Bay 12 Games Forum

Dwarf Fortress => DF Dwarf Mode Discussion => Topic started by: Kishmond on February 05, 2020, 01:06:29 pm

Title: FPS Testing: Need a developed fortress from a recent version
Post by: Kishmond on February 05, 2020, 01:06:29 pm
Hi,

I'm upgrading my PC from DDR4 2400 memory to 4000 (!) and I want to do a direct comparison of Dwarf Fortress performance. The amount of memory and number of sticks is staying the same. Does anyone have a several-year-old fortress with at least 100 dwarves they'd like to donate? It can be either the new version or the previous stable one. I don't have time to commit to a long fort right now.

I'll post the results here when I have them.

PC specs:
i7-9700K 3.6GHz, 4.9 GHz boost, 12 MB "Intel Smart Cache"
32 GB RAM (DDR4 2400 Crucial Ballistix Sport 8 GB x 4) upgrading to (DDR4 4000 Patriot Viper Steel 8 GB x 4)
MSI MPG Z390 Gaming Pro Carbon motherboard
Samsung 960 EVO 500 GB NVME SSD
GTX 1080 Ti
Title: Re: FPS Testing: Need a developed fortress from a recent version
Post by: NordicNooob on February 05, 2020, 02:23:01 pm
You could try Archcrystal (http://dffd.bay12games.com/file.php?id=14574). It may be a bit too large for your tastes, I think it has like 200 pop, and is also ancient (it has been updated through a few versions), which likely has some impact. If you're just comparing FPS between computers it should be fine, though.
Title: Re: FPS Testing: Need a developed fortress from a recent version
Post by: Fleeting Frames on February 06, 2020, 06:17:03 am
https://www.reddit.com/r/dwarffortress/comments/dewey9/pop_1500_megafort/ has a save with 1500 dwarves. I get about 2 FPS on my laptop; though this FPS might be too low for you

A bit higher FPS comparison would be a Breadbowl save: http://dffd.bay12games.com/search.php?string=Breadbowl&id=19, or five fountains http://dffd.bay12games.com/file.php?id=11556 if you want something without vegetation.
Title: Re: FPS Testing: Need a developed fortress from a recent version
Post by: Kishmond on February 13, 2020, 11:09:39 pm
NameBeforeAfter
Cheesesplatter350-450380-500
Fivefountains400Not tested
Archcrystal3035
Lashmoistens (1500 dwarf fort)3-43-4

It's a shame the biggest changes are in the forts with FPSs high enough that the difference doesn't matter, but the difference in Archcrystal is encouraging.
Title: Re: FPS Testing: Need a developed fortress from a recent version
Post by: Superdorf on February 13, 2020, 11:45:44 pm
35 FPS on Archcrystal is pretty amazing. Sethatos was getting, like 14 at the very most.
Title: Re: FPS Testing: Need a developed fortress from a recent version
Post by: Fleeting Frames on February 14, 2020, 05:29:00 pm
Thanks for testing, those are interesting results. (Also very good in comparison to my years-old laptop.)
Title: Re: FPS Testing: Need a developed fortress from a recent version
Post by: knifemind on February 21, 2020, 05:26:39 pm
Quote from: Kishmond on February 13, 2020, 11:09:39 pm
NameBeforeAfter
Cheesesplatter350-450380-500
Fivefountains400Not tested
Archcrystal3035
Lashmoistens (1500 dwarf fort)3-43-4

It's a shame the biggest changes are in the forts with FPSs high enough that the difference doesn't matter, but the difference in Archcrystal is encouraging.

True, but even going to 5 from 3-4 would represent a 25% or more performance increase.  It would be unreasonable to expect that kind of performance jump from any software with just a RAM swap (unless you had REALLY crap RAM to begin with.).  I would think that a fort getting near 100 FPS for an "average" system would be a good benchmark because then we can at least gauge change by (or closely to) individual percentage points.

Also, what about latency?  You made no mention of it for either RAM set and I would think that lower latency would also have at least a measurable impact.