Print

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

[Pirate Bob]

I did some large-scale testing of melee weapons, because I wanted to know what would be the most effective thing to arm my dwarves with.  There are some results posted on the wiki, but I'm not sure how old they are, and Toady fixed several bugs relevant to melee combat so even 34.11 results might not be accurate any more.  Short answer - all weapons besides maces are pretty similar - maces are crap (I think that's pretty well known).

In this test I wanted to see what would be most effective against invaders.  I made a custom arena with an array of 1x3 cells across the whole map.  I then placed goblins in each of the cells, like so

Code: [Select]

-----
|12+|
-----where 1 and 2 are the goblins on sides 1 and 2.  For all tests, goblins were armored with iron helm, iron chain mail shirt, leather leggings, and leather shoes and iron shields.  In other words, fairly typical for what invaders wear - perhaps on the high end.  I then added 1152 goblins for each side, armed as described, using macros which I will post shortly.  All goblins were competent in fighter, axeman, swordsman, spearman, hammerman, maceman, dodger, armor user, and shield user. 

Edit: I forgot to mention that I used a script to count the first 1000 kills in each test, and tabulate how many were for each side.  I stopped at 1000, as if you wait long enough almost all the test subjects on both sides will die from bleeding and infection, so you have to stop somewhere.  I then ran most of the tests twice, switching which weapons was side 1 and side 2, in case there was any bias between sides (it appears that side 2 gets to strike first, and has a slight advantage), and added the scores together.

The results are as follows:

For steel weapons:

axe	836	sword	1164
axe	789	spear	1211
axe	1305	mace	695
axe	943	hammer	1057
sword	944	spear	1056
sword	1422	mace	578
sword	1069	hammer	931
spear	1368	mace	632
spear	1137	hammer	863
mace	728	hammer	1272

In older versions, heavier metals made better blunt weapons.  However, Toady fixed some errors which I think were largely responsible for this.  Any difference between silver and steel blunt weapons is small enough that it is not statistically significant.

silver mace	1019	steel mace	981
silver hammer	991	steel hammer	1009

However, adamantine is light enough that it is significantly worse for blunt weapons, and of course it is a bit better for edged weapons.  I suspect that there is something like a cutoff for the maximum speed a weapon can travel at (this is what happens for projectiles) such that very light things like adamantine (and possibly wood) end up with lower momentum, but all standard metals are the same.  This is purely speculation though.

adamantine axe	534	steel axe	466
adamantine sword	536	steel sword	464
adamantine spear	509	steel spear	491
adamantine mace	214	steel mace	786
adamantine hammer	159	steel hammer	841

In summary, spears appear to be the best melee weapons, followed by swords, warhammers, axes, and finally maces.  The differences between the good weapons are quite small, and may not even be significant in all cases.  These results are for goblins armed as "typical" invaders, and it is possible different weapons could be better against other opponents.

My next step will be to test against larger creatures, such as elephants, dragons, and (if I can figure out how to get the raws) maybe clowns.  Given that spears came out on top against goblins, and they have the highest penetration depth, I expect that they will also prove the most effective against larger creatures, meaning they may be the best overall choice for your military.  However I cannot stress enough that the difference between all weapons (except maces) are quite small, so you should not take these results to indicate that other weapons are bad.

For comparison, I ran anther test where the skills of one set of goblins were Great and the other unskilled, and both sides had silver hammers.  The result was 999 to 1.  Skills far outweigh the small difference between weapons, as the standard weapons are actually balanced quite well.

[Pirate Bob]

Here is a link to a simplified version of the scripts used for testing.  This will allow you to easily duplicate 1152 copies of any combatants you desire, and parse the resulting logs.  For the actual tests I used a larger script which fully automated testing all possible weapon combinations.  I would be happy to provide this if anyone wants it, but it is much harder to modify this to test other things.

[drone1230]

This is definitely great !!science!!, but I'm not sure how to read it. What exactly do the numbers represent?

[Pirate Bob]

This is definitely great !!science!!, but I'm not sure how to read it. What exactly do the numbers represent?

Sorry - I left out the paragraph explaining what the numbers mean in the original post.  I have updated it.  The numbers are how many goblins of the opposition were killed by each side.  To put it differently, out of 2000 1v1 battles, its how many were won by goblins with the indicated weapon.

[drone1230]

Thanks, I looked over it again and understand now. By the way, what kind of leather was the armor made out of? I'm not sure, but I think that different kinds of leather are, well, different. But even if it is, the difference is probably negligible.

[Pirate Bob]

Thanks, I looked over it again and understand now. By the way, what kind of leather was the armor made out of? I'm not sure, but I think that different kinds of leather are, well, different. But even if it is, the difference is probably negligible.

The shoes were made of dwarf leather (it was at the top of the list) and the leggings out of random werecreature leather (it was at the bottom of the list, and leggings can be made of metal so I didn't want to scroll down through all of those).  Dwarves definitely just use the default skin template, and hence will have default leather.  I'm not sure about werecreatures, but I think they do also.    Based on looking at the raws, I am not sure how any creature would produce a different leather - they would have to have a custom tissue layer for skin.  Things like dragons have scales instead of skin, but the scales cannot be tanned into leather.  I'm not saying it doesn't happen - I'm just saying a quick look through the raws suggests most creatures produce identical leather (same yields, fractures, strains and density).  Then again it's also quite possible I don't understand the raws, so if you know something please educate me...

[Art]

This is great work Pirate Bob!

I too noticed the badness of maces in my smaller, less scientific, tests. I have been working on improving them in the thread bellow:
http://www.bay12forums.com/smf/index.php?topic=143680.0

It is kind of a big revelation that silver and steel are so close now, at least against iron armor. I have seen some goblins with steel armor in v.40 though. Stolen from dwarves possibly? Can they do that?

I am also surprised by the lower performance of axes vs swords. Are these long swords or short swords?

Not surprised by spears.

I may be reading into this too much, but it appears that stab attacks are better than slashing.

[Pirate Bob]

This is great work Pirate Bob!

I too noticed the badness of maces in my smaller, less scientific, tests. I have been working on improving them in the thread bellow:
http://www.bay12forums.com/smf/index.php?topic=143680.0

It is kind of a big revelation that silver and steel are so close now, at least against iron armor. I have seen some goblins with steel armor in v.40 though. Stolen from dwarves possibly? Can they do that?

I am also surprised by the lower performance of axes vs swords. Are these long swords or short swords?

Not surprised by spears.

I may be reading into this too much, but it appears that stab attacks are better than slashing.

I would say the take home message for warhammers is that you should make them out of whatever metal is most available to you (probably copper or silver) and that their performance is on par with all the steel edged weapons.  Therefore they are a great choice for your military if you don't have steel (or if you expect undead, of course).

I was not aware that goblins could have steel armor.  I have never seen it, but I tend to spend more time modding/testing than actually playing.  I would expect that against steel armor a warhammer (of any standard metal, since they are all about the same) would perform better than any of the edged weapons.  Steel armor should be able to stop (or convert to blunt in the case of chain mail) most edged attacks from steel weapons, but since warhammers hit hard and have a small contact area I don't think they are stopped.

Have you ever seen invaders with armor other than a mail shirt and helm?  Do they get breastplates, greaves, etc?  In addition to not playing a ton of fort mode, I also have a crappy computer, and tend to cap my population around 50, which may prevent me from seeing higher end invaders?

The swords used were short swords (I wanted to test the weapons you can make in fort mode, at least at first).  I believe the reason axes perform worse is they have 3 (useless) blunt attacks and only one edged attack, while swords have two edged attacks (slash and stab) and two blunt attacks.  Spears also have one edged and only one blunt attack.  I think this makes it more likely that swords and spears will use one of the edged attacks, but I am not certain.  When I get a chance I will parse the logs to confirm this.  If this is true, then you could make axes more equal to swords by duplicating the hack attack, so that it will be used more often. 

This could be totally off.  I know that the AI favors using edged attacks over blunt, so maybe the extra blunt attacks don't matter, but I suspect that they do.

[Art]

I have not seen mail shirts in my latest playthroughs. Only helms and breastplates.

Edit- to be clear, that means no gauntlets, greaves, leggings or boots either.

[Pirate Bob]

I have not seen mail shirts in my latest playthroughs. Only helms and breastplates.

Edit- to be clear, that means no gauntlets, greaves, leggings or boots either.

Huh - that's odd.  In my 40.11 fort all the invaders have mail shirts, and none breastplates.  Maybe it is based on preferences of the enemy civ or something?  I am also dealing with a goblin civ that has so far sent only elves as invaders, so they may not be the most normal...

I feel like mail shirts were more common that breastplates in my 34.11 games also, but I didn't keep close track.  Good to know that they can have breastplates - I will try some tests with that configuration also.  I don't expect it will make much difference, except for maces, as I think steel edged weapons will pretty much bypass all non-steel armor, but I'm not sure.  Maces would likely be deflected by a breastplate, especially an iron one.

[Pirate Bob]

I did a few trials with megabeasts, and no surprises, spears did significantly better than warhammers (as spears have very high penetration, and warhammers low).

The table below shows the results of 1v1 battles between dwarves in full steel armor with Grand Master skills, wood shields, and steel weapons against Dragons and Hydras

steel spear	945	Dragon	55
steel hammer	809	Dragon	191
steel spear	417	Hydra	583
steel hammer	7	Hydra	993

As you can see, both spears and hammers fared well against dragons, with spear suffering almost 4 times less losses.  Speardwarves were nearly even with hydras, but hammerdwarves were slaughtered.  I haven't had time to test yet, but based on penetration depth I would expect axes and swords to perform better than hammers, but worse than spears against megabeasts.

Edit - I had an another idea of why axes may fare worse than swords:  the wiki says that about 20% of dwarves will need to wield axes two-handed due to size, while only 1/49 will need to two-hand a short sword.  Goblins are the same size as dwarves, so I think the goblins in my tests would have the same problem.  The wiki also says that if two hand are not available, there is a penalty to combat, and I gave the test subjects shields, so they should have suffered this penalty.  I guess if I get a chance I should rerun either without shields or using larger creatures, like humans, to confirm this. 

If it turns out to be true, then maybe you want to arm your smaller dwarves with short swords or war hammers, and the larger ones with spears (or axes if you like them and don't care about them not being quite as good in combat, as the differences are small).  Does anyone know an easy way to tell the size of fortress dwarves?  I assume the height and broadness of dwaves is contained in their description, but I don't know exactly what I'm looking for.

[Kryxx]

I wonder how many of those dwarves had a 0 str or agility.

I had a migrant with a 0 agility. I think only decent at like 2 roles with a 50% or higher.

Looking through the raw's.   Standard human liver size 200.  Dwarves have a 300 

the raw is creature_standard.txt


There's a spot in there where it tells the calculated sizes of dwarves.  Everything is variable within set parameters.

[BODY_APPEARANCE_MODIFIER:HEIGHT:75:95:98:100:102:105:125]

Each of the 7 has an equal change of happening.  So it's actually a 14.2857% chance of being short.

[StagnantSoul]

This may be greatly useful against those two megabeasts, have you tried running it against rocs? And then the semi-megabeasts?

[Pirate Bob]

This may be greatly useful against those two megabeasts, have you tried running it against rocs? And then the semi-megabeasts?

I plan to try against bronze colossuses and rocs, and would be happy to try anything else that people think is interesting.  If someone knows how to get the raws for clowns that would be extremely interesting.  I think I will modify my automatic scripts to cycle through all combinations of megabeast and dwarf weapon, although I may not get to that for a few days.

With regard to Kryxx's comments, I am aware of where the relevant modifiers are in the raws.  I'm not sure if this is what he meant, but I definitely will try running again with all the size variations for dwarves (or goblins if that's what I use) removed, to see if this has an impact on the behavior of the weapons that sometimes require two hands.  I might also try removing the attribute (str, agility, etc) variations as well.  I have some "arena dwarves" that already have all this stuff removed, but I think they also don't feel pain and possibly have some other modifications which would be undesirable, so I may start from scratch.

I still would like to know if there is a simple way to determine a fortress dwarf's size from their description page (or perhaps from dwarf therapist), as if I can confirm that part of the issue with axes is some dwarves are too small to use them properly then maybe I would assign said dwarves other weapons (or hauling duty).  Spears have the same size requirements as axes, so dwarves of the correct size might be even more deadly with spears.

[Chevaleresse]

I believe that you can get at the raws for procedural creatures in an uncompressed/decompiled save.