikes that I understand is the AI checking for floor items/structures to be interested in.

Most of kleis public servers that I used to visit just die from someone walking in a swamp and now nobodies pathfinding works and the games running at 3 frames per second.
When you see such a high/low mix on what actually impacts performance losses, its difficult to identify problems elsewhere since if there is an erratic background event that causes intermittent and massive lag spikes it makes all other performance testing extremely unreliable. Its a case where its difficult to talk about other issues because one of them stands out so badly and irregularly.

I know there are mods that actually kill a lot of adversarial npc behavior and get the world to run semi-smoothly for thousands of days, but obviously the living world and complex behaviors are what make don't starve don't starve, but there are tools available to an optimization engineer that are more sophisticated then turning things off, it just would take resources that I hope klei secured access to in their preparation for mobile DST.

Edited Thursday at 07:59 PM by Walrusst

[Frosty_Mentos]

Things like spider AI does mostly affect performance or any big swarms of creatures. For some reason spiders remain biggest issue.

Overall there was a huge performance upgrade when they've changed how lighting works, ultimately upgrading their engine. It was years ago, but playing in a more entity filled worlds they've made it better.

There's still improvement to be done, like with spiders seeking food type of deal, that I've heard.

[Juanasdf]

I don't expect a performance focused QOL update until at least all of the skilltrees (and the last rift content update) are released. And even then, I wouldn't even expect it to be a priority, more like a small possibility.

Performance improvements have come in the past as small additions, those can and will definitely still appear from time to time.

I do expect webber's skill tree to come with a few tweaks to spiders' AI, but nothing major. I think they'll try to give him perks to make stronger spiders instead of recruiting dozens as a bandaid fix.

[RubLog6]

7 hours ago, NSAiswatchingus said:

There's a child somewhere playing DST on the Switch that has spent months gittin gud, researching the wikis, studying YouTube guides, and finally got to a place that they can start making a megabase. Only to find out, well into their project, that their device can't handle it nor any kind of late game world smoothly. Heck, PS4 too. 

It feels odd for a game to release on platforms that can't really handle its own endgame without severe drops in quality. I'd say optimization is a must, but at this point in DST's life cycle I don't think we'll see any MAJOR improvements.

My heart breaks a little inside when i get lag spike or go yellow in my own dang solo server

[Teaose]

Even natural roads are poorly optimized and cause strain on a server. 

So I'd assume there are more legacy code out there that could be improved performance wise.

Heck, some time after moon quay was added, they fixed the pirate monkey AI code to not do redundant checks. Previously, it was a copy paste of splemonkey code, despite them not being able to throw poop etc.

[Draggofroot]

On 7/8/2026 at 1:28 AM, Teaose said:

Do klei forum peeps think klei should one day push an update that focuses on performance and optimization? Or would that be a waste of development resources that could be spent expanding the game?

 

whats the point of a luscious, plentiful game if the game cannot be played 🤔

[Walrusst]

20 hours ago, Frosty_Mentos said:

Things like spider AI does mostly affect performance or any big swarms of creatures. For some reason spiders remain biggest issue.

Overall there was a huge performance upgrade when they've changed how lighting works, ultimately upgrading their engine. It was years ago, but playing in a more entity filled worlds they've made it better.

There's still improvement to be done, like with spiders seeking food type of deal, that I've heard.

Spiders AI cost does seem to be proportional to world entities/things left on the floor.

This isn't as bad a problem say, with wurts merms.
There are probably a few things you could do to cull their search behavior checks, say, if they limited the search to certain behavioral states, say, sometimes when they are idle and stopping to stand still they look around and do their checks instead of it being passive and constant, or chasing a player. Plus, spiders looking around when unalerted would look cute.
I'd need to look at the code to see if they were doing local area based culling, but some sort of masking technique to cull checks to a small local area is a technique I've seen other games do. (IE: Iterating on the whole items array takes time, if you had a small memory sacrifice for world chunk based buffers for whats locally available in your chunk, and nodes to check the neighboring ones, you could check 5% of the world rather then the whole world.)

Its possible they are using all these methods and still getting spider behavior issues, in which its my bad, but I wanted to give examples of things that I've seen be used as measures  elsewhere, I'm not trying to do their job here, just saying what I remember seeing when I was a rabid game modder years back and I did go around reading random games code.

[-Variant]

22 hours ago, Frosty_Mentos said:

There's still improvement to be done, like with spiders seeking food type of deal, that I've heard.

There are so many people still talking about Spiders as if they're this horrifying lag monster.
I don't think I've ever really experienced spider induced lag before, but it should be noted that they're pretty optimized.
I'm sure there's some improvements to be made somewhere, but it seems like it'd be a broader thing to tackle.

[Swiyss]

30 minutes ago, -Variant said:

There are so many people still talking about Spiders as if they're this horrifying lag monster.
I don't think I've ever really experienced spider induced lag before, but it should be noted that they're pretty optimized.
I'm sure there's some improvements to be made somewhere, but it seems like it'd be a broader thing to tackle.

I've experienced the spider lag recently, and it just made the game completely unplayable and it's not an understatement. Took 4 to 5 seconds for me to move an inch, and inputs already given would continuously go until the lag started to settle. It was on a dedicated server on one of the best computers money can buy today and still all it took was 40 spiders to not allow me to play the game. 

However, I cannot seem to get the same exact issue when playing alone. Spawning 200 spiders still makes the game playable alone, it's only when there is a dedicated server that for some reason the game doesn't work anymore.

So yes, we're talking about 2 different issues in this thread, the spider dedicated server lag and the device limitations of not correctly displaying a large base due to hardware problems on switch, android, playstation and xbox. I haven't experienced the latter.

[Walrusst]

I kinda expected there to be something done there, hence why I softened my statements a fair bit by including the "I hope there is something fighting this."

Since there are some servers I've been on where walking by swamp spiders nests have reduced ticks on the server to 3 per second or something absurd, others where its never an issue. I used to get kicked just trying to play webber if I don't immediately grab a razor and show I'm the extermination team.

I suspect there is a complexity where there is a very seldom reached branch in spider behavior where if players are sloppy in a certain way, it can cause extraordinary levels of problems, but the majority case is decently optimized because otherwise we couldn't play. I mean I've seen players who've complained that leaving stumps in the game has lead to lag eventually in long living worlds, so I'm fully willing to believe that there are just a lot of edge cases where player lazyness and unintended behavior can lead to problems. (Or say, a spider war goes on for several years against the one setpiece with 80 tentacles in it, enough glands and silk stack up that what you would normally call a "Player lazyness" extreme case activates anyhow just because the npc on npc combat. After all, I've seen path of exile run into weird cases where they complained of the majority of a regional cluster tick time getting used by single necromancers back during the weird fungal tower defense league.)

It also could just be, you know, people who try to play pikmen in a public server that hasn't been restarted in too long tend to just be a big issue as well. Still, my lived experience is these extrema cases happen enough that I can get kicked on joining with some minion oriented characters, and I've had swamp world death happen maybe 4 times playing with pubs.

Edited Friday at 11:06 PM by Walrusst

[Sanitar]

This will never happen, because the game has to be redone from scratch. Klei made a mistake at the beginning of the development of DST, which is now impossible to fix without wasting billions of time.I really hope that Eslewhere will be many times better in optimization than DST and there will be no such thing as 2 simultaneously running servers for the solo world.

[Walrusst]

Oddly enough one thing thats saved me a few times with the issues dst's really had deaths in its tick rate is abandoning the overworld shard for a few days with the whole server then returning to the surface after everyones gone insane living like mole people.

The shards make a lot of issues and make mods like island adventures very difficult to run, but its also been the savior once or twice so while I acknowledge its a really strong problem, I have much more mixed feelings regarding them because of weird stories like this.

But yeah, if the new engines built on a better foundation this gets much easier without invasive mods that kinda attack the spirit of the game to make long term sandboxy worlds possible. My only hope I have to pick straws at is hopefully we don't lose too many good characters in the move to elsewhere, I imagine the base cast will be slimmed down but hopefully we can catch back up there.

Plus a new game does have one more benefit, it gives us a chance to experiment while being free of dubious legacy decisions to make new decisions. Some will be good, some inevitably probably risk becoming the same sort of legacy, but that means that even if the games audiences are slightly different, we can have two groups of people who are happy. At least thats the hope.

Edited Sunday at 01:32 AM by Walrusst