Dear SwatchPuppy,

You are probably looking for the HyPhy MPI version (which runs in a distributed environment, i.e. clusters).

Depending on your system (I am not familiar with openmosix, we use Scyld Beowulf and IBM's LoadLeveler), you should be able to invoke an MPI run of HyPhy on multiple processors. Their is a test file (MPITest.bf) in the BatchFiles directory that you can use to see if HyPhy is able to use multiple processors, by calling something like:

$mpirun -np xx ./HYPHYMPI BASEPATH=<insert path to HyPhy here> BatchFiles/MPITest.bf

FEL analyses will take advantage of MPI builds, so do REL analyses (with the MPIOPTIMIZER flag passed to HyPhy at start time).

Take a look at some of the other messages on the forum dealing with the specifics of the analyses.

Cheers,
Sergei

Dear SwatchPuppy,

I didn't realize that openMOSIX was an SMP emulator. Try compiling with MP2 option and then calling

$HYPHYMP CPU=n

where 'n' is the number of threads you want to spawn. Most likelihood optimizations will try to take advantage of mutliple nodes, but the performance hits may be severe for network communications in openMOSIX because granularity of threads in HyPhy is fairly high (each thread does a little work).

Cheers,
Sergei

Sergei,

I was trying to start an MPI cluster with LAM/MPI, it worked on compiling when i changed the gcc and g++ compilers to mpicc and mpic++, as well as removed the -lmpi flag on MPI link from the build.sh and gtk_build.sh.

The HYPHYMPI is working perfectly, but the GTK version is giving me this output:

One of the processes started by mpirun has exited with a nonzero exit
code.  This typically indicates that the process finished in error.
If your process did not finish in error, be sure to include a "return
0" or "exit(0)" in your C code before exiting the application.

PID 15612 failed on node n0 (192.168.0.13) due to signal 8.


Even if i only boot one node (the master). Can you give-me a hint?

Thanks in advance

Carlos Família