RFR (S) CR 7177472: JSR292: MethodType interning penalizes scalability (original) (raw)

Aleksey Shipilev aleksey.shipilev at oracle.com
Tue Jun 11 16:03:02 UTC 2013


On 06/11/2013 07:43 PM, Christian Thalinger wrote:

Thanks for this but I was thinking running it with -Xprof to see whether some of the table methods show up on the profile.

I put little trust into this kind of profiling, but here's the sample output from JMH -prof stack, while running with 4 threads:

Before:

Iteration 5 (20s in 4 threads): 380.982 msec/op Stack | 24.8% RUNNABLE jdk.nashorn.internal.runtime.regexp.joni.ByteCodeMachine.matchAt | 6.5% RUNNABLE jdk.nashorn.internal.runtime.regexp.joni.SearchAlgorithm$3.search | 5.2% RUNNABLE java.lang.Class.getClassLoader0 | 3.3% RUNNABLE java.lang.ThreadLocal.get | 3.0% RUNNABLE jdk.nashorn.internal.runtime.regexp.JoniRegExp$JoniMatcher. | 2.9% RUNNABLE jdk.nashorn.internal.objects.NativeRegExp.replace | 2.4% RUNNABLE jdk.nashorn.internal.runtime.regexp.joni.Matcher.forwardSearchRange | 2.3% RUNNABLE jdk.nashorn.internal.objects.NativeRegExp.exec | 2.1% RUNNABLE jdk.nashorn.internal.objects.NativeRegExp.groups | 1.9% RUNNABLE jdk.nashorn.internal.objects.NativeString.replace | 45.6% (other) |

After:

Iteration 5 (20s in 4 threads): 358.121 msec/op Stack | 24.7% RUNNABLE jdk.nashorn.internal.runtime.regexp.joni.ByteCodeMachine.matchAt | 5.7% RUNNABLE java.lang.Class.getClassLoader0 | 5.3% RUNNABLE jdk.nashorn.internal.runtime.regexp.joni.SearchAlgorithm$3.search | 3.5% RUNNABLE java.lang.ThreadLocal.get | 3.4% RUNNABLE jdk.nashorn.internal.runtime.regexp.JoniRegExp$JoniMatcher. | 2.7% RUNNABLE jdk.nashorn.internal.objects.NativeRegExp.replace | 2.3% RUNNABLE jdk.nashorn.internal.runtime.regexp.joni.Matcher.forwardSearchRange | 1.8% RUNNABLE jdk.nashorn.internal.objects.NativeRegExp.exec | 1.7% RUNNABLE jdk.nashorn.internal.objects.NativeString.replace | 1.5% RUNNABLE jdk.nashorn.internal.runtime.regexp.joni.StackMachine.pop | 47.3% (other) |

-Xprof shows the similar data.

-Aleksey.



More information about the core-libs-dev mailing list