SEQUENCE MODE CALCULATOR - finally got it!!

[Anonymate]

FINALLY!!!! I'm so happy!! This was by far the hardest one yet. I spent about 7 hours in total solving this. Actually, about 2 hours to get a solution that required the last node to use 18 instructions (14 if I ignored the fact that when there is a draw I must output 0), and then the remaining time to reduce from 18 to 15 instructions. Getting to 16 instructions wasn't THAT hard, but then taking that last one away... oh well... 1496/7/100 (+ both stack nodes)

My sh**ty solution (SPOILER ALERT): http://steamcommunity.com/sharedfiles/filedetails/?id=1284643533

I have no idea how people solved it with only 5 nodes, or 60-ish instructions... that's quite insane.

Is there any even harder level after this one?

[lisamariefan]

This is definitely one of the harder puzzles, and my score isn't much better than yours (faster, but with more instructions). I think this one might be the hardest just because it's the hardest to fit within the avaliable nodes, but that's subjective. It wasn't the last puzzle I solved, but it's probably one of the ones I had a harder time taking an approach to.

[Anonymate]

Originally posted by lisamariefan:

This is definitely one of the harder puzzles, and my score isn't much better than yours (faster, but with more instructions). I think this one might be the hardest just because it's the hardest to fit within the avaliable nodes, but that's subjective. It wasn't the last puzzle I solved, but it's probably one of the ones I had a harder time taking an approach to.

Yes, it was extremely hard to fit within these nodes. But then you look at the unofficial records, see people solved it using only 3 nodes with 42 instructions and I just wonder: HOW!?

I've finally unlocked all the remaining levels and have already read the statement for all of them (9 left to complete). I think the CHARACTER TERMINAL and IMAGE TEST PATTERN 3 will be the most annoying ones (not necessarily the hardest). Other than those, the remaining problems should be easy.

[lisamariefan]

Not quite sure how it's possible with 3 nodes and 42 instructions, but this post prompted me to give this one another go. While I wound up resuing some code for convenience, I changed the approach to reflect some of the way I've programmed some other "lists" (in this case a frequency list).

It's something you may come across and develop techniques for in later puzzles, but I went from a solution that was 1145/7/105 to 2118/5/64. I suppose I could post my newer solution...though there is something I do want to point out that is not obvious in my stats: In the less compact solution I only used one stack node. In the new one I used both. I've also done a lot of puzzling and have completed both this and Shenzhen I/O in between my initial solution and the new one (it's been months since I even gave it another shot and I wound up copying and pasting some old code and adjusting it).

Another interesting note between my two solutions: In my first solution I had the frequency's going from the highest value to the lowest. In my new solution it's lowest to highest.

Old:
http://steamcommunity.com/sharedfiles/filedetails/?id=881251700

New:
http://steamcommunity.com/sharedfiles/filedetails/?id=1285351206

I dunno how much you consider looking at different solutions spoiling yourself, but those are the two solutions in question.

Edit: You know, I'm not sure exactly how your numbers are flowing off the top of my head. In my solutions they work in a spiral of sorts in the first and a shuttle between the two nodes in the second., with the frequency's being checked and dumped out the right. An additional note of mine is that (and it's probably solvable with a bit of code in an extra node tbh) in the newer solution each sequence dums an additional -1 in the memory, meaning that if it was eer given too many sequences it would lock up. Of course, this never happens in practice.

Edit2: Having looked at your profile and stats I don't think you'll struggle too badly with what's left. I mean, you beat the first campaign after all. Having said that, I might have fuzzy memories of how tough some of the later puzzles were.

I would also like to add that character terminal is probably easier than the image test pattern 3. Image test pattern 3 (at least the way I did it) is a lot like my spacial path viewer, except you it's output is hardcoded, instead of adapting for variables you're given.

[Anonymate]

All nodes with 15 instructions :o, that must have been hard to fit. I won't look at the new solution yet because I might want to try to improve mine later on. To use 5 nodes, the idea I had was the same I used for the NO_MEMORY achievement: each "memory" node solves its part of the problem and passes the result forward, instead of having different nodes solve different parts - and that's actually what I tried to implement first since seemed easier (code one node, then just CTRL+C/V everything), but couldn't fit it within 15 instructions for this problem... not sure if possible.

[lisamariefan]

Yeah, all nodes with 15 instructions was tough. The three rightmost nodes were the toughest to work with on the last part and I had to do some clever trickery with how jumps worked for branching code to be...as optimized as I could.

Interestingly enough, out of now 4 (it was 2 last night) solutions on my leaderboards, my solution with all nodes taking 15 instructions is the fastest.

[DragonBoy]

This one was the last one I solved. Getting my algorithm to fit was definitely the hardest part.

[Zednaught]

Originally posted by lisamariefan:

Yeah, all nodes with 15 instructions was tough. The three rightmost nodes were the toughest to work with on the last part and I had to do some clever trickery with how jumps worked for branching code to be...as optimized as I could.

Interestingly enough, out of now 4 (it was 2 last night) solutions on my leaderboards, my solution with all nodes taking 15 instructions is the fastest.

This is obviously resurrecting an old thread, but I just wanted to thank you for your post. It took me, no joke, 2 weeks to solve sequence mode calculator.

I knew there had to be a simpler way to solve the puzzle; I just couldn't figure out how to spool and unspool the memory stacks.

After looking at your post, it all clicked. I knew I could use your technique to solve the prolonged sequence sorter, and I think I managed to solve it in 30 minutes, and it was pretty straightforward.

Edit: it also makes the T31 puzzle a breeze too.

[lisamariefan]

Originally posted by Zednaught:

Originally posted by lisamariefan:

Yeah, all nodes with 15 instructions was tough. The three rightmost nodes were the toughest to work with on the last part and I had to do some clever trickery with how jumps worked for branching code to be...as optimized as I could.

Interestingly enough, out of now 4 (it was 2 last night) solutions on my leaderboards, my solution with all nodes taking 15 instructions is the fastest.

This is obviously resurrecting an old thread, but I just wanted to thank you for your post. It took me, no joke, 2 weeks to solve sequence mode calculator.

I knew there had to be a simpler way to solve the puzzle; I just couldn't figure out how to spool and unspool the memory stacks.

After looking at your post, it all clicked. I knew I could use your technique to solve the prolonged sequence sorter, and I think I managed to solve it in 30 minutes, and it was pretty straightforward.

Edit: it also makes the T31 puzzle a breeze too.

Heh. I actually wound up changing my approach to not need all nodes in a second solution. It was slower of course, but it has to do with pulling a count in a stack and running from there.

Isn't T31 the RAM simulator? that one was fun, though I wouldn't mind having a solution that doesn't need to reset the reference point to "0" each time to find the right index. It's hard to do when you have limited instructions though.

[Zednaught]

Originally posted by lisamariefan:

Originally posted by Zednaught:

This is obviously resurrecting an old thread, but I just wanted to thank you for your post. It took me, no joke, 2 weeks to solve sequence mode calculator.

I knew there had to be a simpler way to solve the puzzle; I just couldn't figure out how to spool and unspool the memory stacks.

After looking at your post, it all clicked. I knew I could use your technique to solve the prolonged sequence sorter, and I think I managed to solve it in 30 minutes, and it was pretty straightforward.

Edit: it also makes the T31 puzzle a breeze too.

Heh. I actually wound up changing my approach to not need all nodes in a second solution. It was slower of course, but it has to do with pulling a count in a stack and running from there.

Isn't T31 the RAM simulator? that one was fun, though I wouldn't mind having a solution that doesn't need to reset the reference point to "0" each time to find the right index. It's hard to do when you have limited instructions though.

Yeah, the T31 was the RAM simulator; I thought it was cool to design something so functional in the game.

I can imagine how to not have to reset to 0: saving the last position in the stack, subtracting that from the new index value, and then going either up or down depending if the difference is positive or negative. I'm spoiled by EXAPUNKS, though, because it would be much easier to program there than in TIS100.

[lisamariefan]

I haven't grabbed Exapunks yet (and honestly at some point I need to continue Spacechem now that I got that courtesy of dwangoAC). Even so, i do get being spoiled some. There's some stuff you can do even in Shenzhen I/O that ya can't do in TIS that spoil me.

[Zednaught]

Originally posted by lisamariefan:

I haven't grabbed Exapunks yet (and honestly at some point I need to continue Spacechem now that I got that courtesy of dwangoAC). Even so, i do get being spoiled some. There's some stuff you can do even in Shenzhen I/O that ya can't do in TIS that spoil me.

Both games are great; SpaceChem may be old, but it feels just a fresh as any Zachtronics game.

[mass spectre]

wow, loaded this up after a long time away. I got all but 4 on my initial frenzy. Haven't looked at any solutions yet. it is interesting to see the conversations. This has been one of my favorite games. haven't tried exapunks yet (bought the swag for it though). I didn't enjoy Shenzhen IO as much as TIS-100. this has been my favorite of the zachtronics games so far.(I tried several out at the web site as well).
still to go for me:
Sequence Mode Calculator - 11/23 - 2107/7/82
Dynamic Pattern Detector - 11/24 - 625/8/49 - ran first time
Sequence Gap Interpolator - 11/25 - 854/7/65
Prime Factor Calculator - 11/25 - 3682/10/111
Cheers.

[Zednaught]

Originally posted by mass spectre:

wow, loaded this up after a long time away. I got all but 4 on my initial frenzy. Haven't looked at any solutions yet. it is interesting to see the conversations. This has been one of my favorite games. haven't tried exapunks yet (bought the swag for it though). I didn't enjoy Shenzhen IO as much as TIS-100. this has been my favorite of the zachtronics games so far.(I tried several out at the web site as well).
still to go for me:
Sequence Mode Calculator - 11/23 - 2107/7/82
Dynamic Pattern Detector - 11/24 - 625/8/49 - ran first time
Sequence Gap Interpolator
Prime Factor Calculator
Cheers.

Keep at it! You're almost there (and you've already finished the hardest puzzles).

Once you've fiinished the game and had a well deserved break, consider a game called Prime Mover. The problems you solve are extremely similar to TIS-100, but the gameplay is closer to SpaceChem.

The game starts simple, but it gets quite difficult, very quickly. In many ways, your toolset is more limited than even TIS-100's simple commands, though in other ways it's more flexible. I'm learning how crucial it is to compartmentalize every step in solving a problem.

[mass spectre]

Thanks, Zed. I added prime mover to my wish list. Just finished the last one. Kind of a weird feeling after all this time. Was kind of cool seeing that one on the low end of cycle count. I'm still working on Dig or Die (It's on the top of my "rare achievement" thing on my profile). I can see the full completion on TIS-100 being less frequent than say completing The Binding of Isaac. Not sure why the bridge and tower building on Dig or Die are less common than that.