WHY is the nonthreaded process faster inserting 10k data to mongo then 10threads inserting 1000 records

ruby-mongo-nonthreaded.rb

require 'rubygems'
require 'mongo'
require 'digest/md5'

@conn 	= Mongo::Connection.new
@db 		= @conn['testdb']
@coll	= @db['dummy']

@coll.remove

startdate = Time.now

10000.times do |i|
	checksum = Digest::MD5.hexdigest(i.to_s)
	checksum2 = Digest::MD5.hexdigest(checksum)
	@coll.insert({'a' => i+1, 'check' => checksum, 'check2' => checksum2})
end

enddate = Time.now
subtime = enddate - startdate
puts "Final time: " + subtime.to_s

# puts "There are #{@coll.count} records. There they are:"
# @coll.find.each { |doc| puts doc.inspect }

ruby-mongo-threaded.rb

require 'rubygems'
require 'mongo'
require 'digest/md5'

@conn 	= Mongo::Connection.new
@db 		= @conn['testdb']
@coll	= @db['dummy']

@coll.remove

startdate = Time.now

def insert(i)
	2000.times do |j|
		checksum = Digest::MD5.hexdigest(i.to_s + j.to_s)
		checksum2 = Digest::MD5.hexdigest(checksum)
		@coll.insert({'a' => i+j+1, 'check' => checksum, 'check2' => checksum2})
	end
end

@arr = []

5.times do |i|
	@arr[i] = Thread.new {
		insert(i)
	}
end

@arr.each {|t| t.join }

enddate = Time.now
subtime = enddate - startdate
puts "Final time: " + subtime.to_s

# puts "There are #{@coll.count} records. There they are:"
# @coll.find.each { |doc| puts doc.inspect }