Inlining methods

[Juan José Fumero]

Hello,
Ok, I have just realized that I had  -XX:-BootstrapGraal enabled. The 
output now is very verbose.

....
Testing Phase.OptimisticInline
             [thread:11] scope: 
Compiling.GraalCompiler.FrontEnd.HighTier.Inlining.Inlining
               [thread:11] scope: 
Compiling.GraalCompiler.FrontEnd.HighTier.Inlining.Inlining.InliningDecisions
               inlining BitSet.initWords at 4: exact 
java.util.BitSet.wordIndex(int):int: trivial (relevance=1.000000, 
probability=1.000000, bonus=1.000000, nodes=6)
   55835   -1             java.util.RegularEnumSet::add (9 bytes)   made 
not entrant
             [thread:11] scope: 
Compiling.GraalCompiler.FrontEnd.HighTier.Inlining.Inlining
               [thread:11] scope: 
Compiling.GraalCompiler.FrontEnd.HighTier.Inlining.Inlining.InliningDecisions
   55835   -1             java.util.RegularEnumSet::add (40 bytes)   
made not entrant
               inlining BitSet.<init>@47: exact 
java.util.BitSet.initWords(int):void: relevance-based 
(relevance=1.000000, probability=1.000000, bonus=1.000000, nodes=13 <= 
300.000000)
true||
....

Thanks
Juan José Fumero


2013/10/27 Christian Thalinger <christian.thalinger at oracle.com 
<mailto:christian.thalinger at oracle.com>>


    On Oct 25, 2013, at 9:37 AM, Juan José Fumero Alfonso
    <jjfumero at gmail.com <mailto:jjfumero at gmail.com>> wrote:

     > Hi Gary,
     >   I get the same output with the -XX:+UnlockDiagnosticVMOptions, only
     > this:
     >
     >      62    1     n
     >
    com.oracle.graal.hotspot.bridge.CompilerToVMImpl::executeCompiledMethodIntrinsic
     > (native)   (static)
     >    168    2     n       java.lang.System::arraycopy (native)  
    (static)

    This is all the output you get?  How are you running your test?

     >
     > The option -G:+InlineEverything does not exist in my version.
     >
     > Thanks
     > Juan José Fumero
     >
     >
     >
     >
     > 2013/10/25 Frost, Gary <Gary.Frost at amd.com
    <mailto:Gary.Frost at amd.com>>
     >
     >> WRT  -XX:+PrintInlining  I think you may also need
     >> -XX:+UnlockDiagnosticVMOptions if you are using a product JVM
     >> build.
     >>
     >> Gary
     >>
     >>
     >>
     >> -----Original Message-----
     >> From: graal-dev-bounces at openjdk.java.net
    <mailto:graal-dev-bounces at openjdk.java.net> [mailto:
     >> graal-dev-bounces at openjdk.java.net
    <mailto:graal-dev-bounces at openjdk.java.net>] On Behalf Of Juan José
    Fumero Alfonso
     >> Sent: Friday, October 25, 2013 10:37 AM
     >> To: graal-dev at openjdk.java.net <mailto:graal-dev at openjdk.java.net>
     >> Subject: Inlining methods
     >>
     >> Hello,
     >> I am working with the OpenCL Component of Graal
     >> (com.edinburgh.parallel.opencl). This version is not public yet
    but It uses
     >> Graal to get the user code and generates OpenCL kernels. I have
    to create
     >> an Inlining optimization for my own code using Graal.  Now I
    have something
     >> like that:
     >>
     >>        GraalCodeCacheProvider runtime =
     >> Graal.getRequiredCapability(GraalCodeCacheProvider.class);
     >>        ResolvedJavaMethod javaMethod =
    runtime.lookupJavaMethod(method);
     >>
     >> *        OptimisticOptimizations oo = new
     >> OptimisticOptimizations(javaMethod);*
     >>
     >>
     >> oo.add(OptimisticOptimizations.Optimization.UseTypeCheckedInlining);
     >>
     >> The package is com.oracle.graal.phases;
     >>
     >> How can I ensure that the method is inline at the end?  I am
    using the
     >> -XX:+PrintCompilation options but there is not much information.
    Also I
     >> tried with the -XX:+PrintInlining but it seems this options does
    not exist.
     >> Is this the proper way to get it?
     >>
     >>
     >> Thank you
     >> Juan José Fumero
     >>
     >>
     >>