Charm++ parallel programming system

[Takuro Udagawa <udagawa.t.aa AT m.titech.ac.jp>]

Dear All,

I want to use TAU with NAMD2.9 smp mode.
I read http://www.nic.uoregon.edu/tau-wiki/Guide:NAMDTAU and was able to pass build and make, and was able to run NAMD "no smp" mode with TAU.
However when I tried smp mode NAMD failed during startup.
Does charm++ smp require another build procedure?
Does anyone have any idea or suggestion about the issue?
My configurations and error are as follows:

TAU install

./configure -cuda=/opt/cuda/4.1 -mpi -mpiinc=/usr/apps/openmpi/1.4.2/gcc/include -mpilib=/usr/apps/openmpi/1.4.2/gcc/lib -pdt=$HOME/TAU/pdtoolkit-3.18.1
make install

Charm++ build in NAMD_2.9_Source/charm-6.4.0 directory.

./build charm++ mpi-linux-x86_64 mpicxx smp -j16 -g -O0

Tau build in NAMD_2.9_Source/charm-6.4.0 directory.

./build Tau mpi-linux-x86_64 mpicxx smp -j16 --tau-makefile=$HOME/TAU/tau-2.22-p1/x86_64/lib/Makefile.tau-mpi-cupti-pdt

NAMD configure in NAMD_2.9_Source direcotry

./config Linux-x86_64-g++ --charm-arch mpi-linux-x86_64-smp-mpicxx --with-cuda --cuda-prefix /opt/cuda/4.1

And edit arch/Linux-x86_64-g++.arch adding

CHARMOPTS = -tracemode Tau

Then make NAMD

cd Linux-x86_64-g++
make

Finally run NAMD like

charmrun tau_exec -T mpi,cupti,pdt -cupti namd2 +p12 +ppn11 +isomalloc_sync +idlepoll +devices 0  apoa1.namd > output.dat

Then NAMD failed during startup.
The log file was as follows:

Running on 1 processors:  tau_exec -T mpi,cupti,pdt -cupti namd2 +ppn 11  +isomalloc_sync +idlepoll +devices 0 apoa1.namd
mpirun -np 1 -machinefile /var/spool/PBS/aux/428066.t2zpbs05 tau_exec -T mpi,cupti,pdt -cupti namd2 +ppn 11 +isomalloc_sync +idlepoll +devices 0 apoa1.namd
Charm++> Running on MPI version: 2.1
Charm++> level of thread support used: MPI_THREAD_SINGLE (desired: MPI_THREAD_FUNNELED)
Charm++> Running on SMP mode, 11 worker threads per process
Charm++> The comm. thread both sends and receives messages
Converse/Charm++ Commit ID: v6.4.0-beta1-0-g5776d21
Warning> Randomization of stack pointer is turned on in kernel.
[1] Stack Traceback:
  [1:0] +0x329e0  [0x2b819be669e0]
  [1:1] _Z21TauProfiler_PurgeDatai+0x17d  [0xddc77d]
  [1:2] _ZN8TraceTau10traceBeginEv+0x1f  [0xc8eb9b]
  [1:3] _ZN10TraceArray10traceBeginEv+0xa5  [0xb5eebf]
  [1:4] traceBegin+0xbc  [0xb5ef94]
  [1:5]   [0xb6060f]
  [1:6] traceInit+0x15  [0xb6062d]
  [1:7] ConverseCommonInit+0x473  [0xc4c11b]
  [1:8]   [0xc41ad9]
  [1:9]   [0xc40326]
  [1:10] +0x75f0  [0x2b81994645f0]
  [1:11] clone+0x6d  [0x2b819bf0887d

Sincerely yours,

Takuro Udagawa

Department of Computer Science
Tokyo Institute of Technology