Rails App ruby 2.1 MRI vs Rubinius 2.0

gistfile1.rb

Server:

  iMac
  Mac10,1
  Intel Core 2 Duo
  3,06 GHz
 
  Cache L2:	3 MB
  RAM:	8 GB
  bus:	1,07 GHz

App: 
> rails new test_app
> rails g scaffold Post author text

Test:

*Start server puma in prod env*

>rails s puma -e production
=> Booting Puma
=> Rails 4.0.0 application starting in production on http://0.0.0.0:3000
=> Run `rails server -h` for more startup options
=> Ctrl-C to shutdown server
Puma 2.6.0 starting...
* Min threads: 0, max threads: 16
* Environment: production
* Listening on tcp://0.0.0.0:3000

*Start Test*
>ab -n400 -c16 http://127.0.0.1:3000/posts

*Results:*

*Rubinius 2.0.0*

  Benchmarking 127.0.0.1 (be patient)
  Completed 100 requests
  Completed 200 requests
  Completed 300 requests
  Completed 400 requests
  Finished 400 requests


  Server Software:        
  Server Hostname:        127.0.0.1
  Server Port:            3000

  Document Path:          /posts
  Document Length:        916 bytes

  Concurrency Level:      16
  Time taken for tests:   5.672 seconds
  Complete requests:      400
  Failed requests:        0
  Write errors:           0
  Total transferred:      681600 bytes
  HTML transferred:       366400 bytes
  Requests per second:    70.52 [#/sec] (mean)
  Time per request:       226.877 [ms] (mean)
  Time per request:       14.180 [ms] (mean, across all concurrent requests)
  Transfer rate:          117.35 [Kbytes/sec] received

  Connection Times (ms)
                min  mean[+/-sd] median   max
  Connect:        0    0   1.1      0      11
  Processing:     7  225 148.2    203     861
  Waiting:        7  222 146.9    202     824
  Total:          8  225 148.1    204     861

  Percentage of the requests served within a certain time (ms)
    50%    204
    66%    280
    75%    322
    80%    341
    90%    422
    95%    521
    98%    563
    99%    579
   100%    861 (longest request)

 
 *MRI*
 
 Benchmarking 127.0.0.1 (be patient)
 Completed 100 requests
 Completed 200 requests
 Completed 300 requests
 Completed 400 requests
 Finished 400 requests


 Server Software:
 Server Hostname:        127.0.0.1
 Server Port:            3000

 Document Path:          /posts
 Document Length:        916 bytes

 Concurrency Level:      16
 Time taken for tests:   3.187 seconds
 Complete requests:      400
 Failed requests:        0
 Write errors:           0
 Total transferred:      681600 bytes
 HTML transferred:       366400 bytes
 Requests per second:    125.49 [#/sec] (mean)
 Time per request:       127.496 [ms] (mean)
 Time per request:       7.968 [ms] (mean, across all concurrent requests)
 Transfer rate:          208.83 [Kbytes/sec] received

 Connection Times (ms)
               min  mean[+/-sd] median   max
 Connect:        0    0   1.2      0      14
 Processing:     4  125  75.5    119     394
 Waiting:        4  120  73.9    110     394
 Total:          4  126  75.6    120     398

 Percentage of the requests served within a certain time (ms)
   50%    120
   66%    149
   75%    168
   80%    184
   90%    235
   95%    272
   98%    292
   99%    357
  100%    398 (longest request)

Any chance you can do a test that runs for at least 10 seconds? These short benchmarks make runtimes that use a JIT (like Rubinius and JRuby) look worse than they are in practice.

yea this test means nothing..

I'd suggest you to use siege with -i and, as mentioned above, for a little more time.