#EXAPUNKS SOLUTUIONS CODE#
#EXAPUNKS SOLUTUIONS SERIES#

Otherwise it idles until an M is populated at the start of the cycle. If they do, it will use the value of one of them at random.

When a bot reads from M, it checks if any bots have a local M value at the start of the cycle. Any number of bots can have values in there own M registers, and the bot will wait idle until the value is read out. It turns out that a bot has a local M register that it can write to. I made a test program to just try to sync 6 bots as fast as possible as a simpler test case. Here’s where I needed to really understand exactly how the global M register worked to get more improvements.

#EXAPUNKS SOLUTUIONS CODE#

I had to move to having all the code start in a single bot and replicate to reduce the size: All the NOOP I needed for padding was making the solution go over the maximum allowed size. My third solution extended this so that all the bots were using the global register. I did this by using idle instructions ( NOOP) to make sure they were out of phase: My second solution was similar to the first, but but had two nerves sharing the global register. I thought this would be a good chance to really understand this functionality. If two or more EXAs attempt to read from another EXAĪt the same time (or vice versa), one will succeed but which one succeeds will be If an EXA reads from the M register, it will pause execution until a If an EXA writes to the M register, it will pause execution until that value is read byĪnother EXA. The instructions for the game come in the form of a fictional zine. Playing through the game so far, I had never really had to master the exact way the global register was handled when multiple bots try to use it at once. The number of cycles is pretty much dominated by whichever nerve is handled slowest. Looking at the data progress in the chart, you can see how far ahead the direct communicating nerve gets from the other two. My initial solution was to have the most distant set of nerves use the global communication, and have the other two nerves handled by moving back and forth.

However, there’s also a global communication register that all the bots can read and write to for communicating at a distance. The most basic solution would be to have the robots move back and forth between the nerves. There’s 3 source nerves and 3 sink nerves, and you have to move the data between them. You need to write a program to connect your hand to your central nervous system. In the game fiction, your character is dying of a strange disease that’s turning you into a machine. However, on the puzzle “Mitsuzen HDI-10” I decided to dig in and try to go for an optimal solution. I don’t usually go back to make multiple passes at optimization unless I’m unhappy with what I implemented, or there’s an obvious major improvement. I usually write my solutions focusing on optimizing for cycles, as I find it most interesting.

Activity - How much the bots moved between networks or attacked other bots.

Size - The number of instructions you wrote.

Cycles - The number of cycles it takes to run through the test data.

One interesting feature of these games is that they let you see how your solution stacks up against all the other players. The gimmick for Exapunks is that your instructions are executing on tiny robots moving around inside computer networks. I actually made a lot of typos this time around since I recently did some x86 assembly and kept typing MOV instead of COPY to move values between registers.

#EXAPUNKS SOLUTUIONS SERIES#

Almost all of their games boil down to solving a series of puzzles in a simulated assembly like programming language with unusual constraints. It’s probably not a surprise that I’m a fan of Zackatronics games. For something a little different, I thought I’d look at the process I went through to optimize a solution in the game Exapunks.

Exapunks solutuions