hacker.org

I again analyzed and shortened your SuperHack program:
Happy holidays!

Hmm, I cannot find my original solution. My current bublesort like solution looks following way (6110 cycles): 0,,@2@@@@@@@@@@ xv2vx2^d1-1^d1+v2v1+x^?$1[,21^?$g1vpP22^?$1vp! (Too long for Kolmogorov KOTH challenge) I am currently working on challenge 'Tiny Sort' and since I could not beat your code...

Thanks for the congratulations!

My solution to challenge 'Super Fast Brainfuck' has code size 1885, but the SuperHack program consists of only 447 instructions. Furthermore, I still see room for improvement.

Although it is required according to the challenge text, I wrote no general-purpose Brainfuck interpreter. Even worse, my solution does not compute all test cases correctly and the SuperHack program was submitted to the server as long as new King of the Hill benchmarks were set: %+.<>f /\ x, 5+ 91 ...

Yes, I have climbed the hill!

Next, I might try to solve challenge 'Super Fast Brainfuck', although I still do not know how to use or synchronize threads in SuperHack. For questions, I get back to you via PM if that is okay?

How small is your smallest solution? My last approach resulted in an interpreter with code size 189: /%,x\/-\ /02^<x01gd?s!-11^0\ /-x35*-?\14^<p$ /9+-xx*1\/,15^>$ :g10//s+\?d9^2\//s=0s/s*x+0s/-=\g60$?^2\xx*1:x=\8=sv3v3\:14^<+\ \6]s\20g1@/1[1/\:x3^?/116^<?/+s\3v+2v2v/ \0s\4v1v-3v/$0>^41/ I have no cl...

The loop implementation is really (time) bottleneck in the interpretter (it costs a lot of time to detect te loop and a lot of time to find where to jump). I would expect jumps to be precomputed in interpret. (Actually this is the only place where interpretter gains against (original) compiller as ...

Actually nested loops are simple in the already implemented code, which didn't try to be fast so no s's are used for back loops ... you count number of opened ['s and put '/','\' given number of lines above or under ('\' is switched with '/' when [ is switched with ]). When using s's an option is t...

Actually using } instead of 2] would short the code, but for >>> the 6] would become better (what I have not implemented) and actually test cases seems do not contain more than two consecutive >'s so it's definitely thing to consider not only in code size optimization challenge, but even in time op...

Wow, I am even more impressed. You wrote a multi-threaded compiler which is smaller than my interpreter. (In fact, I have a smaller interpreter, but it requires too many cycles.) First, the whole Brainfuck program is loaded into memory cells (0++, 0). The memory cells (0++, 1) are used as the tape. ...

That is right, if the program starts with %00 instead of %0, then you can replace 1[, by +,. Or you can use a different arithmetic operation. Nice trick. I will post my solution of challenge 'Small Brainfuck' within the next few days in the appropriate thread. And yes, the SuperHack challenges are r...

Oh yeah, now that you mention it. Initially, I had also considered to solve the Brainfuck challenges that way. The idea was to translate Brainfuck programs into SuperHack using the following substitutions: Brainfuck command | SuperHack equivalent ------------------+--------------------- (Program Sta...

Did I miss something? How are the Brainfuck challenges solved with a compiler?