Ruby comparison of BigDecimal vs. Float

rub_bigdec_float.rb

# I've compiled this in order to test the speed on using a BidDecimal, which some in the community assert is slower than using a float.  
# So far, it looks like BigDecimal is faster, and I've created this file to solicit feedback.  Cheers.

run_size = 10000
p_fin = []
p_fin << "Each run #{run_size}"

t1 = Time.now
run_size.times do
  r = BigDecimal.new
  r = rand
  s = BigDecimal.new
  s = rand
  t = r/s
end
p_fin << [Time.new - t1, "BigDecimal = rand"]

t1 = Time.now
run_size.times do
  r = BigDecimal.new( rand.to_s )
  s = BigDecimal.new( rand.to_s )
  t = r/s
end
p_fin << [Time.new - t1, "BigDecimal(rand.to_s)"]

t1 = Time.now
run_size.times do
  r = Float.new
  r = rand.round(14)
  s = Float.new
  s = rand.round(14)
  t = r/s
end
p_fin << [Time.new - t1, "float.round."]

t1 = Time.now
run_size.times do
  r = Float.new
  r = rand
  s = Float.new
  s = rand
  t = r/s
end
p_fin << [Time.new - t1, "float."]

pp p_fin

Careful, every time you do this you're not doing anything different:

r = Float.new
r = rand

and

r = BigDecimal.new
r = rand

In both cases, the object created in the first line is thrown away and r is reassigned to the return value of Kernel#rand, which is a Float