Swift vs Kotlin performance on sorting array
I tried sorting an array of size 100000000 in Swift and Kotlin and i can see a huge performance gap between them. For this number, Kotlin is almost 18 times faster than Swift (on my machine).
I recorded some results and i found that swift is faster when the size is around 10000 or less but once the number goes up, Swift becomes significantly slow as compare to Kotlin .
Code for Swift and Kotlin is below,
Swift
let n = 100000000
var arr = Array(repeating: 0, count: n)
for i in 1...n {
arr[i-1] = Int(arc4random_uniform(UInt32(n)))
}
//Record time before sort
arr.sort()
//Record time after sort
Kotlin
val n = 100000000
val arr = IntArray(n)
for (i in 1..n) {
arr[i-1] = Random().nextInt(n)
}
//Record time before sort
arr.sort()
//Record time after sort
Time recorded for both is below,
Swift
Size: 1000 Time: 0.001 sec
Size: 10000 Time: 0.009 - 0.01 sec
Size: 100000 Time: 0.122 - 0.127 sec
Size: 1000000 Time: 1.392 - 1.409 sec
Size: 10000000 Time: 16.115 - 16.569 sec
Size: 100000000 Time: 187.346 - 187.71 sec
Size: 1000000000 Waited more than 6 minutes and gave up!
Kotlin
Size: 1000 Time: 0.06 sec
Size: 10000 Time: 0.063 - 0.084 sec
Size: 100000 Time: 0.083 - 0.105 sec
Size: 1000000 Time: 0.23 - 0.501 sec
Size: 10000000 Time: 1.098 - 1.807 sec
Size: 100000000 Time: 10.759 - 11.141 sec
Size: 1000000000 Time: 124.252 - 127.54 sec
So, here you can see Swift is becoming extremely slow when size increases although it is faster when the number is small.
As MartinR highlighted to compile with release build configuration for Swift so i changed it to release and with that it seems Swift is faster for any number as shown in the below results,
Swift with Build Configuration as release
Size: 1000 Time: 0.001 sec
Size: 10000 Time: 0.001 sec
Size: 100000 Time: 0.006 - 0.007 sec
Size: 1000000 Time: 0.076 - 0.081 sec
Size: 10000000 Time: 0.891 - 0.898 sec
Size: 100000000 Time: 9.01 - 10.14 sec
Size: 1000000000 Time: 113.87 - 117.285 sec
This might be helpful for someone to check Swift results in release build configuration when doing some kind of performance benchmark as shown above the results vary significantly in debug and release configurations.
In debug and release configurations, compiler uses different optimization levels so it affects the performance. Some discussion on optimization levels can be found in this question
链接地址: http://www.djcxy.com/p/31684.html