« Return to Thread: [Fwd: [drlvm]several performance optimizaion can improve situation on Dacapo.jython bench]

Jython performance

by Ian Rogers (nabble) :: Rate this Message:

Reply to Author | View in Thread

Hi,

as noted before it'd be good to get Jython performance up. Here's a
performance break down of where we're losing time with Jython based on
the AOS profiling of things making up >1% of the execution time:


Class library related (3.37%):
97.0 (1.0420023633043292%) Ljava/util/Hashtable;.put
(Ljava/lang/Object;Ljava/lang/Object;)Ljava/lang/Object;   BOOT
109.0 (1.1709098721667204%) Ljava/lang/String;.toLowerCase
(Ljava/util/Locale;)Ljava/lang/String;       BOOT
205.0 (2.2021699430658503%) Ljava/lang/String;.toUpperCase
(Ljava/util/Locale;)Ljava/lang/String;       BOOT

Opt compiler related (3.70%):
344.0 (3.6953485873885485%)
Lorg/jikesrvm/compilers/opt/OPT_NormalBURS;.reachableChild
(Lorg/jikesrvm/ArchitectureSpecific$OPT_BURS_TreeNode;Lorg/jikesrvm/compilers/opt/OPT_SpaceEffGraphNode;I)Z      
BOOT

Jikes RVM runtime related (2.49%):
232.0 (2.4922118380062304%)
Lorg/jikesrvm/runtime/VM_StackTrace;.walkFrames (Z)I        BOOT

Memory manager related (13.45%):
178.0 (1.91212804812547%) Ljava/lang/ref/ReferenceQueue;.dequeue
()Ljava/lang/ref/Reference;    BOOT
209.0 (2.245139112686647%) Ljava/lang/ThreadLocal;.get
()Ljava/lang/Object;     BOOT
433.0 (4.651412611451284%) Lgnu/java/util/WeakIdentityHashMap;.get
(Ljava/lang/Object;)Ljava/lang/Object;       BOOT
433.0 (4.651412611451284%)
Lgnu/java/util/WeakIdentityHashMap$WeakBucket;.getEntry
()Lgnu/java/util/WeakIdentityHashMap$WeakBucket$WeakEntry;   BOOT

Misc (5.93%):
552.0 (5.929745407669997%)
Lorg/jikesrvm/runtime/VM_Runtime;.resolvedNewArray
(III[Ljava/lang/Object;IIII)Ljava/lang/Object;    BOOT

In total this is about 28.94% of execution time. A compilation system
report for ~69second run looks like:

                Compilation Subsystem Report
Comp    #Meths  Time    bcb/ms  mcb/bcb MCKB    BCKB
JNI     23      0.71    NA      NA      5.5     NA
Base    3334    350.21  540.42  6.29    3056.3  485.8
Opt     557     14050.64        14.14   5.28    552.9   104.7

with BURS accounting for 73.15% of compilation time.

Ian


-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >>  http://get.splunk.com/
_______________________________________________
Jikesrvm-core mailing list
Jikesrvm-core@...
https://lists.sourceforge.net/lists/listinfo/jikesrvm-core

 « Return to Thread: [Fwd: [drlvm]several performance optimizaion can improve situation on Dacapo.jython bench]