Print

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

[darkflagrance]

As far as I know, there is no conclusive testing that shows that central stairwells or up/down stairs lead to greater lag than, say, horizontal fortresses with ramps. It is entirely theory based, using what is hypothesized about how z-level pathfinding works.

[wuphonsreach]

Main advantage of a cube / vertical layout, however is that average distance between any two points can be driven down to 1/2 or 1/3 of a horizontal layout.

I'm running on a 3x3 embark, with a 2x2 main stairs that goes from Z145 to Z110, with the trade depot at Z140, residences down around Z125, smelters down at Z110, surface farms up at Z150.  Everything (nearly) in the fort is within 20-25 tiles of the 2x2 stairs.  So a worst-case path length is only about 80 tiles.  The only hard part about the design is finding the gaps in the caverns so you can drive straight down to the magma sea with a single 2x2 stairway.  With the ( primary hotkeys of F1-F8, I tend to set them to focus on eight different Z-levels, which helps a lot with navigating a vertical fortress.

Sitting at anywhere from 60-90 FPS with 70-some dwarves.

[echonic]

I don't understand how you can be running that slow with what you described...

I have a several year old machine and even on a 4x4 embark where I'm pumping up magma 120 z-levels with an artificial water generator giving me power I can generally run a steady 60+.
I have 120 dwarves, I hardly ever atom smash anything, and there's hundreds of corpses everywhere from the constant invasions...

[jellsprout]

In one my of early .31 forts I decided to test this out. I had a fortress with a central staircase. When my fps was round 20, I decided to wall of the staircase and dig a linear ramp from my fort to the surface. The result? Nothing. The fps was the exact same when I converted to ramps as it was when I used stairs.
In my experience it doesn't matter if at all if you use stairs or ramps. It doesn't matter if I use a central stair or ramp, all my fortresses tend to run at the same speed and get the sudden fps drop at the same time. The problem with ramps is that they take up a lot of space. You can simply have a 2x1 pillar of up/down stairs down to your fortress, but a linear ramp with a width of 2 tiles can run across the entire length of the map to reach the magma sea. If you go for a spiral ramp design, it takes up even more space and your dwarves will generally have to take a longer path.

A possible way to test this and finally confirm or bust this myth is by having a fortress with three ways to get from the surface to the magma sea. One with up/down stairs, one with a linear ramp and one with a spiral ramp. Construct floodgates or a raising bridge at both ends of each path and make sure at all times two of these paths are closed. Every once in a while have all your burrowed to your meeting hall or something through the use of military alerts, and open a different path. Cycle through this a few times and it should become clear if one of these methods is significantly slower or faster than the others.

[Niseg]

I fixed my waypoint example. it saves waypoints now but  click sqrt(dx^2+dy^) and then Path Cache update. results :

the fixed vector heuristic A* :
visited nodes:849, length= 131 . 
A waypoint based navigation:
visited nodes:37, length= 133
path smoothing the end with A* :
visited nodes:81, length= 135

Path cache size is path cache Size:1142  which translates to 4568 bytes in current representation (x,y in 2 bytes each ) or  571 bytes~ in a more optimal representation (nibble mode 16 directions stepping ) for  (n-1)² = 16 paths. Even with admissible heuristic A* visits every single node on the map .

Think about how fliers try to enter the fort through a tiny hole and you'll understand some of the issues that cause FPS drops.

[Khift]

The argument about ramps vs. stairs is entirely reliant on pathfinding to be the main FPS draw. And honestly, I call balogna on that. I'm sure that the ramps do improve pathfinding efficiency. But it'll never improve FPS by more than a single point. Pathfinding is simply not something that happens hundreds of thousands of times per second needed for the small amount of efficiency generated by using ramps going to add up into noticeable improvement.

If you can take a slow fort (20 FPS or less), remove everyone's jobs so they all go idle in the dining hall and barely see an improvement in FPS then it isn't pathfinding that's wrecked your FPS. It has to be something else.

[agatharchides]

Yet sieges regularly cause a 10 FPS drop for me.

[Kogut]

And ambushes around 50

[EmperorJon]

Got 50 goblins, about 30 giant bats running around outside, all 3 caverns breached, 130 dwarves all moving around doing stuff, 50 or so animals, a river flowing, etc. etc. etc.

85FPS. It's not DF that's the problem. Now, the clothes bug may hit me soon... in which case, it will be.

[Korva]

One of my forts' FPS plummeted inexplicably lately so that I shelved it. That was in 31.18. I don't understand it really, other forts had more space dug out, a bigger population, more junk, more liquids flowing around ...

Can't seem to fix it for the life of me. Ah well. It's as good an excuse as any to finally get "serious" about a fort in the new versions.