The Nineteenth Byte

Nathan Merrill

2:03 PM

@Rainbolt I wonder what the dev manager said: "I have no idea! I've donated a can of spam!"

Lembik

@MitchSchwartz done :)

Poke

I don't want to look through a day's worth of scrollback so forgive me if this has been posted already

reddit.com/r/netsec/comments/591pvu/…

Nathan Merrill

oooh, I have another challenge idea: I'll take all of the KC historical alphabet challenges we've had, and pass in the output string, and a number from 1 to 26. They need to scale the output to the appropriate size.

Basically pattern matching

Dennis

@Adnan Done.

Lembik

hi @Dennis

Dennis

2:15 PM

Hello.

Dennis

2:25 PM

@Lembik Did you try dropping to runlevel 2? That might help, especially when multi-threading is involved.

Lembik

@Dennis I did exactly that!

Oh I wanted to ask.. do you think it's the kind of thing that would be sped up by a good GPU?

just out of interest

@Dennis my theory is that there are some complicated cache effects... maybe profiling could tell me?

Dennis

@Lembik Maybe. I have no idea how to program on a GPU (I also don't have an AMD GPU, so there's no point in trying), but given that the big loop is perfectly parallelizable, it might work pretty well.

Lembik

@Dennis cool. I didn't mean on my computer. I don't have a proper GPU

I was just interested

Dennis

Neither do I. I just use the integrated GPU of my processor.

Eᴀsᴛᴇʀʟʏ Iʀᴋ

@Rainbolt that's amazing

@TuxCopter you can't put tags in links, tags are links

Dennis

2:38 PM

@Lembik No clue. I'm hardly an expert when it comes to writing fast code. Could you share the Python wrapper for my submission?

Lembik

@Dennis bpaste.net/show/070089d369d1

Adnan

@Dennis Thanks :)

Lembik

can I complain about what a pain it was to get it work too :)

Actually Ton Hospel's was worse because he had some odd way of exiting that screwed up all pipes

so you were both a pain :)

Poke

@GLASSIC If that's truly the reason you got fired it sounds like a pretty shitty place to work

New Sandboxed Posts

0

A: Sandbox for Proposed Challenges

Generate antsy permutations code-golf permutations Introduction I defined the class of antsy permutations in an earlier challenge. As a reminder, a permutation p of the numbers from 0 to r-1 is antsy, if for every entry p[i] except the first, there is some earlier entry p[i-k] such that p[i]...

0

A: Sandbox for Proposed Challenges

Alphabet pattern matching We've had a lot of alphabet challenges. For this challenge, you are passed the output of an alphabet challenge, and you need to output the pattern scaled to size N. For example, if N=5 and you were passed the L-phabet: ABCDEFGHIJKLMNOPQRSTUVWXYZ BBCDEFGHIJKLMNOPQRST...

DJMcMayhem

2:52 PM

@Dennis Clearly:

59

A: Calculate the number of primes up to n

C, 0.026119s (Mar 12 2016) #include <math.h> #include <stdint.h> #include <stdio.h> #include <stdlib.h> #include <string.h> #include <time.h> #define cache_size 16384 #define Phi_prec_max (47 * a) #define bit(k) (1ULL << ((k) & 63)) #define word(k) sieve[(k) >> 6] #define sbit(k) ((word(k >> 1) >

@Lembik Hey, I have an AMD GPU (r9 290, which is slightly old but still pretty powerful). Want me to test something?

Dennis

@Lembik Sorry... I've been wanting to ask you: Does our code have to be correct for arbitrary nxn matrices or just the randomly generated ones? I'm asking since 35x35 will work perfectly fine with 128-bit integers for these, but the edge case of all 1's won't fit into the data type anymore...

@DJMcMayhem You've picked the one time I had a successful fastest code submission. :P And as you can tell from the revision history, it took some time to get it right...

@Lembik Hm, that's quite consistent with my timings in Bash.

Lembik

@DJMcMayhem would be cool but someone has to write some code first :)

DJMcMayhem

@Dennis Your first revision would have won the entire challenge

Lembik

@Dennis just with randomly generated ones is fine. In fact I slightly regret having integer coefficients instead of real numbers in the range 0 to 1

Dennis

Heh, my approach wouldn't even have been possible with real numbers... That said, I should try Gray codes instead of that.

Lembik

3:00 PM

@Dennis It is probably too late to change it now and in any case, I think numerical accuracy would then come into play

but I think it might be a better challenge

DJMcMayhem

@Lembik OK. Well, if you want me to test anything, feel free to ping me and I'd be happy to help.

Lembik

my understanding is that the Glynn formula is much better for numerical accuracy

@DJMcMayhem thanks!

DJMcMayhem

I'm usually free in the 0-6 UTC range

Lembik

cool

Dennis

@DJMcMayhem Unfortunately, to try GPU programming, I'd need ssh access to your computer. :P

Lembik

3:01 PM

so sad for GMT though....

@DJMcMayhem you could set up a virtual machine for him?

:)

DJMcMayhem

@Dennis Really? You couldn't just send me the code and have me run it?

New Sandboxed Posts

0

A: Sandbox for Proposed Challenges

The Last Monday code-golf date Monday, October 31st, is Halloween. And it got me thinking -- I wonder what other months have the last day of the month also be a Monday? Input A positive integer in any convenient format representing a year, y > 0. Output A list of the months of that year wh...