3 ms·

I'm 100% a ruby fan, but it seems crazy they went from Go to ruby if ultimate performance isn't their goal. Wouldnt a static type language be worlds faster?

pkulak 14y ago
Go is pretty slow right now. I wrote a prime number finder in Go and Ruby 1.9 and Ruby smoked it. Granted that's probably mostly due to all the math in Ruby being C, but still, you can't just say that Go is faster because it's compiled before hand.
- NateDad 14y ago
  Honestly, my guess is that you were doing something wrong in the Go code. Go is definitely faster than Ruby, even on a single core machine. http://benchmarksgame.alioth.debian.org/u64/benchmark.php?test=all&lang=go&lang2=yarv http://benchmarksgame.alioth.debian.org/u64/benchmark.php?te...
  - pkulak 14y ago
    EDIT: haha, thanks so much, guys! Stupid mistake. Go is _significantly_ faster if written properly. I'd love to know what I screwed up. The results surprised me: https://gist.github.com/pkulak/99d2d5a5968b5fc03754 https://gist.github.com/pkulak/99d2d5a5968b5fc03754 https://gist.github.com/pkulak/909d882614e0781e4525 https://gist.github.com/pkulak/909d882614e0781e4525 bender:Desktop phil$ time go run primes.go Found them! 78702 real 0m21.349s user 0m21.304s sys 0m0.033s bender:Desktop phil$ ruby -v ruby 1.9.3p194 (2012-04-20 revision 35410) [x86_64-darwin11.4.0] bender:Desktop phil$ time ruby primes.rb Found them! 78702 real 0m7.656s user 0m7.651s sys 0m0.005s bender:Desktop phil$ time go build primes.go real 0m0.269s user 0m0.231s sys 0m0.033s
    - hndc 14y ago
      You're using integer arithmetic in Ruby and floating-point arithmetic in Go. Try replacing math.Mod(float64(i), float64(j)) with i%j My results: $ time go run primes.go Found them! 78702 real 0m0.835s user 0m0.787s sys 0m0.039s $ time ruby primes.rb Found them! 78702 real 0m21.013s user 0m21.005s sys 0m0.016s
      - agentS 14y ago
        It's worth noting that the time on the go side includes the time to run the compiler, and linker, since you're using `go run`, instead of `go build`. Not saying that's bad, just something to keep in mind when benchmark numbers are in the second range. (I understand that you did because the original poster did it)
    - dmit 14y ago
      You should use the same integer remainder operator (%) as in the Ruby version instead of math.Mod(float64(i), float64(j)): <...> for j := 2; j < limit; j++ { if i % j == 0 { found = true <...>
- pjscott 14y ago
  I assume you're comparing the relative performance of their bignum libraries? If so, that's not exactly a representative comparison.
NateDad 14y ago
I didn't see anything in the article that mentioned performance at all. I think they just wanted to try out Go. Ruby being the industry standard for new websites these days, it seems logical they'd choose it if they didn't like their experience with Go.

pkulak 14y ago

Go is pretty slow right now. I wrote a prime number finder in Go and Ruby 1.9 and Ruby smoked it. Granted that's probably mostly due to all the math in Ruby being C, but still, you can't just say that Go is faster because it's compiled before hand.

NateDad 14y ago
Honestly, my guess is that you were doing something wrong in the Go code. Go is definitely faster than Ruby, even on a single core machine. http://benchmarksgame.alioth.debian.org/u64/benchmark.php?test=all&lang=go&lang2=yarv http://benchmarksgame.alioth.debian.org/u64/benchmark.php?te...
- pkulak 14y ago
  EDIT: haha, thanks so much, guys! Stupid mistake. Go is _significantly_ faster if written properly. I'd love to know what I screwed up. The results surprised me: https://gist.github.com/pkulak/99d2d5a5968b5fc03754 https://gist.github.com/pkulak/99d2d5a5968b5fc03754 https://gist.github.com/pkulak/909d882614e0781e4525 https://gist.github.com/pkulak/909d882614e0781e4525 bender:Desktop phil$ time go run primes.go Found them! 78702 real 0m21.349s user 0m21.304s sys 0m0.033s bender:Desktop phil$ ruby -v ruby 1.9.3p194 (2012-04-20 revision 35410) [x86_64-darwin11.4.0] bender:Desktop phil$ time ruby primes.rb Found them! 78702 real 0m7.656s user 0m7.651s sys 0m0.005s bender:Desktop phil$ time go build primes.go real 0m0.269s user 0m0.231s sys 0m0.033s
  - hndc 14y ago
    You're using integer arithmetic in Ruby and floating-point arithmetic in Go. Try replacing math.Mod(float64(i), float64(j)) with i%j My results: $ time go run primes.go Found them! 78702 real 0m0.835s user 0m0.787s sys 0m0.039s $ time ruby primes.rb Found them! 78702 real 0m21.013s user 0m21.005s sys 0m0.016s
    - agentS 14y ago
      It's worth noting that the time on the go side includes the time to run the compiler, and linker, since you're using `go run`, instead of `go build`. Not saying that's bad, just something to keep in mind when benchmark numbers are in the second range. (I understand that you did because the original poster did it)
  - dmit 14y ago
    You should use the same integer remainder operator (%) as in the Ruby version instead of math.Mod(float64(i), float64(j)): <...> for j := 2; j < limit; j++ { if i % j == 0 { found = true <...>
pjscott 14y ago
I assume you're comparing the relative performance of their bignum libraries? If so, that's not exactly a representative comparison.

NateDad 14y ago

I didn't see anything in the article that mentioned performance at all. I think they just wanted to try out Go. Ruby being the industry standard for new websites these days, it seems logical they'd choose it if they didn't like their experience with Go.