Naive benchmark while generating 15M rows of CSV

benchmark.md

Code is over here

go1.0.2

[master] thbar@~/git/golang-playground/bulkload: time go run generate_csv.go 

real  0m41.522s
user	0m39.015s
sys	0m1.825s

jruby 1.7.2

[master] thbar@~/git/golang-playground/bulkload: time rake generate_csv

real    1m19.560s
user    1m20.972s
sys    0m4.928s

mri 2.0.0-p0

[master] thbar@~/git/golang-playground/bulkload: time rake generate_csv

real	1m56.004s
user	1m50.204s
sys	0m3.403s

mri 1.9.3-p392

[master] thbar@~/git/golang-playground/bulkload: time rake generate_csv

real    2m0.235s
user	1m55.443s
sys	0m2.993s

If you use opencsv (Java lib) instead of the csv (Ruby), you can beat Go with JRuby 1.7.2.

require 'java'
require 'opencsv-2.3.jar'

java_import 'au.com.bytecode.opencsv.CSVWriter'
java_import 'java.io.FileWriter'

writer = CSVWriter.new(FileWriter.new('java-data.csv'), "\t".ord)

writer.writeNext(['guid', 'some_field', 'some_other_field'])

15000000.times do |counter|
  writer.writeNext(["guid_#{counter}", "some_field_#{counter}", "some_other_field_#{counter}"])
end

writer.close

And here's my results (MBA i7 2.0Ghz):

35.15s user
1.48s system
107% cpu
34.031 total

I've got the same time as you for the Go version.