Benchmarking if versus map-matching (a kind of pattern-matching in ruby)

fizz_buz.rb

require "benchmark"

def log msg
  # NOOP
end

def with_if(x)
  x.times do |n|
    if (n % 3 == 0) && (n % 5 != 0)
      log 'Fizz'
    elsif (n % 3 != 0) && (n % 5 == 0)
      log 'Buzz'
    elsif (n % 3 == 0) && (n % 5 == 0)
      log 'FizzBuzz'
    else log n
    end
  end
end

def without_if(x)
  x.times do |n|
    h = {
      'Fizz' => (n % 3 == 0) && (n % 5 != 0),
      'Buzz' => (n % 3 != 0) && (n % 5 == 0),
      'FizzBuzz' => (n % 3 == 0) && (n % 5 == 0)
    }
    log(h.key(true) || n)
  end
end

n = 5_000_000
Benchmark.bm(10) do |x|
  x.report("with if:") { with_if    n }
  x.report("without:") { without_if n }
end

I'm a CleanCode maniac and love the no-ifs-approach tho ... it's almost 5x slower in this case.
I prefer clean and beauty stuff but sometimes we need brachial brutality 😀

                 user     system      total        real
with if:     1.470000   0.010000   1.480000 (  1.484849)
without:     7.560000   0.130000   7.690000 (  7.735552)

Restructuring the benchmark to use IPS drops the performance difference from ~5x to ~2.6-2.8x (though without conditionals, there's a much higher variance in individual call performance), so it may perform worse on larger data sets. It's still definitely slower, but maybe less of an outright sacrifice if you're going for the no-conditionals thing.

Calculating -------------------------------------
            with if:    41.975k i/100ms
            without:    33.080k i/100ms
-------------------------------------------------
            with if:      2.712M (± 8.1%) i/s -     13.474M
            without:      1.029M (±17.2%) i/s -      4.962M

Comparison:
            with if::  2712057.2 i/s
            without::  1029371.0 i/s - 2.63x slower