{"id":5590,"date":"2018-01-03T02:41:53","date_gmt":"2018-01-03T02:41:53","guid":{"rendered":"http:\/\/www.garysieling.com\/blog\/?p=5590"},"modified":"2018-01-03T02:41:53","modified_gmt":"2018-01-03T02:41:53","slug":"scala-time-methods-track-average-variance","status":"publish","type":"post","link":"https:\/\/www.garysieling.com\/blog\/scala-time-methods-track-average-variance\/","title":{"rendered":"Scala: time methods and track average & variance"},"content":{"rendered":"

Using this simple statistics library [https:\/\/chrisbissell.wordpress.com\/2011\/05\/23\/a-simple-but-very-flexible-statistics-library-in-scala\/], a timing function [[https:\/\/stackoverflow.com\/questions\/9160001\/how-to-profile-methods-in-scala#9160068]] and Hazelcast, we can track the variability of code performance in Scala.<\/p>\n

Hazelcast setup:<\/p>\n

\nval cfg = new Config(\"concepts\")\nval hazelcastInstance = Hazelcast.newHazelcastInstance(cfg)\nval timings = hazelcastInstance.getList(\"timings\").asInstanceOf[ICollection[Map[String, Double]]]\n<\/pre>\n
Then, for timing functions:<\/p>\n
\n  var timer = 1\n\n  def time[R](name: String, block: => R): R = {\n    val maxMemory1 = runtime.maxMemory()\n    val allocatedMemory1 = runtime.totalMemory()\n    val freeMemory1 = runtime.freeMemory()\n    val totalFree1 = freeMemory1 + (maxMemory1 - allocatedMemory1)\n\n    val t0 = System.nanoTime()\n    val result = block\n    val t1 = System.nanoTime()\n\n    val maxMemory2 = runtime.maxMemory()\n    val allocatedMemory2 = runtime.totalMemory()\n    val freeMemory2 = runtime.freeMemory()\n    val totalFree2 = freeMemory2 + (maxMemory2 - allocatedMemory2)\n\n    timings.add(\n      Map[String, Double](\n        timer + \".1 \" + name + \".time\" ->1<\/a><\/sup>\n        ).map(\n          (kv) => (\n            kv._1,\n            mean(kv._2),\n            2 * stddev(kv._2)\n            )\n        ).map(\n          (kv) => kv._1 + \": \" + format.format(kv._2) + \" \u00b1 \" + format.format(kv._3)\n        ).toList.sorted\n          .mkString(\"\\n\") +\n        \"\\n********\\n\"\n    )\n\n    hazelcastInstance.shutdown()\n<\/pre>\n
This is a really easy way to track performance <\/p>\n
  1. t1 - t0) \/ 1000000000.0),\n        timer + \".2 \" + name + \".totalFree\" -> ((totalFree1 - totalFree2) \/ 1024.0 \/ 1024),\n        timer + \".3 \" + name + \".totalFree\" -> ((allocatedMemory2 - allocatedMemory1) \/ 1024.0 \/ 1024)\n      )\n    )\n\n    timer = timer + 1\n\n    result\n  }\n<\/pre>\n
    Then, right before you finish, you can compute metrics, and print them out:<\/p>\n
    \n    import scala.collection.JavaConverters._\n\n    println(\n      \"*********\\n\" +\n        \"Averages:\\n\" +\n\n        timings.asScala.flatMap(\n          (map) => map.toList\n        ).groupBy(\n          _._1\n        ).map(\n          (kv) =>\n            (kv._1, kv._2.map(_._2 [↩<\/a>]<\/span><\/li><\/ol>","protected":false},"excerpt":{"rendered":"
    Timing code, getting timing variance in Scala<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"om_disable_all_campaigns":false,"_monsterinsights_skip_tracking":false,"_monsterinsights_sitenote_active":false,"_monsterinsights_sitenote_note":"","_monsterinsights_sitenote_category":0,"footnotes":""},"categories":[4],"tags":[480],"aioseo_notices":[],"aioseo_head":"\n\t\t\n\t\n\t\n\t\n\t\n\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t