All SMP versions of LS-DYNA allow the user to set the number of
processors at the beginning of the job.  This means that if you
choose to run on 4 processors the entire run will use 4 processors.  
The user DOES NOT need to set the number of processors
by an environment variable, such as KMP_PARALLEL, OMP_SET_NUM_THREADS
or NCPUS.

For example, use one of following two methods to run an SMP version
of LS-DYNA job with 4 CPUs:

  a. Define NCPU=4 in the LS-DYNA input deck

     Set NCPU of *CONTROL_PARALLEL for KEYWORD input decks or
          set Column 1-5 of Contrl Card 16 for NON-KEYWORD(STRUCTURED)
          input decks.

  b. Use the command line option NCPU. For example,

     ls-dyna ncpu=4 i=(input file)
                or
             ls-dyna ncpu=-4 para=1 i=(input file)

     The command line option NCPU will override the number of CPU's
      defined in the input file. A positive number will set
     the consistent result flag to 2 (off), the default for 
     LS-DYNA 940.2 and subsequent versions, and a negative number 
     will set the consistent result flag to 1 (on). 

     you will get better performance by setting the consistent 
     result flag to 2.

     Also, the new command line option PARA will override the parallel
     force assembly flag in the input file.