Hi David, If you have multiple sockets on your compute node (or multiple compute node sharing a parallel file system) you might also give a few tries with the MPI support enabled: Sometimes performances are better using a number (N_PROCS) of mpi processes per socket running each one over the N_CPUS of the socket (or the node). You can do this by running your application using "mpirun --bind-to socket -nb $N_PROCS otbcli_YourApp" (or similar to bind process to socket) and ITK_GLOBAL_DEFAULT_NUMBER_OF_THREADS equals to $N_CPU
Keep us updated if you have multiple sockets (or nodes) and the faith to benchmark this ;) Regards, Rémi Le mardi 31 octobre 2017 16:51:40 UTC+1, davidmich...@gmail.com a écrit : > > Hi, > > I'm pretty new to OTB. > > I'm currently building a platform for the rapid processing of Sentinel > products on HPC. During tests, I have seen that the otbcli_* commands are > using all available processors on a compute node. > > My question is: > > Is there a way to limit / set the number of processes / threads used by > the OTB programs ? > > Best regards, > David > -- -- Check the OTB FAQ at http://www.orfeo-toolbox.org/FAQ.html You received this message because you are subscribed to the Google Groups "otb-users" group. To post to this group, send email to otb-users@googlegroups.com To unsubscribe from this group, send email to otb-users+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/otb-users?hl=en --- You received this message because you are subscribed to the Google Groups "otb-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to otb-users+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
