Print

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

[Rafal99]

So, that leads to a question:  Toady, just how random is the RNG in DF?  How robust are the seeds?  Have you done, are you planning to do, would you consider outside auditing of the quality of the random numbers the RNG used in game produces?

I'm guessing it's a pretty good Pseudo-RNG. In terrain generation it's very easy to notice periods that are too small.

AFAIK DF uses Mersenne Twister, which is a very good random number generator.
It could be found in Battle Champs source code.

If there are any quirks in the terrain generation they probably come from the imperfection of terrain generation algorithm rather than from short periods of the RNG.

[Siquo]

If there are any quirks in the terrain generation they probably come from the imperfection of terrain generation algorithm rather than from short periods of the RNG.

That's possible, it's a personal experience

[dreiche2]

Circling back just ever so slightly to the RNG questions (while tapdancing around the drop rate statistical analysis dead horse), the kind of drop wierdness displayed in MMO's, and other types of simulations in general, has a great deal to do with ty KIND of RNG being used. [...]

I'd really like to see some evidence that this is an actual problem in practice.

[HiEv]

Circling back just ever so slightly to the RNG questions (while tapdancing around the drop rate statistical analysis dead horse), the kind of drop wierdness displayed in MMO's, and other types of simulations in general, has a great deal to do with ty KIND of RNG being used. [...]

I'd really like to see some evidence that this is an actual problem in practice.

The one MMO that I play has repeatedly had myths about its RNG coming up, primarily around its refining mechanic.  However, every single statistical analysis that's been applied using sufficiently large numbers shows that it always has the same odds regardless of time of day or location in the game where it's done.

In my experience, people see patterns in random noise and assume it's not random because they think that random noise never has patterns in it.  Honestly, if results from a RNG never had patterns in it, that would be proof that it's not random.

[BradUffner]

The one MMO that I play has repeatedly had myths about its RNG coming up, primarily around its refining mechanic.  However, every single statistical analysis that's been applied using sufficiently large numbers shows that it always has the same odds regardless of time of day or location in the game where it's done.

In my experience, people see patterns in random noise and assume it's not random because they think that random noise never has patterns in it.  Honestly, if results from a RNG never had patterns in it, that would be proof that it's not random.

I remember a story about WWII cryptography where the army was generating random numbers for one time pads by using a group of church ladies pulling bingo balls with the number written on them.  The cypher was eventually cracked because the ladies thought they were making the numbers "more random" by picking a new number if the one they drew was the same as the previous one.  It reduced the entropy enough that the Germans were able to crack the code.

[Arkenstone]

So... Um, derail thread time?

[Mr S]

BINGO!!

[PTTG??]

YAHTZEE!!

[Captain Crazy]

stacking isnt a bug but an unimplemented feature
you can follow bugs status on the bugtracker, i think it's updated in realtime

elaboration:a whole stack (bones, teeth, etc) makes only one craft (figure, crown, bracelet, etc)

must be one weird bracelet made out of 1000 giant whale bones

No, it doesn't.  I've watched my bonecrafters make crafts.  One bone makes one craft.

Decorating with bones uses an entire stack, and artifacts demand entire stacks of bones, so that's kind of bugged there, but making discrete objects like bolts and crafts uses one bone at a time.

The "whole stack of bones used for one craft" bug was fixed in either 34.07 or 34.06, I forget which exactly, but that's not exactly the important point. 

Anyway, that's not what people mean by "stacking" usually, so it was likely to confuse people.  Stacking is when you can take, say, 50 individual plump helmets and put them all into the same tile to make plump helmet [50].  Currently, you can split stacks apart, but you cannot stack them back together. 

Combined with the fact that vectors lag the game when an overly large number of items are stored in them, (and a typical fortress can have ten thousand items easily, and that's before you count contaminants like evil cloud stuff,) and stacking is a way to keep FPS up while having more small item types, or have dwarves carry around more small items.

well, uh

* butcher animal in adventure mode
* take bones, make flute
* 13 bones for a single flute

I think it's impossible to seperate a stack of certain things (bones) in adventure mode (throwing it throws the entire stack, dropping it/picking it up gives now "how much?" prompt, etc)

the flute is either silly-straw long, or my guy is just terrible with bones (or it's a bug)

[irdsm]

Toady just confirmed cart dumping and you can do it without stopping the cart. This could make it ridiculously efficient. Might even be able to make a magma dump trap.

[Arkenstone]

When the contents are auto-dumped, does that include occupants?

And while I certainly like the idea of magma carts, I feel like they're somewhat redundant given that we already have pump stacks.  I'm imagining much more use for dumping stuff into magma.

[Chronas]

And while I certainly like the idea of magma carts, I feel like they're somewhat redundant given that we already have pump stacks.  I'm imagining much more use for dumping stuff into magma.

Don't forget this moves magma faster laterally and requires less planning and materials to set up.
Energy efficiency will be through the roof too, only 1-2 rollers would be necessary on a small magma fall circuit.

[Himmelblau]

And while I certainly like the idea of magma carts, I feel like they're somewhat redundant given that we already have pump stacks.  I'm imagining much more use for dumping stuff into magma.

Don't forget this moves magma faster laterally and requires less planning and materials to set up.
Energy efficiency will be through the roof too, only 1-2 rollers would be necessary on a small magma fall circuit.

I'm not sure what a magma fall circuit means (a kind of a trap?), but I don't see magma carting as an efficient alternative to pump stacks; if every z-level requires a roller and a mechanism, the energy cost is roughly twice that of a pump stack. Unless rollers can be somehow designed to transmit power vertically, or if they spend less than 10 uristwatts, that is. You're right about material efficiency, though.

I also don't know about "moving magma faster laterally". Pumps are pretty fast. Magma carts could be safer, though, and more versatile as they can move through non-empty corridors without leaving 1/7 magma everywhere.

[Osmosis Jones]

I also don't know about "moving magma faster laterally". Pumps are pretty fast. Magma carts could be safer, though, and more versatile as they can move through non-empty corridors without leaving 1/7 magma everywhere.

Yah, carts are definitely slower than pumps laterally; as long as you're pumping into a confined tunnel, magma will be teleported to the first open tile. This means that in an empty tunnel, magma travels a number of tiles equal to the number of pump cycles, while in a full tunnel, magma can travel an arbitrary distance in one tick (may be some IRL lag as the computer paths to the free tile though).

[Sphalerite]

* butcher animal in adventure mode
* take bones, make flute
* 13 bones for a single flute

This is impossible in the vanilla game, so you're talking about a modded-in reaction.  Modded-in reactions don't respect stack sizes, this is true.  Making bone crafts in fortress mode is completely different, doesn't use a reaction, and does work properly with only taking a single bone off a stack.