Github user gengliangwang commented on a diff in the pull request:
https://github.com/apache/spark/pull/22663#discussion_r223221336
--- Diff: core/benchmarks/KryoBenchmark-results.txt ---
@@ -0,0 +1,29 @@
+================================================================================================
+Benchmark Kryo Unsafe vs safe Serialization
+================================================================================================
+
+Java HotSpot(TM) 64-Bit Server VM 1.8.0_131-b11 on Mac OS X 10.13.6
+Intel(R) Core(TM) i7-6920HQ CPU @ 2.90GHz
+
+Benchmark Kryo Unsafe vs safe Serialization: Best/Avg Time(ms)
Rate(M/s) Per Row(ns) Relative
+------------------------------------------------------------------------------------------------
+basicTypes: Int with unsafe:true 27 / 31 117.9
8.5 1.0X
+basicTypes: Long with unsafe:true 34 / 37 93.8
10.7 0.8X
+basicTypes: Float with unsafe:true 29 / 32 107.6
9.3 0.9X
+basicTypes: Double with unsafe:true 31 / 32 101.3
9.9 0.9X
+Array: Int with unsafe:true 3 / 3 1130.4
0.9 9.6X
--- End diff --
Hi @dongjoon-hyun , thanks for pointing it out.
This is because the length of the arrays is random
https://github.com/apache/spark/blob/669ade3a8eed0016b5ece57d776cea0616417088/core/src/test/scala/org/apache/spark/serializer/KryoBenchmark.scala#L90-L91
> The benchmark result is also reason compared to the original one.
I will fix the randomness.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
