Print

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

[LegendaryTyper]

Hello Urists!

I am looking for potential candidates to help conduct a fps experiment. The experiment concerns how quickly fps drops for different fortress sizes ( I like to call it "how long it takes a DF fort to experience a heat death") fortress fps information will be collected and graphed.

Everyone is welcomed to participate
Those who wish to conduct this experiment just need to start a fort somewhere far away from goblins and preferably on a world with no invaders of any kind in order not to interrupt the experiment.
Then make the Fort self sufficient and leave it to run until an fps drop is noticed


Here are some fort ideas

1- A small Fort locked underground (not necessarily if no invaders or killer animals). The Fort should contain a minimum of 25 - 30 dwarfs no animals.

2- A medium Fort with up to 60 dwarfs and 20 animals max

3- A large Fort with 60+ dwarfs and 20+ animals

4- !!!!Mega Fort!!! A gigantic Fort! Basically you goal is to crush the fps as quickly as possible

Create a copy of the save file every 5 years (1 is better) and then upload your saves once an fps drop is noticed (20 fps or less) to Dropbox or something like that and post a link here.

If you wish you could also post an fps average for every 5 years.

This information will be used to plot a graph and estimate the fps for say a Million df years or so ( should be fun  )

With this data we'll be able to further research how long can a certain sized Fort survive with different fps drop prevention method and which are most effective in fort life span lengthening and not to mention if you seal your fort we'll be able to see if your fortress "emos itself to death " ( quote from wiki) before it experiences an fps death  .

Anyways thanks for reading this through and please tell me what you think about this experiment or whether you have any ideas,improvements or questions.

If you wish to contribute here is a set of rules (might be updated)
1- post a link to the saves
2- also post your specs cpu is very important
3- make a regular world but in advanced settings turn off all vampires and other monsters also change the limit on immigrants to whatever fort you're making (small large etc) and finally settle away from goblins and everything else preferably to avoid invasions.
4- only vanilla df no tile sets
5-for now keep enough stockpile space to store necessary supplies
6- take a bunch of pics showing the fps or make a video and post it with the saves
7- whoever starts gets to choose the df version (I couldn't decide which version to choose but the newer the better)

[Sanctume]

Won't you need to define a few variables and know their initial values in order to compare?

1. CPU specs
2. RAM capacity and latency specs
3. OS tasks control, as in are there other processes running concurrently with DF? i.e. browsers, YouTube/Twitch streams, other Window's services., etc.
4. DF.exe affinity as in low, normal, high, etc.

And then, the parameters for each world-gen the fortress is tested on
1. World size
2. History length
3. Historical figures
4. Cavern count
5. Embark size
6. Number of biomes, what type of biomes
7. Temperature on?
8. Weather on?
9. Number of liquid tiles flowing

10. Population / Idle count
11. Number of stockpiles open / number of job tasks
12. Number of multi-tree tiles, seasonal changes, are trees shedding leaves?
13. Invader count, mounts?
14. Visitor count

There are plenty of variables that will affect FPS. 

Narrow down a set of parameters so that test fortresses have references to compare with each other.

[Melting Sky]

You could collect all the saves and then run them on the same machine for comparison, but I think for the most telling results there should be a lot of controls. For instance running the same exact fortress where as little is changed as possible before and after a cat explosion or before and after generating 7,000 misc. items. Another good experiment would be to keep the fort population and item count roughly the same but increase it's size by 200% etc.

[Malakor]

I would post the PC/Laptop specs then.
- CPU
- RAM
- OS with processes given (youtube/Soundsense/Whatever you desire, adding it as BG-Feature in the "data")
                                       
-DF on Normal to get the most basic

And for the World-gen i would set everything to medium as standart, but would use different sizes.(Temp and weather on)
So it would be:
(WS = Worldsize)
Small Fort , Pocket WS, Small Fort, Smaller WS, Small Fort, Small WS etc.
And same with the other ones, medium, large and !!!!MEGA!!!.

Embarksize i would go for 3x3. (Except if you choose !!!!MEGA!!!, maybe) and one biome only, if possible.

As in ingame i would fiddle around with the population being idle. If the Fps drop lower with more idlers or more non-idlers(adding it to data). And Stockpiles would depend on the Forts size. Maybe we could figure out something for that. Also i would mention if its a surface or underground Fort. Plus everything vanilla, so no edited raws other than with LNP, like tweaks, Tilesets etc.
(Would also mention it in the data)

As in for trees i would go for an sparse area to chop them all right at the beginning! (beds bins barrels)
And about Invaders, i would add them if the FPS drops after thier arrival, and then count them with/without mounts etc.
Dunno about Visitors tho, maybe try to get as less as possible in that case.

At the other hand, to try out every difference possible would take waay too long(unless we get tons of people playing a fort at the same time and being done nearly at the same time too) and i assume, since the game also keeps developing, the FPS informations would be outdated before even done collecting all needed data. So yeah, that's what i would suggest in that case, standardization.

[Sanctume]

Let's say you use advance world gen parameter and give the highest embark points 99,999?
And you bring however many male cats you can get in the embark like 10,000 cats?

How would you be able to tell when the game starts after striking the earth prompt screen?

How will you measure FPS? Do you rely on the dfhack FPS monitor to measure things?

[Malakor]

In that case i dont need any testing, i know i will have below 0 fps. 
Tho, i would rather go for chickens. 

[Sanctume]

Well, if 10,000 cats will be instant 0 FPS, how about testing 5k, and keep reducing to get a range for a number of objects looking for pathing from day 1.

And let's say 2,000 is passable at 30 FPS.   The next step is to build 1,000 objects (or dfhack create 1,000 blocks). 
Then test 10 stockpiles for blocks at 10x10 size.  Enabled the 7 dwarves to haul them, and see if there is an FPS hit when only labors on are hauling. 

Now repeat by embarking 100 dwarfs from day 1. 

How about 200 dwarf embark?  Will 1,000 stockpile hauling labor do a significant FPS hit?

[LegendaryTyper]

Thanks everyone for your responses and ideas. I updated the rules accordingly. I'll post tomorrow the details on how everything will be tested as it's rather late now for me.For now I think once someone's Forts fps starts dropping another person with better specs will take over and continue until fps starts dropping again then pass it on etc then a chart will be made with this information

As for fps testing such as the catsplosion that's for later once we have the bare bones of the experiment.

Thanks again for the ideas and recommendations

[LegendaryTyper]

I'll also add some more rules that everyone recommended tomorrow

[Corune]

I'll be willing to do some tests.  Right now I have four  computers I can test on, soon to be five, all of various power.  If you want specs, I can compile a list in just a few hours when I'm off work.

[Daris]

A crocsplosion would be better than a catsplosion, as one can easily go from 10 crocs to 200 in minutes, providing a natural experiment.  There is a noticeable difference in FPS, and another noticeable difference when the crocs are caged up.

[LegendaryTyper]

I'll be willing to do some tests.  Right now I have four  computers I can test on, soon to be five, all of various power.  If you want specs, I can compile a list in just a few hours when I'm off work.

That's great! A spec list will be very useful. But for now start with your lowest spec pc and either upload it here with its specs or continue on a better spec pc once the fps drops (20 or less).

[Oafsalot]

I may do this. Not sure I hae the time yet.

I would add one thing, RAM speed and if possible latency. I'e always suspected RAM as being a big factor in fps death.

[Malakor]

RAM is, and so is the CPU!

I wish me a Single-Core CPU with around 6+ghz - stable.
Plus 16GB RAM, all that only for Dwarf Fortress. And then making a Fort with around 500 people on an 10x10 embark.
(or more, til it gets under 30 fps)
I wonder what would be faster, to acquire that CPU or DF being able to run Multicore.

[Overspeculated]

Since everybody has vastly different rigs you need to upload some reference saves in order to normalise the fps values.