I modified the original to be correct, theoretically when your AtomicLong
reaches the *Long.MAX_VALUE* + 1 it will shift to *Long.MIN_VALUE* making
the mod (%) function return negative which will make the application fail
due to invalid list index *-list.get(neg_index)-*
In practice that is unlikely to happen but there might be some crazy lab
running DNA sequencing so in practice infinite is now reachable, could take
hours, days or weeks, but once it happen, original *RoundRobinRountingLogic*
will fail, perf tests concluded that at least in Java 8 atomic adder + mod
is faster than CAS, but the class needs to be modified from this:
final class RoundRobinRoutingLogic extends RoutingLogic {
val next = new AtomicLong(0)
override def select(message: Any, routees: immutable.IndexedSeq[Routee]):
Routee =
if (routees.isEmpty) NoRoutee
else routees((next.getAndIncrement % routees.size).asInstanceOf[Int])
}
To this, not sure if it is going to compile, I'm not a Scala programmer,
not sure if the "asInstanceOf" is needed anymore, Integer or Long they will
eventually overflow and go to their MIN_VALUE so why bother with a Long?
It looks to me Long was put there with fear of reaching MAX_VALUE causing a
failure, well, it won't fail because overflows will just shift a bit, but
it will fail because mod(%) will return negative *-list.get(neg_index)-*,
so here is the original with Math.abs and AtomicInteger instead:
final class RoundRobinRoutingLogic extends RoutingLogic {
val next = new AtomicInteger(0)
override def select(message: Any, routees: immutable.IndexedSeq[Routee]):
Routee =
if (routees.isEmpty) NoRoutee
else routees(Math.abs(next.getAndIncrement % routees.size).asInstanceOf[
Int])
}
*JMH perf tested results:*
Benchmark Mode Cnt Score Error Units
CASvsMOD.intCAS_1 thrpt 5 76444721.566 ± 813686.931 ops/s
CASvsMOD.intCAS_4 thrpt 5 73853879.600 ± 689854.722 ops/s
CASvsMOD.intCAS_8 thrpt 5 76352315.163 ± 2300065.091 ops/s
CASvsMOD.intMOD_1 thrpt 5 131541916.778 ± 3400828.887 ops/s
CASvsMOD.intMOD_4 thrpt 5 96787272.406 ± 1540417.674 ops/s
CASvsMOD.intMOD_8 thrpt 5 96954428.161 ± 1007965.370 ops/s
CASvsMOD.longMOD_1 thrpt 5 129730846.046 ± 23522696.956 ops/s
CASvsMOD.longMOD_4 thrpt 5 96930440.731 ± 1549403.070 ops/s
CASvsMOD.longMOD_8 thrpt 5 96882920.242 ± 1342011.843 ops/s
JMH perf source code:
import org.openjdk.jmh.annotations.*;
import org.openjdk.jmh.runner.Runner;
import org.openjdk.jmh.runner.RunnerException;
import org.openjdk.jmh.runner.options.Options;
import org.openjdk.jmh.runner.options.OptionsBuilder;
import java.io.IOException;
import java.util.concurrent.atomic.AtomicInteger;
import java.util.concurrent.atomic.AtomicLong;
@State(Scope.Benchmark)
public class CASvsMOD {
private final AtomicInteger nextInt = new AtomicInteger();
private final AtomicLong nextLong = new AtomicLong();
private int nextInt(int routees) {
return nextInt.accumulateAndGet(1, (index, increment) -> ++index <
routees ? index : 0);
}
@Benchmark
public void intCAS_1() {
nextInt(1);
}
@Benchmark
public void intCAS_4() {
nextInt(4);
}
@Benchmark
public void intCAS_8() {
nextInt(8);
}
private int nextIntWithMod(int routees) {
return Math.abs(nextInt.incrementAndGet() % routees);
}
@Benchmark
public void intMOD_1() {
nextIntWithMod(1);
}
@Benchmark
public void intMOD_4() {
nextIntWithMod(4);
}
@Benchmark
public void intMOD_8() {
nextIntWithMod(8);
}
private long nextLong(int routees) {
return Math.abs(nextLong.incrementAndGet() % routees);
}
@Benchmark
public void longMOD_1() {
nextLong(1);
}
@Benchmark
public void longMOD_4() {
nextLong(4);
}
@Benchmark
public void longMOD_8() {
nextLong(8);
}
public static void main(String[] args) throws RunnerException, IOException
{
final Options opt = new OptionsBuilder().
include(CASvsMOD.class.getSimpleName()).
warmupIterations(5).
measurementIterations(5).
forks(1).
build();
new Runner(opt).run();
}
}
Why am I proposing code here?, well, I tried to import and compile Akka on
IntelliJ *-I have latest IntelliJ with Scala plugin-*, but I gave up
because several issues I didn't have the time nor patience to resolve like
SBT, so this is the best I can do.
--
>>>>>>>>>> Read the docs: http://akka.io/docs/
>>>>>>>>>> Check the FAQ:
>>>>>>>>>> http://doc.akka.io/docs/akka/current/additional/faq.html
>>>>>>>>>> Search the archives: https://groups.google.com/group/akka-user
---
You received this message because you are subscribed to the Google Groups "Akka
User List" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to akka-user+unsubscr...@googlegroups.com.
To post to this group, send email to akka-user@googlegroups.com.
Visit this group at http://groups.google.com/group/akka-user.
For more options, visit https://groups.google.com/d/optout.
