[ Performance Trace Data Visualizer] Mapping instruction address back to Java source?

Hi.

I am currently learning how to profile a java application with PEX and the Performance Trace Data Visualizer tool.

I have managed to collect data on a java program using

java -Dos400.enbpfrcol=1 -Djava.compiler=jitc -cp scimark2lib.jar jnt.scimark2.commandline

(I'm using the http://math.nist.gov/scimark2/ benchmark to do a lot of work and give a number back. Lot of CPU, almost no I/O, perfect :) )

using

ADDPEXDFN DFN(TPROF1MS16) TYPE(*TRACE) JOB((*ALL)) MAXSTG(1000000) INTERVAL(1) TRCTYPE(*SLTEVT) SLTEVT(*YES) BASEVT((*PMCO *NONE *FORMAT2))

and visualize it with PDTV. If I understand correctly I can now go in Hierachical Tprof and get a hit count on each method hit by the probe every 1 ms (due to the INTERVAL), and I can then click on a given method and get an analysis of the data collected for that method.

Here the "Hit count by address" is very interesting to me as it shows where in the actual method that each probe found it. Unfortunately these addresses are so raw that I cannot immediately see what java statemetns inside the method they correspond to. An example:

Instruction address Hit count Statement number
FF81D30F38007050 2 0
FF81D30F38007090 2 0
FF81D30F3800709C 2 0
FF81D30F380070A0 2 0
FF81D30F380070A8 4 0
FF81D30F380070B0 2 0
FF81D30F380070B4 12 0
FF81D30F380070B8 4 0
FF81D30F380070BC 8 0
FF81D30F380070C0 14 0
FF81D30F380070C8 14 0
FF81D30F380070CC 12 0
FF81D30F380070D0 2 0
FF81D30F380070D4 8 0
FF81D30F380070DC 2 0
FF81D30F380070E8 2 0
FF81D30F380070F8 4 0
FF81D30F380070FC 4 0
FF81D30F38007104 4 0
FF81D30F38007110 308 0
FF81D30F38007114 274 0
FF81D30F38007118 1.556 0
FF81D30F3800711C 336 0
FF81D30F38007120 2.728 0
FF81D30F38007124 1.770 0
FF81D30F38007128 54 0
FF81D30F3800712C 738 0
FF81D30F38007130 1.208 0
FF81D30F38007134 1.112 0
FF81D30F38007138 290 0
FF81D30F3800713C 46 0
FF81D30F38007140 572 0
FF81D30F38007144 2 0
FF81D30F38007150 2 0

Is there a simple way to map these numbers back to the source file? Have I overlooked something coming from the Java world?

/Thorbjørn