I just re-ran an extensive Monte Carlo involving a lot of minimization, kernel regression, mixed oct and .m files, using the new MPI extensions, and Octave 3.2.3. On 4 machines with 32 cores, it took 6.9 hours. The same code using Octave 3.0.x and MPITB on the same hardware took 15.6 hours. I believe that the difference in times is mostly due to improvements in Octave, because the code is written to keep MPI traffic low. Nevertheless, the replication indicates that there are no serious problems with the new MPI bindings (the functions tested are MPI_Send, MPI_Recv and MPI_Iprobe). So, thanks to you both, and thanks to everyone else responsible for the speed improvements in Octave since 3.0. Michael
------------------------------------------------------------------------------ Return on Information: Google Enterprise Search pays you back Get the facts. http://p.sf.net/sfu/google-dev2dev _______________________________________________ Octave-dev mailing list Octave-dev@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/octave-dev
