Print

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

[xmakina]

Switch of invasions on the medium world. I read that that can be a huge contributor to FPS as the game has to analyse the whole world to see if anyone is invading you.

No, that's silly. I'm playing FOR the invasions.

Well, yeah, what else are you going to pour magma on? I would still be curious if this was an FPS hit or not though, if only so we can add/verify it in the wiki.

[Urist Imiknorris]

Well, yeah, what else are you going to pour magma on?

Did you actually ask this question? There's always something to pour magma on.

[Rask]

You have problems with ram or it is a glitchy embark maybe? Below is my testing:
2x2 embark on small world: 1200+ FPS
2x2 embark on large world: 1200+ FPS
2x2 embark on a particularly wierd large world: 300+ FPS

Additional tests:

new medium world: 180 FPS   (with much increased civilisation count)
new small world: 500 FPS (standard civilisation count)
new pocket world: 420 FPS (increased civ count)

Seems the FPS is both dependent on the number of civilisations in the world and on the size of the world.

Current theory: checking for invaders on larger worlds takes longer, even if there are more civilisations on the smaller world? Did you run your tests with invaders on or off?

[rephikul]

You have problems with ram or it is a glitchy embark maybe? Below is my testing:
2x2 embark on small world: 1200+ FPS
2x2 embark on large world: 1200+ FPS
2x2 embark on a particularly wierd large world: 300+ FPS

Additional tests:

new medium world: 180 FPS   (with much increased civilisation count)
new small world: 500 FPS (standard civilisation count)
new pocket world: 420 FPS (increased civ count)

Seems the FPS is both dependent on the number of civilisations in the world and on the size of the world.

Current theory: checking for invaders on larger worlds takes longer, even if there are more civilisations on the smaller world? Did you run your tests with invaders on or off?

my large worlds have 200 civs and I play with invader on. I happen to have 6 worlds right now so lemme try all of them

[Rask]

Generated some new worlds, standardised to produce 5 civs and run for 2 years of history.
Embarks on 2x2 square with play now on flat terrain:

medium: around 300 FPS
small: around 750 FPS
pocket: 540+ FPS

For all of these, invaders were OFF.

The data for the small and pocket world looks pretty weird, though medium is slower, as expected. Possible explanation: small and pocket worlds are both so tiny that the number/type of civilisations (the remaining random factor, aside from embark terrain) becomes the relevant factor?

[rephikul]

tested again with normal playing set up, "PLAY NOW" on warm biomes and fps capped and monitored

2x2 large, 400 fps with weighted average
2x2 pocket fluctuate alot, but it's around 1900-2000

So yes, world size DOES affect fps to some degree. It's not known whether this is a scaling issue, or it's a constant fps drain and shouldnt be worried about.

[shlorf]

If you could gen some worlds to allow 200 dwarfs on embark and get near identical embarks, you could check whether the civilization related calculation only eat a meaningful amount of fps in high fps scenarios or if it will still be a 4x difference with some more pathing calculations going on.

[Girlinhat]

Keep in mind the local terrain and such.  If you're going to do an empirical test, you need empirical data.  This means weather and temperature off, as these can be unaccounted for variables, and consistent embarks.  I'd suggest a world designed to be 100% flat, with no volcanism and below freezing, and no caverns/HFS/etc.  Then, you can make various sized embarks on various sized worlds and compare data.  It may be required to add biomes to entities, so that they'll settle on glaciers, to run the test about civ counts, but this shouldn't actually interfere with any FPS counts.

[Rask]

Sadly, creating such a world is beyond me.

In any case, recent imperfect experiments suggest that world size does indeed affect FPS significantly from medium world map sizes upwards.

[daishi5]

Doesn't DF do weather modeling, that might be a big part of why world size affects the FPS, bigger world has more weather to track? 

[Mickey Blue]

Tried my own test (just one large and one small, would like to test with more..), both with similar setups, genned to 250 w/ only one cavern layer and everything else set as normal.  Tried to select two similar sites (both temperate heavy forests w/ no running water or magma).

Large FPS hovered around 200, small FPS hovered around 550... I'll have to test this more in the future with some more maps keeping track of number of civs and all that but it does appear from this and other tests that world size might have a notable effect on FPS.


Other things I have done are to limit myself to 2x2 map size, turn caverns to 1, try to avoid running liquids (such as rivers, etc), keep animals caged, use a dwarf cleaning system (I use the trench.. But the shower listed earlier sounds neat) and I use DFHack to get rid of blood if it gets too much (which I don't consider cheating as there really isn't another realistic way to get rid of blood/fluids from the map and I'm only doing so because of issues with FPS, not to give myself some unintended advantage in game.  If there were a more realistic way to clean fluids and/or if fluids didn't have such a damning effect on FPS then I wouldn't care).

[blue emu]

 

[Urist Imiknorris]

I believe I have posted a screenshot of an identical event before...

EDIT: Not quite, but close:

[Buttery_Mess]

Large worlds certainly do reduce FPS noticeably. I presume that the reason for this was that a larger world will produce a significantly larger save file. I'm guessing that most of the save file (if not all) will be loaded into the RAM during play. Again, I'm guessing that DF also performs regular checks and small calculations on every tile of the game world, every civ, megabeasts and so on, for the purposes of creating history, weather patterns, politics and so on. I don't know how often it does this, but I do know that doubling the dimensions of a world quadruples the amount of data required to describe it and thus the length of time it takes to perform calculations and checks.

This is mostly guesswork, I don't know the finicky details of how the game works. All I know is, smaller maps with less stuff on them mean a faster game.

[abadidea]

Large worlds certainly do reduce FPS noticeably. I presume that the reason for this was that a larger world will produce a significantly larger save file. I'm guessing that most of the save file (if not all) will be loaded into the RAM during play.

I run on Linux with no swap, so I can attest that it isn't a memory size problem, since nothing's getting paged out for me. I am guessing it's more with repeatedly traversing linked lists of other civilizations and what they are doing, and weather, and forgotten beasts roaming around.