Q: We would like to install MPP version of LSDYNA software on our computer here. Can you please guide us with regards to the steps involved in installation and setup for the MPP version of DYNA? A: The details will vary somewhat depending on your computing environment. Start by reading README.first, located in the respective folder for the LS-DYNA MPP release of interest, e.g., http://ftp.lstc.com/user/mpp-dyna/R7.1.2 . One thing for certain is that you'll have to install an MPI (Message Passing Interface). For your convenience, various MPIs are located in http://ftp.lstc.com/user/mpp-dyna/Tools . Some guidance for installation of the MPI is provided in the README.howtoexec file which is in the same folder as the README.first file. Beyond this, we don't offer support for installation of the MPI. _________________________________________________________ Q: What’s the difference between platformmpi and platformmpi_sharelib? A: The only reason you might be interested in files containing "sharelib" would be if you're creating user-defined subroutines, e.g., as for a user-defined material model (*MAT_041 to *MAT_050). For more infor, see ../doc/mpp.dynamic ___________________________________________________________ Q: How do I submit an MPP LS-DYNA run? A: A typical execution line looks something like this... mpirun -np 8 /home/bin/ls-dyna_mpp_d_r8_0_0_95359_x64_redhat54_ifort131_sse2_platformmpi i=input.k memory=100m memory2=50m See the Appendix on MPP in Volume I of the LS-DYNA User's Manual. _____________________________________________________________ Q: Will my input deck need to be modified? A: Maybe. Contact routines in MPP are not identical to those in SMP, and so results can vary. Note the "MPP" option for *CONTACT in the User's Manual for input variables specific to MPP. _____________________________________________________________ Q: What other complications might there be when running MPP. A: How the model gets decomposed to the various processors can affect results to some degree. The default decomposition will normally offer the best load balance between processors. User's can visualize the decomposition by adding *control_mpp_decomposition_show to the input deck. User's can modify the decomposition by either i (a) adding *control_mpp_decomposition_... commands(s) to the input deck, or (b) creating a "pfile" containing decomposition directives as explained in the MPP appendix in the User's Manual and appending "p=pfile" to the execution line. ______________________________________________________________ Q: I ran MPP LS-DYNA but didn't get any so-called ASCII output files, except for glstat. Where's the missing data? A: In the binout* files. For details on how to postprocess the data, see http://ftp.lstc.com/anonymous/outgoing/support/FAQ/ASCII_output_for_MPP_via_binout . _________________________________________________________________________________