[Inori]

In honour of 500,000 posts, Serenity is holding a recruitment challenge. The challenge is to create a method that finds all primes under or equal to n.

Constraints:

• total time for 1,000 tests under 2 seconds
  
• average time per test under 0.01 seconds
  
• get the correct number of primes (1,229 for our test case)
  
• same result as test method
  

The following method dictates whether or not a submission passes:

Code:

def benchmark(method): """ Used for calculating a pass/fail in Serenity's 500k recruitment challenge constraints: - <5 seconds total runtime - <0.01 seconds average runtime - correct number of primes (1229 primes <= 10000) - correct primes (compared to sample method) """ time=__import__('timeit').timeit( '%s(10000)'%method, setup='from __main__ import '+method, number=1000 ) res=eval(method+'(10000)') pss=round(time,6)<5 and time/1000<0.01 and len(res)==1229 and test(10000)==res return '\n'.join([ method+': '+('Fail','Pass')[pss]+ ('',' (distinction'+'+'*(time<0.0025)+')')[time<0.00324851038689], 'Average: '+`round(time/1000,6)`, 'Total: '+`round(time,6)`, 'Score: '+`(((0,3)[pss]+round(5-time,5))*(1,(1.25,1.5)[time<0.0025])[time<0.00324851038689])*10` ])

and the scores for my benchmark method:

Code:

test: Pass Average: 0.003354 Total: 3.354264 Score: 46.4574

Bonus points: The method I used is (one of) the most efficient for this task. Correct identification of who devised it and the name of it is +2 points

Submission: PM me (on here or discord) with your entry to submit. If you'd prefer to do this challenge in another language, let me know, and I'll write something up.

Recruitment: The top 50% of entries will be recruited to Serenity

[Rick]

I like the idea of the thread but I don't see why you used such a high threshold, I mean very few people will be able to beat established prime counting algorithms. I could go and search for very efficient algorithms if I wanted and learn how they work but would that show my intellect?

Wouldn't a better idea be to have no threshold and then judge off of the responses individually off of efficiency, creativity etc?

[Inori]

I like the idea of the thread but I don't see why you used such a high threshold, I mean very few people will be able to beat established prime counting algorithms. I could go and search for very efficient algorithms if I wanted and learn how they work but would that show my intellect?

Wouldn't a better idea be to have no threshold and then judge off of the responses individually off of efficiency, creativity etc?

I'm not even considering the a "failed" program a fail. Passing is just a point bonus.

[lux]

Uuuh, I made a Go Web api for calculating primes... Will write benchmarks.

---

HTTP Endpoint:
Server response for primes <= 1229 - 5ms

Runtime Logic:
Single instance benchmark - 1526696 ns/op
Total time for 1000 ops - 1.7

[insidious]

Is there an ETA for when winners will be announced?

[Inori]

Is there an ETA for when winners will be announced?

No idea. I'm just waiting for more responses.

[m0dem]

I need to get off my bum (...not literally...) and do this.

---

Could we do a lookup into an internet API?
(probably cheating...)

[Inori]

I need to get off my bum (...not literally...) and do this.

---

Could we do a lookup into an internet API?
(probably cheating...)

you could, but an API call would probably take way longer than anything local

[m0dem]

Aww...
Finally, I got my prime function to work...

Code:

sieve: Fail Average: 0.005121 Total: 5.12078 Score: -1.2078

I might try to make it better, somehow.

[Inori]

Aww...
Finally, I got my prime function to work...

Code:

sieve: Fail Average: 0.005121 Total: 5.12078 Score: -1.2078

I might try to make it better, somehow.

Jesus, my scoring method needs work...