@@ -2,8 +2,26 @@ A quick benchmark vs Starling doesn't deal with contention running in Production mode - with 50,000 pre-existing RudeQueue items + + # empty queue + user system total real + rudequeue 30.630000 1.050000 31.680000 ( 44.472249) + starling - one 1.450000 0.550000 2.000000 ( 11.144641) + starling - many 1.160000 0.420000 1.580000 ( 18.012299) + + # after 1st run -> 10,000 :processed items + user system total real + rudequeue 30.910000 1.090000 32.000000 ( 45.863217) + starling - one 1.380000 0.530000 1.910000 ( 12.189596) + starling - many 1.110000 0.390000 1.500000 ( 17.231946) + + # after 5th run -> 50,000 :processed items + user system total real + rudequeue 33.670000 1.090000 34.760000 ( 47.945849) + starling - one 1.340000 0.510000 1.850000 ( 12.792256) + starling - many 1.140000 0.410000 1.550000 ( 18.573452) >> @starlings.length => 148 + # with the deprecated TokenLock user system total real rudequeue 49.060000 2.970000 52.030000 ( 77.439360) starling - one 1.710000 0.890000 2.600000 ( 11.684323) @@ -13,7 +31,11 @@ MySQL 5 ruby 1.8.6 - To run this, + key points: + If you don't need items to persist, should destroy on processed! + (need to implement this!) + + To run this benchmark; :~ $ sudo gem install starling :~ $ starling -d performance/benchmark_vs_starling.rb cbad1f3096a196e94675ee950d43fd17b93eedf6 MatthewRudy MatthewRudyJacobs@gmail.com http://github.com/matthewrudy/rudeq/commit/d5b9d93c50ec1396da89c9dc8e94927c907a1bd3 d5b9d93c50ec1396da89c9dc8e94927c907a1bd3 2008-07-25T06:44:28-07:00 2008-07-25T06:44:28-07:00 revised benchmarks since move to PessimisticLock => MUCH FASTER eb83314672cb5d0c75f18f63998bf06c824e549b MatthewRudy MatthewRudyJacobs@gmail.com