Charm++ parallel programming system

[Cristian RUIZ <cristian.ruiz AT inria.fr>]

Hello,

I compiled charm++ with support for fast message-logging protocol:

./build charm++ net-linux-x86_64 mlogft


And then I tried to execute the example 'jacobi3d' located in:

net-linux-x86_64-mlogft/tests/charm++/jacobi3d/

using the following line:

./charmrun ++p 16 ++nodelist /root/nodelist ./jacobi3d 512 512 256 64 64 64

arg_read_pes 16 arg_requested_pes 16
Charmrun> started all node programs in 6.788 seconds.
Converse/Charm++ Commit ID:
Charm++> scheduler running in netpoll mode.
CharmLB> Load balancer assumes all CPUs are same.
[0] Fast Message Logging Support

STENCIL COMPUTATION WITH BARRIERS
Running Jacobi on 16 processors with (8, 8, 4) chares
Array Dimensions: 512 512 256
Block Dimensions: 64 64 64
Start of iteration 0 at 0.065944
Start of iteration 10 at 2.821627
Start of iteration 20 at 5.352906
Start of iteration 30 at 7.841674
Start of iteration 40 at 10.262645
Start of iteration 50 at 12.006342
Start of iteration 60 at 14.013436
Start of iteration 70 at 16.391883
Start of iteration 80 at 18.997955
Start of iteration 90 at 21.442106
Start of iteration 100 at 23.218660
Start of iteration 110 at 24.616721
Start of iteration 120 at 26.255881
Start of iteration 130 at 28.471377
Start of iteration 140 at 31.068913
Start of iteration 150 at 33.561893
Start of iteration 160 at 36.025725
Start of iteration 170 at 38.209047
Start of iteration 180 at 40.535497
Start of iteration 190 at 43.061998

        

The application hangs when performing the first checkpoint.
I dont have any problem when I used 'syncft'
I have this problem with version 6.6.0 of charm and with the current 
version from the git repository.

Any ideas about what I'm doing wrong? Should I add more options when 
compiling?

Thank you