VEC_$IREC316_i                    Domain/OS                     VEC_$IREC316_i


NAME
     vec_$irec316_i - calculate third-order linear recurrence of 16-bit
     integer vector

SYNOPSIS (C)
     #include <apollo/base.h>
     #include <apollo/vec.h>

     void vec_$irec316_i(
          short int *start_vec,
          long int &start_inc,
          short int *vec_2,
          long int &vec2_inc,
          short int *vec_3,
          long int &vec3_inc,
          short int *vec_4,
          long int &vec4_inc,
          long int &count,
          short int *result_vec,
          long int &result_inc)

SYNOPSIS (Pascal)
     %include '/sys/ins/base.ins.pas';
     %include '/sys/ins/vec.ins.pas';

     procedure vec_$irec316_i(
          in start_vec: univ vec_$integer16_vector;
          in start_inc: integer32;
          in vec_2: univ vec_$integer16_vector;
          in vec2_inc: integer32;
          in vec_3: univ vec_$integer16_vector;
          in vec3_inc: integer32;
          in vec_4: univ vec_$integer16_vector;
          in vec4_inc: integer32;
          in count: integer32;
          var result_vec: univ vec_$integer16_vector;
          in result_inc: integer32);

SYNOPSIS (FORTRAN)
     %include '/sys/ins/base.ins.ftn'
     %include '/sys/ins/vec.ins.ftn'

           parameter (nvec = 10)

           integer*2 start_vec(nvec), result_vec(nvec), vec_2(nvec), vec_3(nvec), vec_4(nvec)
           integer*4 count
           integer*4 start_inc, vec2_inc, vec3_inc_2, vec4_inc, result_inc

           call vec_$irec316_i(start_vec, start_inc, vec_2, vec2_inc, vec_3, vec3_inc,
          & vec_4, vec4_inc, count, result_vec, result_inc)

DESCRIPTION
     Vec_$irec316_i calculates a third-order linear recurrence based on
     start_vec, vec_2, vec_3, and vec_4.  It differs from vec_$rec3_i in that
     the vectors being handled contain 16-bit integers.

     This call, like all vec_$ calls ending in _i, takes a set of extra stride
     arguments, one for every vector argument.  The stride arguments determine
     which elements in the array are actually processed.  For instance, if the
     stride for a particular array is set to 3, every third element in the
     array will be processed by the routine.  The stride arguments need not be
     identical.  If all stride arguments are set to 1, this call behaves
     exactly like the version without the _i in its name.

     This routine reads the first three entries in the array result_vec; it
     then writes count entries into result_vec based on the following formula:

     Initialize the counter variables J, K, L, M, and N to the low indices of
     the arrays start_vec, vec_2, vec_3, vec_4, and result_vec.  In Fortran,
     the low index will be 1;  in C, it will be 0; in Pascal, it varies
     depending on the declaration.
     Execute the following equations count times:
     result_vec(N+3) = start_vec(J) + (vec_2(K) x result_vec(N+2))
                        + (vec_3(L) x result_vec(N+1))
                        + (vec_4(M) x result_vec(N))
     J = J + start_inc
     K = K + vec2_inc
     L = L + vec3_inc
     M = M + vec4_inc
     N = N + result_inc

     Notice that result_vec is used both for input and output, and that it
     must be large enough to hold count + 4 entries.

     start_vec
          An input vector.

     start_inc
          The stride for start_vec.

     vec_2
          An input vector.

     vec2_inc
          The stride for vec2_vec.

     vec_3
          An input vector.

     vec3_inc
          The stride for vec3_vec.

     vec_4
          An input vector.

     vec4_inc
          The stride for vec4_vec.

     count
          The number of elements to be operated on.

     result_vec
          The vector containing three input variables and the output from the
          recurrence calculations.

     result_inc
          The stride for result_vec.

NOTES
     When vec_$irec316_i is used to operate on matrixes in C and Pascal,
     start_vec, vec_2, vec_3, and result_vec are row vectors; in FORTRAN, they
     are column vectors.

     As in all the vec_$ calls,  the result array must not overlap any of the
     input arrays; the result array may be identical to an input, but must not
     contain any subset of an input.   Because of pipelining, using overlap-
     ping input and output arrays may cause incorrect results.

     vec_$irec316_i, like all 16-bit integer routines, performs poorly when
     compared to the 32-bit integer routines.  Its use should be avoided wher-
     ever possible, especially on high-performance workstations.

SEE ALSO
     vec_$rec3, vec_$rec3c, vec_$rec3_i, vec_$drec3, vec_$drec3_i, vec_$irec3,
     vec_$irec3_i, vec_$irec316_i.