Hash#fetch with second argument is slower than Hash#fetch with block

[HarlemSquirrel]

require "benchmark/ips"

MY_HASH = (1..99).each_with_object({}) { |n,h| h[n.to_s] = n }

def fast
  MY_HASH.fetch('nope') { 'default' }
end

def slow
  MY_HASH.fetch 'nope', 'default'
end

Benchmark.ips do |x|
  x.report("Hash#fetch with block") { fast }
  x.report("Hash#fetch with second argument") { slow }
  x.compare!
end

Ruby 2.5.7

Warming up --------------------------------------
Hash#fetch with block
                       288.905k i/100ms
Hash#fetch with second argument
                       315.096k i/100ms
Calculating -------------------------------------
Hash#fetch with block
                          5.247M (± 0.7%) i/s -     26.290M in   5.011189s
Hash#fetch with second argument
                          6.183M (± 5.2%) i/s -     30.879M in   5.009613s

Comparison:
Hash#fetch with second argument:  6182898.3 i/s
Hash#fetch with block:  5246593.2 i/s - 1.18x  slower

Ruby 2.7.0

Warming up --------------------------------------
Hash#fetch with block
                       328.574k i/100ms
Hash#fetch with second argument
                       394.964k i/100ms
Calculating -------------------------------------
Hash#fetch with block
                          6.444M (± 1.1%) i/s -     32.529M in   5.048419s
Hash#fetch with second argument
                          8.438M (± 0.6%) i/s -     42.261M in   5.008428s

Comparison:
Hash#fetch with second argument:  8438304.4 i/s
Hash#fetch with block:  6444222.1 i/s - 1.31x  slower

[HarlemSquirrel]

Upon further testing, this seems to not be helpful.