Trying to prove the expensive use of mailboxSize in a feature of Akka-2.0, in this case, showing the bottleneck on the receive timeout feature.

Uncomment lines 43 to 47 and comment line 49 to perform the benchmark differences.

The benchmark was performed in a machine with 24 AMD Opteron 6174's and 66111860 kB of RAM

Benchmark results: (jvm parameters: -Xmx4096m -Xms4096m)

a)
=-= Using receive timeout =-=
Workers: 8
Messages: 10000
Hops: 100
Repetitions: 5

Average execution time = 481584 ms
............................................................

b)
=-= Without receive timeout =-=
Workers: 8
Messages: 10000
Hops: 100
Repetitions: 5

Average execution time = 2772 ms

Difference = 481584 / 2772 ~ 173.731

These results show that when enabling the receive timeout feature on Akka, the performance degrades as shown on benchmark a).

Adding the changes from commit https://github.com/jboner/akka/pull/81 fixes this problem, which means that when enabling the ReceiveTimeout feature, performance will no longer degrade. This is not shown in this benchmark.