Chunks sometimes take ages to load

[yolistenupheresthestory]

Hi! I just got this game, and I'm loving it so far, but when I recently began actively exploring and rapidly moving long distances in a single direction I ran into a pretty obnoxious issue: sometimes I'll reach the world's end! The missing chunks have so far eventually generated given time, but it takes several minutes sometimes and is happening somewhat frequently since I began playing nomad; what gives?

Please see included screenshots for examples.

[yolistenupheresthestory]

This is going from immersion-breaking to unplayable. Now I refused to load the new chunks at all, even after waiting a long time, and when I finally just restarted the game I found myself stuck in a single isolated chunk surrounded by void for several minutes before all the neighbouring chunks finally started generating a few minutes later. Is this normal? My savefile is currently about 950mb.

[Thorfinn]

That's typical if you have a machine on the bottom end of the specs, memory or CPU, or if you are running higher view distances. You might try lowering view and choosing the setting for more aggressive memory management. I don't remember what it's called, Optimize RAM, maybe, but I think its on the Graphics tab in Settings. 

[EDIT]

I'd add that I don't know what mods you are using, other than the one, but some can make mapgen slow to a crawl. That would be the first thing I'd rule out. Just create a new game with all but the core 3 disabled, and go for a run.

Edited February 4, 2025 by Thorfinn

[gilt-kutabe]

I’d add that the ‘Hydrate or Diedrate’ mod (I think that’s the one you have?) seemed to cause some lag/loading issues for me when I tried it on the current version. Not exactly like you have, but if a mod is causing the problem, I’d start with that one.

[yolistenupheresthestory]

Hi, thanks for the input! I'm running 32gb RAM, 4x3.60ghz processor, RTX 3060 and have over 70gb free space, so I don't think the resources should be lacking at all; I am, however, running a whole laundry list of mods (see screenshot lol). I hope none of them is causing it, since I've been enjoying them a lot, but I did change and finally remove one which was supposed to add new types of gravel blocks earlier since it either didn't work as intended or had a conflict with one of my other worldgen mods; might that change have broken something when generating new chunks on the same save?

I should perhaps mention that this didn't start until I'd played a good deal; early on the game ran like butter and even with this strange ghost-chunk issue there are still zero other performance issues to speak of.

For now, I'll try what you suggested and if that works fine I'll try flying around through a new world in creative mode with all the mods enabled to see if it still works; hopefully it was the mid-game disabling of a mod all along.

 

[Thorfinn]

Doesn't Windows itself complain about having only 70g? There could be a lot of swapping going on there, and, yeah, that would happen as you get larger files, and your machine is constantly having to clean parts of the recycle bin to get space.

I take back the mod thing now. Figure out a way to get more disk space. On a machine like that, I'd probably add another SSD dedicated to VS. $50 for a 500g drive is a no-brainer.

FWIW, only the things that affect worldgen should be giving you those kinds of hiccups. So the flowers series, the biome, rivers, underground mines, that kind of thing. Last I checked, the FotSA added only runtime spawns, so I wouldn't worry about them at all.

 

Edited February 4, 2025 by Thorfinn

[yolistenupheresthestory]

70 is bad? The sysrequirement page on the VS homepage itself says 3 is minimal, 6 is decent and 10(!) is "optimum" for "60+ fps on high and very high graphics quality settings, with mods". Am I missing something? Windows only ever complained about disk space when I got down below 30 or something thereabout at some point..

[yolistenupheresthestory]

Well, I tried running it without the swarm of mods enabled, made a new world, started running straight through it and sure enough: voidlands.

As a bonus, this time the log that shows up on shutdown just lingered without any text for a very long time.

My machine runs games like Shogun 2 and Stalker Anomaly just fine at the highest settings, but this one game for some reason has a problem with me seems like; this hasn't happened once ever since I invested in a powerful gaming PC years ago, so I'm stumped..

From the log:
5.2.2025 04:52:12 [Warning] Server suspend requested, but reached max wait time. Server is only partially suspended.
5.2.2025 04:52:12 [Notification] Unable to autosave, was not able to pause the server
5.2.2025 04:52:12 [Notification] A client reconnected, resuming game calendar.
5.2.2025 04:52:12 [Notification] Server ticking has been resumed
5.2.2025 04:52:12 [Warning] Server overloaded. A tick took 5290ms to complete.
5.2.2025 04:52:17 [Warning] Server suspend requested, but reached max wait time. Server is only partially suspended.
5.2.2025 04:52:17 [Notification] Unable to autosave, was not able to pause the server
5.2.2025 04:52:17 [Notification] A client reconnected, resuming game calendar.
5.2.2025 04:52:17 [Notification] Server ticking has been resumed
5.2.2025 04:52:17 [Warning] Server overloaded. A tick took 5122ms to complete.
5.2.2025 04:53:04 [Notification] Server ticking has been suspended
5.2.2025 04:53:04 [Notification] Last player disconnected, compacting large object heap...
5.2.2025 04:53:06 [Notification] UDP: client disconnected insideandoutside
5.2.2025 04:53:06 [Event] Player insideandoutside got removed. Reason: Server shutting down - Exit request by client
5.2.2025 04:53:06 [Notification] Server stop requested, begin shutdown sequence. Stop reason: Exit request by client
5.2.2025 04:53:06 [Notification] Entering runphase Shutdown
5.2.2025 04:53:06 [Notification] Defragmented listener lists

[Steel General]

Tall worlds take a lot longer to generate - if you've got a huge depth of ground under you, that might explain part of it.

I'm running on Win7 with 1GB vram and ~7GB swap space, and I only get effects like that if I'm running continuously with a large view distance, and even then it loads very quickly after I reach the edge.

[Thorfinn]

I've noticed the same thing, @Steel General. I get better performance on an older i5 running Win7 than I do with an i7 with better specs all the way around running Win10. I am so not looking forward to the inevitable change to .NET8. Probably by then, I'll have to switch something over to Linux rather than deal with all the Windows bloat.

I have no idea how you get by with so little swap space. I can watch the specs on my second monitor start to get worse when I get below about 25% of my SSD. At 15% the full-scale spikes on usage start getting disturbing wide -- over half the time, it's pegged. And that's at a piddly 512 view.

@yolistenupheresthestory,  5 seconds is a long time for a server to be bogged down. I've not seen 5 seconds, but waiting a couple seconds for those blocks to fill in is not unheard of. With your system, I'd think the mapgen would probably keep up, but it would also depend on what else your 'puter is doing in the background. Browsers in particular are awful. I've taken to not only closing them down, but also in opening up Task Manager and making sure there's no residual crap still taking up space.

[yolistenupheresthestory]

@Steel General The only parts I tend to mess with when it comes to worldgen (apart from gameplay settings etc) are land coverage and equator location, so it shouldn't be terribly tall, though.. It's a bummer, but for now I'm just back to moving slowly through the world and settling in one place rather than wandering too far since the only real performance issue I've encountered has been chunks refusing to generate.

@Thorfinn So the game actually runs better on Win7 even with older hardware? I was already planning on jettisoning my old dual-boot to go Linux-only once EoL comes for yet another Windows install I'd finally gotten used to, and it just seems like a better idea all the time lol

Somehow, I hadn't even thought of opening up the task manager since the PC doesn't seem stressed out in general, but I'll definitely have a good look at it whenever I have chunk-loading issues again; thanks!

[Thorfinn]

Y'know, maybe I have seen this on my Win10 machine. It is prone to just locking up and eventually crashing after a month or so of game time, and always after a lot of sustained sprinting, though not with me standing at the void. But maybe that's because I'm in already explored territory. Single-block resolution is somewhat less than 500, so it's not obvious that the cache is the problem because my new blocks are all being created at a greater distance than that. My normal exploration is sprinting between things -- flax, copper, parsnips, that kind of thing, and taking fairly frequent pro-picks, so things have a chance to catch up.

That would explain a lot. As things start to stutter, I'm frequently seeing more than 2g of RAM taken up by shared, i.e., due to be shifted to virtual memory, or possibly delay in updating the world file, and when it goes much above that, if I don't stop, it eventually crashes.

Edited February 6, 2025 by Thorfinn

[yolistenupheresthestory]

Yeah, staying in explored territory has given me zero issues, which is why I didn't even notice anything was wrong until my initial practice-base finally got boring; if you want it to trigger, then what I mentioned in the other thread is what'll do it: travel in a single direction without deviation to make sure the game has no time to sluggishly generate the chunks long before you get there (it will sporadically trigger even at slower exploration, but this is a sure-fire method on my machine), made easier by generating a world with hunger at slow and enemies at their weakest to remove any distractions from sprinting.

It would be interesting to see if it's the same on your Win10, too.

As I mentioned in the other thread, my go-to playstyle in these kind of games is generally the opposite of yours (and, I assume, most players): I'll learn the ropes of the gameplay, gather up a set of essential survival gear (in VS, that was a stack of flint, a raft+oar, bed, firestarter and a portable "tent" of haybales, with anything else being procure-on-site)  and then go exploring to see what the world looks like and how far I can get; I was hoping to make it to one of the poles eventually, but seems like that won't happen anytime soon lol

[Thorfinn]

Huh. My crashes have been mostly in explored areas. Spent some time digging through the crash logs, and noticed this:

System.IndexOutOfRangeException: Index was outside the bounds of the array.
   at Vintagestory.GameContent.RoomRegistry.FindRoomForPosition(BlockPos pos, ChunkRooms otherRooms) in VSEssentials\Systems\RoomRegistry.cs:line 395
   at Vintagestory.GameContent.RoomRegistry.GetRoomForPosition(BlockPos pos) in VSEssentials\Systems\RoomRegistry.cs:line 379
   at Vintagestory.GameContent.EntityParticleInsect.playsound() in VSEssentials\Systems\ParticleEntity\EntityParticleInsect.cs:line 112
   at Vintagestory.GameContent.EntityParticleInsect.doSlowTick(ParticlePhysics physicsSim, Single dt) in VSEssentials\Systems\ParticleEntity\EntityParticleInsect.cs:line 39
   at Vintagestory.GameContent.EntityParticleSystem.OnNewFrameOffThread(Single dt, Vec3d cameraPos) in VSEssentials\Systems\ParticleEntity\EntityParticleSystem.cs:line 427
   at Vintagestory.GameContent.EntityParticleSystem.onThreadStart() in VSEssentials\Systems\ParticleEntity\EntityParticleSystem.cs:line 192
   at System.Threading.ExecutionContext.RunInternal(ExecutionContext executionContext, ContextCallback callback, Object state)
--- End of stack trace from previous location ---
   at System.Threading.ExecutionContext.RunInternal(ExecutionContext executionContext, ContextCallback callback, Object state)

There appears to be a problem with insects. Might be why the silly thing is so hard to duplicate.

[EDIT]

Incidentally, next time it happens to you, check your log files. So far, all of my unmodded crashes have been "explained" in client-crash.log

Among them:

System.IndexOutOfRangeException: MapRegion data, index was 2004147647 but length was 512

Which appears to be butterfly-related.

Edited February 8, 2025 by Thorfinn

[Thorfinn]

On 2/6/2025 at 11:59 AM, yolistenupheresthestory said:

As I mentioned in the other thread, my go-to playstyle in these kind of games is generally the opposite of yours (and, I assume, most players): I'll learn the ropes of the gameplay, gather up a set of essential survival gear (in VS, that was a stack of flint, a raft+oar, bed, firestarter and a portable "tent" of haybales, with anything else being procure-on-site)  and then go exploring to see what the world looks like and how far I can get

Nah, that's pretty much the same, except for the preps. I gather stuff on the way. I usually only choose a place to settle down when I have 60 copper, or I'm out of inventory space, which is generally with linen sacks by then. The only time I deviate is if I find somewhere way too sweet to pass up.

I turned on coordinates the other day just to see, and covered at least 14,000 blocks in one game-day, sprinting almost all the way. At least, because while I was intending to keep heading south, I was having to go east and west to get around mountain ranges and other crappy terrain. That and how else am I supposed to accumulate all that flax if I don't try to stick to regions where it grows?

Edited February 8, 2025 by Thorfinn

[Steel General]

On 2/5/2025 at 12:33 PM, Thorfinn said:

I've noticed the same thing, @Steel General. I get better performance on an older i5 running Win7 than I do with an i7 with better specs all the way around running Win10. I am so not looking forward to the inevitable change to .NET8. Probably by then, I'll have to switch something over to Linux rather than deal with all the Windows bloat.

I have no idea how you get by with so little swap space. I can watch the specs on my second monitor start to get worse when I get below about 25% of my SSD. At 15% the full-scale spikes on usage start getting disturbing wide -- over half the time, it's pegged. And that's at a piddly 512 view.

The 1.20.x series have been giving me a lot of hard crashes, so it might not last much longer as a game machine (I'm already off Steam).

I have a single 1TB HDD, usually with 2-10 GB free; I use a view range of 208.  Many versions ago I used a much larger view radius and could see the March of the Time Being, but at some release the system could no longer handle the large view radius, so I turned it down until I stopped getting the frequent, huge lag spikes.

For some reason Vintage Story uses more RAM than Task Manager can find - the Vintage Story process claims to use ~2GB RAM, but another 7GB RAM go missing at the same time and return when Vintage Story is closed. All processes should be visible, but something isn't - I know it's running separate client and server processes, so I suspect one of those isn't appearing in Task Manager.

[Thorfinn]

4 minutes ago, Steel General said:

The 1.20.x series have been giving me a lot of hard crashes,

Noticed that. 16g is no longer enough to run the game smoothy at higher view distances. Even on my "crappy" machine, on 1.19, I was running 1024 without a hitch. Now trying that is a pretty much guaranteed crash, somewhere between immediately and 30 minutes, usually. 512 would occasionally barf. 32g fixed that, at least so far. No, not the 1024; that's likely gone forever, at least on that CPU/GPU. But 768 seems to work.

[yolistenupheresthestory]

Hopefully it will be solved relatively quickly; not unexpected for an indie game and totally forgivable assuming work is done to fix it as it feels kinda unreasonable to be able to run snazzy AAA games at the highest settings with zero issues but have a blockbuilding game try to devour my machine alive for some reason. I've heard that butterflies did cause some sort of issue due to their spawning patterns or something like that, so maybe that has something to do with it?

[Thorfinn]

Are you absolutely certain it's a VS issue, @yolistenupheresthestory? Reason I ask is I've started enabling all the dev debug aids, and I'm amazed at how many mods there are that have memory leaks. Dispose errors, I guess they call them. I don't know of any in your list, but then again, I only use 2 of them very often at all. I have not tried any of the rest of them in 1.20. And there were a lot of API changes there. Most of the content -only mods will probably be fine, but those with code are a whole 'nother story.

Anyway, I'd enable some of the dev options, for sure VAO logging, as that's one of the more common I'm seeing, and check your client-debug logs early and often.

[Jake Marshall]

@Thorfinn It might still be a VS issue, this thread is about the same issue:

which you might recall. The author of that thread expressed not using mods at all.

 

I've experienced this same issue myself, but I do have mods. My own personal guess is I noticed it after I added in a world gen mod, Plainsandvalleys in anticipation for my next fresh start. I'm wondering if maybe the pre-existing world is not acting friendly with new world gen guides. But I don't really know enough about how under the hood works to determine that.