Bay 12 Games Forum

Please login or register.

Login with username, password and session length
Advanced search  

Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Messages - Urist McNobody

Pages: [1]
1
DF Dwarf Mode Discussion / Re: Pathfinding is not a major cause of FPS death
« on: November 04, 2023, 10:42:44 am »
> Helix rampcases without a central hole are significantly better than staircases, since dwarves can see up and down stairs

Do hatches block line of sight?  Do glass hatches and doors block line of sight?

2
DF Gameplay Questions / Re: Writing New Music
« on: October 28, 2023, 10:01:07 am »
Yes.  And various guests do filter through to entertain the populace.  On rare occasion, a guest or resident will perform something that they know, and my citizens can learn it from them.  But I've never seen anyone come up with a new musical form that they didn't already know at the time that they entered the map.

Scholars will routinely publish new written works, and there is a leveling system for their knowledge (albeit a slow one).  Is there a similar system for artistic works?

3
DF Gameplay Questions / Writing New Music
« on: October 27, 2023, 07:37:04 pm »
5.11 steam version.  Is there a way for dwarfs to write new music / discover new musical forms?  My citizens include some dwarfs with musical skills, and know how to make many instruments, but know very few songs that use those instruments.

4
DF Gameplay Questions / Re: On the Dwarves and their Lye
« on: October 02, 2023, 07:58:39 am »
> I never have managed to get an oil branch to work out reliably

For completeness's sake, the step in the oil pipeline that doesn't work out for me is accumulating the pressed oil into a large pot/barrel in specific storage.  The oil just stays in the jug, which gets sent back to hang out with musical instrument parts**.  Lye and milk both successfully accumulate from their buckets to space- and labor-efficient storage.  It would be nice if oil did so, too.

** I could be weirdly specific about how the jugs are made.  Make them from nickel silver or somesuch, just to artificially set aside a pair of tool stockpiles for empty/full jugs that would at least keep them separated from instrument parts.  But the important missing element is the accumulation.

5
DF Gameplay Questions / Re: On the Dwarves and their Lye
« on: October 01, 2023, 09:35:07 am »
I don't have any trouble at all with the lye branch of the pipeline.  I never have managed to get an oil branch to work out reliably, so we rely exclusively on tallow-based soap.

Ashery: make 4 lye when lye available (the default option) < 10.

The soap-maker's workshop has adjacent stockpiles for lye and tallow.  Lye makes its way (slowly) into a large pot, which right now is holding 13 lye.

The soap-maker's order is: make 1x soap from tallow, when lye-containing items at least 1 and soap bars is less than 20.  By fiddling with the manager's criteria, I can observe that the one pot of 13 lye counts as exactly one lye-containing item.

One job "make soap from tallow" consumes a stack of tallow and a lye-containing item, producing min(n_tallow, n_lye) from them, returning excess to their stockpiles.

We're currently sitting on 44 bars of soap, 6 of which are currently in the hospital.  That's more over-production than I'd like.  Based on fiddling with the quantity criteria, I'm quite certain that partial bars of soap are not being counted by the manager when figuring out how many bars of soap are in the fort: the manager only "sees" 25 bars.  I'm pretty sure that this is the root cause of so many soap over-production issues, combined with the hospital's flow for partial items.  When a caretaker checks out a bar of soap to clean a patient, the hospital goes out and fetches another fresh bar, returning the partial bar to the general stock.  This leads to an accumulation of partial soap bars which are also not observed by the manager, which triggers the production of more fresh soap.  Rinse and repeat.  Thread and cloth have the same issue, leading to slow accumulation of partial spools of thread and bolts of cloth which cannot be turned into cloth/clothing. 

6
DF Gameplay Questions / Re: [Spoilers] How do you dig in the cotton candy packages?
« on: October 01, 2023, 09:13:12 am »
The gems won't smolder when they are still embedded in the wall.  So, channeling out obsidian is effective for allowing the fluids to drain without lighting off a !!diamond!!.  Also, a bucket brigade is sufficient to put out a flaming gem.  Set up a pond one level above the !!gem!! and the first bucket of water will put it out.

7
DF Gameplay Questions / Re: Military not properly equipping gear
« on: September 22, 2023, 11:30:05 pm »
Military uniform assignment is pretty buggy.

Yellow means that something in that category has been assigned to the soldier, but the soldier hasn't picked it up, yet.  Red means that nothing is available to assign to them.  I frequently run into a problem whereby the same item gets assigned to two different dwarfs.  Dwarf the second doesn't get what they were assigned.  When its traceable to named items, its always dwarfs in different squads that are claiming the same item.

8
DF Dwarf Mode Discussion / Re: Armor Material Science: how do materials compare against goblin weapons? 50.07
« on: March 14, 2023, 09:50:56 pm »
Thank you, this is very helpful.  I thoroughly appreciate frequency histograms like this for non-Gaussian problems.  A cumulative probability distribution is also nice, since it allows you to recover something meaningful from the vertical axis independent of the number of bins.  Therefore the bin size can just be the natural discretization level of the data: 1 wide.  It also lets you answer strategic questions like "what percentage of the time will a defender die in N or fewer blows in this situation?"  To the extent that all of the cases under test have long fat tails** thanks to the strength distribution of a Standard Goblin Horde, the "core" of the cumulative distribution should still help to tease out meaningful details between the armor classes.  Its hard to tell just by eyeballing the frequency diagrams and trying to imagine what their integral looks like, but there does appear to be a shift to the right for the bronze->iron->steel armor sequence.


** Those lashers, though.  If it wasn't for the fact that my silly dorfs have no respect for target priority, those guys just bubbled up to the top of the target priority list.

9
DF Dwarf Mode Discussion / Re: FPS Science and observations with 1000 dorfs population
« on: March 14, 2023, 08:32:49 am »
Putnam clarified in this thread that:  The bounding box for skipping the calculation is a cube, and that the early-out is for greater-than 26.  So dwarfs on level N can consider dwarfs on level N+26 for further evaluation, but not on level N+27.

10
DF Dwarf Mode Discussion / Re: Armor Material Science: how do materials compare against goblin weapons? 50.07
« on: March 14, 2023, 08:27:46 am »
OK, I'm familiar with RMSE.  Using standard deviation divided by sqrt(N) is perfectly reasonable when the samples are uncorrelated and gaussian-distributed.

I'm struggling with the level of dispersion required to calculate mean of 62 with sample standard deviation of 80 from positive numbers alone.  Is it highly clustered?  Are there cases that caused fatal injuries far more or less rapidly than the mean?

11
DF Dwarf Mode Discussion / Re: FPS Science and observations with 1000 dorfs population
« on: March 13, 2023, 08:46:58 pm »
The expression max(dx, dy) + dz still defines an octahedron.  For each dwarf on level N, only a single tile would be <= 26 away on level N+26.  So shouldn't that have worked?  Or is the early-out defined a little differently, here?

My pitiable fort is about to get retired at 11-13fps despite 140 levels of vertical sprawl, so this question is especially important to me right now.

12
DF Dwarf Mode Discussion / Re: FPS Science and observations with 1000 dorfs population
« on: March 13, 2023, 08:10:34 pm »
Is there a typo?  Reading these two statements:

Quote
2. The most performance efficient embark for my testing conditions is 16x1 (and probably 1x16, but I haven't tested that one). It exploits the 26 tile skipping mechanism the most while still having relatively small amount of land to calculate/path around. With 1000 dwarves and their migrated animals on 16x1 embark i get around ~50 fps

Quote
3. 26 tile skipping doesn't work on z-dimension. On my recent test day I made 4 areas as in point I.5. separated by 26 z-levels, and spread 1000 dorfs on those areas resulting in roughly 250 dorfs on each of them, then removed stairs connecting them. FPS results are absolutely the same as with single 16x1 z-level.

If 16x1 checkerboarded was the best-performing configuration, and separating 4x occupied levels by 26 floors performed the same as just one level of a 16x1 embark then why do you say that the distance check doesn't work?  Doesn't that imply that the distance check is also sensitive to the vertical dimension?

In at least one other context (work scheduling) Putnam reported that the distance used for scheduling weight was max(dx, dy) + dz.  That's a little weird, but it would account for the vertical.

13
DF Dwarf Mode Discussion / Re: Armor Material Science: how do materials compare against goblin weapons? 50.07
« on: March 13, 2023, 07:39:04 pm »
There's something suspicious about this data.  I'm picking on one row in particular for exposition, but this anomaly applies to all of your reports.

Code: [Select]
| weapon          | mean | stdev | lower_bound | upper_bound |
|-----------------|------|-------|-------------|-------------|
| iron mace       | 62.1 | 79.7  | 46.1        | 78.0        |

Upper and lower bounds from a 95% confidence interval, typically defined as mean +/- 2*standard deviations.  The reported mean is centered in-between the bounds (avg(lower, upper) == mean).  But the separation on the bounds implies a one-sigma standard deviation of about 8 ((upper - lower) / 4).  Where does 79.8 come from?

Pages: [1]