Bay 12 Games Forum

My method is to keep all males caged and all females tethered until I reach a stable female population of 5 or 6.  Then, each time winter comes around, I kill half of the males and any caged mature females.  That leaves me with a steady-state population of 4 or 5 males, a half-dozen females for production, and a few immature females waiting in the cage just in case.  I typically alternate going down the list "You live / you die / you live / you die" so that I retain the same age profile across the population.

It has to do with pathfinding.  When you carve a big swath of ramps (in order to hollow out vertical space safely), you turn lots and lots of squares into viable paths between Z-levels.  Then, as you get a bunch of them carved, individual ramps collapse or become non-viable.  You're creating and destroying lots of pathfinding options, so of course it's going to chug to a halt for a bit.

I always thought that pushing the fluid flows, temperature effects, and weather off onto extra cores would be a fairly easy way to improve performance.  Since they're the functions that low-end computer users turn off first anyway, the multi-threading would have the same effect: DF would use Core 1 at near 100% (like it does right now) but would run with the performance of DF-minus-aero-thermo-fluids.  DF-Weather, DF-Liquids, and DF-Thermo would run on additional cores, or all together on a second core, or whatever.

As a design matter, it seems simple, but I'm not a coder and I'm not Toady, and I haven't seen the source.  Perhaps Toady would consider handing off that code to someone like he did with the OpenGL stuff, and letting them write a standalone chunk of code that runs in parallel with DF.  Then when DF launches it could launch the others as their own processes, passing data back and forth.

I expect a legion of coders to all jump down my throat and tell me what a terrible idea this is.  Also: if your reply is of the format "Toady cancels progress: interrupted by multithreading" or "Toady has said he ain't doing it," peace - I don't want to start a fire.  Just thinking out loud.  In silent ASCII.  Something.

Yes, I fully expect to see "MORE MAGMA / LESS KITTENS" as a response.  Seriously, though, the project -- if realized -- would probably appeal more to gadget-happy early adopters than to housewives, so I'm trying to get a cross-section.  I've got feelers out in other places, too; I doubt that the DF community will skew my results too horribly.

Thanks for all the quick responses, everyone!  I've already got a bunch of good information.

kutulu, are you suggesting compressing the Mayday pack to PNG will result in a performance boost?  Will that work with earlier 40dX versions?  I'm running 40d9 still, and considering upgrading to 40d11 now that it appears stable.  I'm also looking to move its processor affinity to an unused core and try a few other performance boosting tweaks.

Penguin avatars are clearly the reigning tr
ALL GLORY TO THE HYPNOTOAD

Prepared food takes up to four stacks of ingredients, gives each one (Ingredient Value * Ingredient Prep Quality), then adds (10*Meal Prep Quality), and sums the value of all four ingredients before multiplying by the combined stack sizes of all four ingredients.  A single piece of dragon meat, masterfully sliced (50 * 2 * 12 = 1200) plus 25 no-quality quarry bush leaves (5 dorfbucks each), prepared at no quality mod (plus 10), gives a stack of 26 with each item worth 1215 for a grand total of 31,590.  Because one ingredient is masterfully-prepared, the meal stack gets marked with the masterwork tag, too!

Now imagine you're at a wedding reception and someone brings around little hors d'ouevres of perfectly-sliced, perfectly-cooked filet of Kobe beef on top of sloppily hand-ripped leaves of generic iceberg lettuce.  Is that really a $500 bite of food?  I think not.

Doesn't stop me from running a Dwarven Drive-Thru, though.