Better gate creation sequence for IC assembler #115
Conversation
|
Sorting every time it picks a new target is indeed suboptimal. I was thinking about implementing a proper algorithm there but decided to skip it as I wanted to get a first release out at that time. In general, a spiral doesn't look all that bad, but there is the problem that turning takes longer than etching which may result in your circuits being assembled slower with more lasers for certain designs. Calculate the path of every laser before it starts running and cache it, if you want to improve this. Have a look at this algorithm: http://en.wikipedia.org/wiki/Nearest_neighbour_algorithm Sidenote: Bonus points for moving it to a new thread, doesn't have to cause tick lags 😉 |
|
Do you mean that the laser has limited angular speed and not something else? As a side note, this "nearest neighbour algorithm" is more or less what I have implemented. |
Example implementation for #114
I guess it is highly suboptimal (sorts all the gates when it chooses the next one to create), but computations still take miserable amount of time. If you start too many (like thousands) assemblers simultaneously, it might lag quite a bit.
Might also have placed some code in the wrong file.
Feel free to reject it.