Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Skuggen]

He said he uses small region.

Do old versions run well on the same machine? As mentioned it's pretty odd you get so poor performance on that system.
How long histories do you generate with?

[MDFification]

Just a reminder: there's never been a single test that indicated that world size has anything to do with FPS. When DF2014 first came out, people came out of the woodwork claiming that was the case. These claims were groundless. The main cause of FPS drops in DF2014 are the giant amounts of contaminants dropped by the new trees and changes to how pathfinding calculations work (what with climbing and jumping needing to be factored in - and remember at first animal pathfinding was completely broken in regards to doors, and flying pathing in regards to trees, causing giant FPS drops and Trees becoming DF2014's King of Beasts).

A suggestion; it's pretty simple to mod your game so that all subterranean creatures spawn in the first cavern layer. If you do this, you can generate worlds with much, much better fps by making sure only 1 cavern layer spawns.

[Putnam]

The way the game seems to speed up when unretiring suggests some other cause.

[Eldin00]

If I recall, contaminants don't persist across a retire/unretire cycle, which would make a speedup on unretire consistent with contaminants from trees being a major contributor to FPS death.

[Putnam]

Cleaning contaminants with DFHack doesn't seem to have as much effect.

[Eldin00]

Yes, it's definitely true that cleaning the contaminants with DFHack has far less impact than a retire/unretire cycle. But I haven't looked closely at what exactly DFHack does when cleaning contaminants, so I guess it's possible that some remnant of the contaminant hangs around in memory (even if it's just some space allocated in a memory structure somewhere or a now blank entry in a list) after it's removed, but that said remnant is simply never created when unretiring. Or it's possible that the contaminants are a red herring, and the major contributor to FPS death is something which has yet to be identified.

[omniclasm]

So, in the save folder...

Pre-retire, the "world.sav" file is 7,250kb with 40-50ish FPS

Post-reclaim, the "world.sav" file is 6,250kb with 80-90fps

Which, while monitoring that during the time after the previous reclaim, that roughly corresponds with the FPS experienced while building up to that 7,250kb. World.sav seems to gradually grow the longer a fortress is around. Retiring and reclaiming seems to chop a big chunk off that number.

[omniclasm]

Looking at an older fort from 34.11, that had been around for a long time with 120+ dwarves, and the world.sav file is 6,900kb. The FPS in game is ~70, which corresponds to roughly the same curve. In the same save game as above, I deleted every item on the map (again), and world.sav size went down to 6,800kb, and FPS went from 40fps to ~70fps.

[utunnels]

Needs further research, but for now I'm blaming this fort's issues it on what I'm calling an "undead cascade".

Yeah, undead can be a problem if you let them inside your caverns.
You'll experience FPS death eventually no matter how powerful your hardwares are.

[omniclasm]

Cleaning all contaminants with dfhack dropped world.sav size from 7,375kb to 7,300kb, with no noticeable FPS change

Using exterminate to kill the 10 dogs in my fortress then increased it from 7,300 to 7,325

[Putnam]

Could you upload before+after unretire saves? It could really help determine exactly what's going on with DFHack.

[omniclasm]

Where would one post a dfhack script?

Threw one together that makes undead attributes start to deteriorate after 6 months, and then has them die (for good) after a year. I figure that might help with the occasional undead cascade in the caverns, as well as a mildly interesting mechanic.

Of course, with modifiable thresholds

[Jorn Stones]

I've been capable of running a 5x5 embark with about 2/3 of that being woodland and full of contaminants, with about 110 adult dwarves and 90 childs, and about 120 or so animals just fine at a roughly stable 20-30 fps.

Until the giant Kea's came along. They destroyed the fps in an instant because of something.

So.. Whenever I my fps die, I take a look at the units and if I see giant keas; I exterminate giant_kea with Dfhack, otherwise its just unplayable.

its possible other creatures can cause this to happen as well, but its a guessing game to figure out which.

[variablenonsense]

I do know that making doors pet unpassable seems to drop my fps dramatically, whereas making them forbidden doesn't; for some reason pets keep trying to path through them, I assume, even though they aren't allowed to do so. Animals seem to still have some oddities regarding pathing - I had an extremely tiny fort that crawled to a halt for no explainable reason (had never breached the caverns) .. never could figure out what was wrong, but there -were- giant owls on the map. Another map (same size, same world) has at least twice as many dwarves, caverns open, and is chugging along in the 40s-50s.

It's anectodal, but it's possible fliers freak out the game, or animal pathing in general. On a reclaim, does the fps increase persist (if you play it for a while)? Do animals on the map change on reclaim?

[arbarbonif]

I had a giant lag spike (100+ FPS down to single digits) in a fort a while ago while I was building a floor bridge over a river canyon.  Turned out that the dwarves were climbing over the trees that overhung the river to place blocks on the other side, but only after exhausting EVERY possible path that didn't involve climbing.  It recovered as soon as there was a non-climbing path to get there (of if I suspended the far side).  I haven't had problems with flying things causing lag, haven't had giant keas but the buzzards should do similar pathing (and didn't cause any problem).