Please help! 2-Parallellas Cluster, 32 e-cores Hello world

Discussion about Parallella (and Epiphany) Software Development

Moderators: amylaar, jeremybennett, simoncook

Please help! 2-Parallellas Cluster, 32 e-cores Hello world

Postby claudio4parallella » Fri Sep 01, 2017 4:36 pm

Hi,

please suggestions, as well as my attempts till now were unsuccesfull.

- I've two Parallellas "parabuntu-2016.11.1-hdmi-z7010.img"
- I've ssh passwordless within the two Parallellas
- I've copied /epiphany-examples/apps/hello-world on both
- I've modified ./build.sh on both substituting <<gcc>> with <<mpicc>>
- I've modified ./run.sh on my master node substituting <<./hello-world.elf>> with <<mpiexec -machinefile machine_file -n 4 ./hello-world.elf>>
- I've the same path and folder <<hello-world>> copied on both
- I've prepared <<machine_file>> with the two IPs
- Simple executable will run on the 2-Parallelas 2-ARM-Cores with << mpiexec -machinefile machine_file -n 4 ./example>> (the cluster is ok!)

--> but when I run ./run.sh on my master node, or << parallella$ mpiexec -machinefile machine_file -n 4 ./hello-world.elf >>

I got the following exception:

Code: Select all
-------------------------------------------------------
Primary job  terminated normally, but 1 process returned
a non-zero exit code.. Per user-direction, the job has been aborted.
-------------------------------------------------------
./hello_world.elf: error while loading shared libraries: libe-hal.so.0: cannot open shared object file: No such file or directory
./hello_world.elf: error while loading shared libraries: libe-hal.so.0: cannot open shared object file: No such file or directory
--------------------------------------------------------------------------
mpiexec detected that one or more processes exited with non-zero status, thus causing
the job to be terminated. The first process to do so was:

  Process name: [[58863,1],0]
  Exit code:    127
--------------------------------------------------------------------------


Please anybody could help me?

My aim is to run on my 2-nodes cluster a program (OK it works) that by each one on its own Parallella run the 16 Cores, in order to have the 32 Cores activated in parallel.

Where am I wrong ?, ever if it's possible to have what I'm looking for ......

The following <<./run.sh>> does work on a single ARM of a single Parallella !
Code: Select all
mpiexec -n 1 ./hello_world.elf


As well as the following <<./run.sh>> does work on the ARM Cores of a single Parallella !
Code: Select all
mpiexec -n 2 ./hello_world.elf



Somewhere I read:

<< There is also the Brown Deer Technology version of MPI for programming the Epiphany >>

back to coprthr 1.6 or 2.0 issue ??
claudio4parallella
 
Posts: 68
Joined: Thu Aug 10, 2017 3:48 pm

Re: Please help! 2-Parallellas Cluster, 32 e-cores Hello wor

Postby claudio4parallella » Sat Sep 02, 2017 7:03 am

Here I've found an << how to >>, even if not recent.

https://sites.google.com/a/complexsys.info/scattershot/home/parallella-raspberry-pi-cluster-computing

That is requiring installing both openMPI, from native adapteva image, and MPICH2, from scratch.

I'm assuming that within the last image of parabuntu-2016 both are already installed, or only openMPI. MPICH2 I do not understand if it is required and involved in the error I've detected.

Thanks for any help, regards
claudio4parallella
 
Posts: 68
Joined: Thu Aug 10, 2017 3:48 pm


Return to Programming Q & A

Who is online

Users browsing this forum: No registered users and 4 guests

cron