Print

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

[Retro]

I think if you alternate floor tiles, you can make all the pumps at once without waiting for each one to be built.

For a while built the entire pump stack at once, since you can designate pumps to be built on top of other pump designations, but if one of them gets cancelled mid-construction the entire stack collapses, even if some of them had already been built. And then you can't redesignate them immediately because all the parts lying around have been dibs'd by dwarves looking for stockpile hauling work. So now I just wait it out and do the whole stack one layer at a time.

[Gus Smedstad]

I think if you alternate floor tiles, you can make all the pumps at once without waiting for each one to be built.

The standard, compact pump stack works by stacking pumps on the same two squares on all z-levels, leaving open spaces below pumps, for power transmission from the pump below.  This makes the pump "hanging," like a waterwheel or a vertical axle.  It's not legal to start a hanging machinery item unless there's something at least planned below, and if it completes before the supporting item, it immediately deconstructs.  I've had this happen with conventional pump stacks, placed 10 at a time and had half of them fall apart because they were completed in the wrong order.

This happens even if only one of the two tiles below the pump is an open space (which is the way I always did it).

My alternate stack looked like this:

Code: [Select]

level z

##.###
#0>>.#
##*#.#
######

level z+1
####.#
#.#*.#
#.<<0#
######

>, < = pump direction
* = gear over open space to pump below
O = open space to level below
The gears must be built after the pump below is complete, but that's the only dependency, the pumps can be built in any order.  The drawbacks are it requires an extra gear for each level, and takes a 4 x 4 block for the pumps instead of a 4 x 2 block.  The time required to make the mechanisms and place them wasn't too large, but the extra 5 power is kind of significant.

 - Gus

[numerobis]

That extra spot of outflow area probably imposes a lot of the computational cost in your setup.  With only a single-tile output, I predict it would be much less painful.  That's from first principles, not from experiments.  A myth has been proposed; pray bust it for me.

[blastrogath]

What might make more sense if you want magma (and this is what I did) is to spend effort early on to find a volcano.

I got magma up on z level 245, savana down at 149, and the magma sea starts at about -15.

My fort is on a small rise isolated from the moutain the volcano is on, I'd need to get several z levels of magma going before it hit the base of my walls.

[Gus Smedstad]

I have a hard time seeing it as being purely the extra 140 or so tiles that might have liquid in them.  I'd drained enough water out of the cavern lake that there were at least 100+ tiles in there with less than 7/7 water, and while that probably contributed to the 6 fps while the pump stack was off, the frame rate dropped by a factor of 6 or more, not a factor of 2, as you might expect if I'd (at worst) doubled the amount of fluid computation going on.

Further, the frame rate climbed back to pretty much "normal" for that point when the stack was off, even though a good number of those tiles (probably 60 or so) now had partial (3/7 or 4/7) magma in them.

 - Gus

[ManaUser]

That extra spot of outflow area probably imposes a lot of the computational cost in your setup.  With only a single-tile output, I predict it would be much less painful.  That's from first principles, not from experiments.  A myth has been proposed; pray bust it for me.

I was thinking the same thing. A standard pump stack, with floors (as pictured here) might well be easier on the CPU.

[numerobis]

Fluid teleportation appears to be expensive.  The 1-tile outflow means one unit teleports every cycle (a distance of 140).  The two-square outflow means 140 units teleport every turn (a distance of two each).  Hence my prediction that there's potential for a big difference.  Not certainty, by any measure.

I agree that fluid diffusion is not likely the problem.

ManaUser: Gus clearly knows pump stacks, having chosen this design to avoid them.  There is s nice design to make a helical pump system with 1-tile outputs.  It would take just an extra couple hundred or so power for the axles and gears (the power stack can't be vertical in this design).

[steveman0]

That extra spot of outflow area probably imposes a lot of the computational cost in your setup.  With only a single-tile output, I predict it would be much less painful.  That's from first principles, not from experiments.  A myth has been proposed; pray bust it for me.

I was thinking the same thing. A standard pump stack, with floors (as pictured here) might well be easier on the CPU.

I use this standard pump stack design for my 140z magma stack and I notice little to no difference when I turn it on.  Having only 1 tile might well do the trick since the pumps work faster that the magma can reasonably propagate there is little fluid action taking place.

[Gus Smedstad]

I use this standard pump stack design for my 140z magma stack and I notice little to no difference when I turn it on.  Having only 1 tile might well do the trick since the pumps work faster that the magma can reasonably propagate there is little fluid action taking place.

That's a good point.  Pumps have a very high capacity for transport; in most plumbing I've done, i.e. waterfalls and power plants, the input tile for the pump typically remains completely dry.  My alternate power plant design showed that a single pump can keep up with at least 4 one-tile wide channels of water.

So it may very well be that a 1-tile output pump stack has a near-zero computational cost compared to 2-tile approach I took, since there's never any fluid in most of the output tiles.

I hadn't considered a helical design.  That requires two gears per level, one above each pump, and one to transfer the power to the next pump, for a total of 20 power per level.  That seems worthwhile if it avoids both the build-order dependency and the computational hit I took.  I've started a new fort, it remains to be seen where the magma will be on this map.

 - Gus

[derekiv]

The answer is a volcano. My last fort had dwarves climbing stairs to go to the forges.

[smigenboger]

So are you going to make it a 'spiral' square pattern? the pumps could all be separate, or joined by gears, if gears transfer power

[Corona688]

To avoid the fixed bottom-to-top or top-to-bottom build-order of a vertical pump stack, build grounded gears beside them first.  They'll hold the pump up.  Once the pumps are all built you can remove them.

[Reese]

To avoid the fixed bottom-to-top or top-to-bottom build-order of a vertical pump stack, build grounded gears beside them first.  They'll hold the pump up.  Once the pumps are all built you can remove them.

better, of you have enough free wood, is to built grounded horizontal axles, because those will deconstruct into wood instead of a bunch of gears you may or may not have any use for.

[Gus Smedstad]

I didn't think of that.  Planting "scaffolding" machinery next to the pumps seems like a fairly good solution.  Aesthetically, I think I prefer the spiral, but not enough to pay the 100% premium in power requirements.

 - Gus

[Corona688]

To avoid the fixed bottom-to-top or top-to-bottom build-order of a vertical pump stack, build grounded gears beside them first.  They'll hold the pump up.  Once the pumps are all built you can remove them.

better, of you have enough free wood, is to built grounded horizontal axles, because those will deconstruct into wood instead of a bunch of gears you may or may not have any use for.

I suppose.  I'm a big enough mechanical nut I probably could use them all eventually so that's what leaped to mind.