AND gate design challenge

[rtm223]

The biggest thermo drain on normal piston OR/AND gates is the piston timing. As you make a piston timing shorter, you can actually see the thermo increasing below a certain time. It's actually fairly significant at 0.1 seconds.

I did not know this, thanks.

However, this might not be caused by the piston, but rather the predicted movement of the object the piston is attached to. If this is the case it is possible that this gate uses less thermo. However, it is also possible, that having multiple pistons and chains on one object makes it more complex for the engine and therefore increases the thermo.

2D vector addition of the multiple forces (or speeds) is a really simple calculation. Plus you would need to add these up anyway to do the pistons in series... Collision detection on multiple objects is far more complex. however:

The only way to know for sure is to experiment

Agreed. Are you volunteering?

[dawesbr]

Wow triple post

rtm, I was kinda actually referring to my design for NO moving parts at all, in case you didn't notice.

trickster, year 10 DT here as well 741 op-amps woo!

[dcf]

I did not know this, thanks.

you're welcome.

2D vector addition of the multiple forces (or speeds) is a really simple calculation. Plus you would need to add these up anyway to do the pistons in series... Collision detection on multiple objects is far more complex. however:

True, although we don't know the algorithm LBP uses to calculate thermo, which is the real issue, not the actual computation time. Also, for nested ANDs/ORs I have not placed pistons in series. I use a simple, two input AND/OR and link multiple together, rather than making a single device. The advantage to this is that there is no weird piston problems from placing pistons in series. Each object is directly hooked to dark matter. This is obviously less efficient because, for example, to have 4 inputs, one has 6 moving parts (2n-2). However, in LBP it might be more thermo friendly because of the assumptions that are made when something is directly linked to dark matter.

I haven't really considered any of these things with my first tool set. I came at the problem from the 'programmer time more valuable than computation time' mindset, which may be wrong for LBP engineering.

Agreed. Are you volunteering?

Yeah, I'll give it a shot. As someone who puts ~100-200 logic gates in a level this is a big deal for me. If this is much more thermo friendly it has huge implications.

[tameturtle]

Tested a configuration with six winches against one piston, all set to 0.1s. Confirmed working. Great!

[dawesbr]

Still can't think of an n-input XOR. I was thinking maybe something to do with a rotating circle with an angled switch, surrounded by emitted keys. One emitter off, circle spots it and stops. But this would work as an OR. Perhaps by linking my previous idea to this one, or...

[rtm223]

thanks dcf, I'm looking to see how that turns out - would be funny if it's horrendously inefficient lol.

dawesbr, as an xor with multiple inputs will be true is any odd number of inputs is true, it's gonna need some really funky contraption, which probably would be less easy than just stacking lots of 2 input ones. I don't intend to try and solve that one, because I don't have any use for it yet!

[dcf]

XOR might actually be pretty straight forward.

Traditional two input XOR is AND(OR,NOT(AND)). Nesting them together means one and only one.

In order to do this we need to use the multi-input OR. Instead of using a multiple input AND we need one that activates if any two are active. This might be doable by setting the winch strength on the multiple input AND to 5. This way when only one winch is triggered it can't overcome the piston. If two or more are triggered then the combined strength might overcome the piston. If this works you only need to put these two together with a traditional two input AND to complete a very simple XOR switch. It's not quite as simple as these multiple input ANDs/ORs but XORs rarely are.

[Pinchanzee]

Probably for most levels a disolve based one would work, but the design I have in mind is fully reversible.

How the diagram works:
[SPOILER]
Green mag key is inverted so if the moving bar is pulled in, output is false; bar pushed out, output is true.

The weak piston is always pushing out (it's switch is inverted direction).

If all of the chains are set to out, then none of them affect the moving bar and the piston pushes it away.

Otherwise, the chain(s) that are set to "in" will pull in the bar setting output false. A single chain is enough because it's stronger than the piston.

You could always replace the piston with gravity, but I prefer having the movement a bit more controlled.
[/SPOILER]

This sounds like all chains have to be activated at once though?
What if you wanted a reversable gate, that the effect was permanent, but reversable if one of the AND inputs were changed??

[rtm223]

Pinchanzee, a single chain set to input is enough to pull the moving part back in. It's a fully working combinatorial AND gate.

[quote=dcf]Traditional two input XOR is AND(OR,NOT(AND)). Nesting them together means one and only one.[quote]
This is wrong. (a XOR b XOR c XOR d) will be true if 1 or 3 of the inputs are true.

[dcf]

[quote=rtm223;184919]Pinchanzee, a single chain set to input is enough to pull the moving part back in. It's a fully working combinatorial AND gate.

[quote=dcf]Traditional two input XOR is AND(OR,NOT(AND)). Nesting them together means one and only one.

This is wrong. (a XOR b XOR c XOR d) will be true if 1 or 3 of the inputs are true.

Bah, you're right

[RangerZero]

I'd be curious about you memory test results.

While I understand the piston based idea can be more "complex" engine wise, I know that emitters simply take alot more memory than connectors though. There's definetely a interesting test to be made in this case because it can go either way really. I would even bet the pistons wins memory wise since emitters are taking so **** much in my exp.

.

[rtm223]

I was thinking the memory test would be for traditional AND gates vs the multiple chains one. I'm not really convinced the emmiters AND gate is that great, simply because you can't negate the inputs internally to the switch. Plus it's so noisey! I'd use it for timing requirements only I think.

[dawesbr]

It has no moving parts though - that was the challenge xD

[dcf]

Test Procedure: Gates built and tested in blank level. Captured as object and placed into new blank level once. Object copied 100 times with dark matter aligned to big grid. Thermo rounded to nearest 1/2 bar and recorded.

Additional notes: Only dark matter and white wood used for all tests. Starting gate left in level. No other objects in level.

AND Gates 2 Inputs
Parallel Config: 6 bars
Normal Config: 4.5 bars

AND Gates 3 Inputs
Parallel Config: 7 bars
Normal Config: 9.5 bars

AND Gates 4 Inputs
Parallel Config: 8 bars
Normal Config: 14.5

**************************************

The results speak for themselves. If you are using even 3 inputs there is an enormous thermo advantage to using the parallel gate. In addition, the parallel gate should have a faster response time compared to the series configuration.

***************************

I have also run an additional test. I took a captured 4 input AND gate, placed it in a level and deleted one input. I then copied it so there were 100 of them and measured the thermo. It appears identical to the thermo for the 3 input gate. Therefore, it does not appear necessary to clutter your profile with parallel gates for each input. Just make one for the highest number of inputs you will use and capture it. If you need less inputs just delete the chains.

[rtm223]

Nice work dcf, I was hoping this would be one of those things where the more inputs you have the better the payoff.

[snowflakecat]

good stuff! great challenge! It's a shame I haven't picked up a new PS3 since moving (left mine to a friend). I found that winches have many interesting uses since they don't exert force when extending.

[dcf]

Oh, if anyone wants the 4 input AND/OR gates let me know and I'll send them to you. You must be careful with the lengths (2.5 instead of zero, etc) or else the thing oscillates.

[Pinchanzee]

Hmm.. I'm still not really understanding how this idea works..
Has anybody made a level?
Or would somebody be able to make a tutorial level?
I'm sure many people would benefit from it, not just the members here

[dawesbr]

Think about it like this. Each chain is worth, say an arbitrary unit of 2. The piston is worth 7. Only when all the chains are on, and their total worth is 8, can they overcome the 7 that the piston has. If only 3 are on, they amount to 6, so the piston overpowers them.

Also, the piston can act as an override

[Pinchanzee]

Ahh.. I see.. great idea!
I'll be sure to try it out
Thanks