datatype: map builtin MPI datatypes to internal types

[hzhou]

Pull Request Description

Many of the MPI Datatypes are redundant, for example, MPI_INT, MPI_INT32_t, MPI_INTEGER. The new ABI proposal requires setting some of these types at runtime, for example, MPI_Abi_set_fortran_info. In this PR, we create a set of fixed-size internal datatypes and then map all external builtin types to internal ones. This allows runtime setting and resetting any builtin types. And internally, we only need to support a fixed set.

/* Internal supported types
 * The handle bit pattern follows the convention of builtin types, except
 * the last two bytes are type categories rather than indexes.
 *
 * Internally we support FIXED, SIGNED, UNSIGNED, FLOAT, and COMPLEX.
 * Other types can be added when supported, e.g. bfloat16 or logicals that
 * doesn't follow the C convention of zero/nonzero.
 */
#define MPIR_TYPE_FIXED         0x80    /* communication-only, op not supported */
#define MPIR_TYPE_SIGNED        0x81    /* including C boolean */
#define MPIR_TYPE_UNSIGNED      0x82
#define MPIR_TYPE_FLOAT         0x83    /* C native floating point */
#define MPIR_TYPE_COMPLEX       0x84    /* 2 C native floating point */
/* Pair types support communication and MPI_MINLOC and MPI_MAXLOC ops. The 1st value type
 * is SIGNED, UNSIGNED, FLOAT, COMPLEX as defined above. We add a mask for pair types.
 * E.g. MPI_FLOATINT become MPIR_TYPE_FLOAT | MPIR_TYPE_PAIR_MASK */
#define MPIR_TYPE_PAIR_MASK      0x10   /* the 2nd type is C int */
#define MPIR_TYPE_PAIR2_MASK     0x20   /* the 2nd type is the same as 1st type, e.g. MPI_2REAL */


#define MPIR_FIXED0            ((MPI_Datatype)0x4c000080)       /* 0-size, internally equivallent to MPI_DATATYPE_NULL */
#define MPIR_FIXED1            ((MPI_Datatype)0x4c000180)
#define MPIR_FIXED2            ((MPI_Datatype)0x4c000280)
#define MPIR_FIXED4            ((MPI_Datatype)0x4c000480)
#define MPIR_FIXED8            ((MPI_Datatype)0x4c000880)
#define MPIR_FIXED16           ((MPI_Datatype)0x4c001080)
#define MPIR_SIGNED_INT1       ((MPI_Datatype)0x4c000181)
#define MPIR_SIGNED_INT2       ((MPI_Datatype)0x4c000281)
#define MPIR_SIGNED_INT4       ((MPI_Datatype)0x4c000481)
#define MPIR_SIGNED_INT8       ((MPI_Datatype)0x4c000881)
#define MPIR_SIGNED_INT16      ((MPI_Datatype)0x4c001081)
#define MPIR_UNSIGNED_INT1     ((MPI_Datatype)0x4c000182)
#define MPIR_UNSIGNED_INT2     ((MPI_Datatype)0x4c000282)
#define MPIR_UNSIGNED_INT4     ((MPI_Datatype)0x4c000482)
#define MPIR_UNSIGNED_INT8     ((MPI_Datatype)0x4c000882)
#define MPIR_UNSIGNED_INT16    ((MPI_Datatype)0x4c001082)
#define MPIR_FLOAT1            ((MPI_Datatype)0x4c000183)
#define MPIR_FLOAT2            ((MPI_Datatype)0x4c000283)
#define MPIR_FLOAT4            ((MPI_Datatype)0x4c000483)
#define MPIR_FLOAT8            ((MPI_Datatype)0x4c000883)
#define MPIR_FLOAT16           ((MPI_Datatype)0x4c001083)
#define MPIR_COMPLEX2          ((MPI_Datatype)0x4c000284)
#define MPIR_COMPLEX4          ((MPI_Datatype)0x4c000484)
#define MPIR_COMPLEX8          ((MPI_Datatype)0x4c000884)
#define MPIR_COMPLEX16         ((MPI_Datatype)0x4c001084)
#define MPIR_COMPLEX32         ((MPI_Datatype)0x4c002084)


extern MPI_Datatype MPIR_Internal_types[];
#define MPIR_DATATYPE_REPLACE_BUILTIN(type) \
    do { \
        if (HANDLE_IS_BUILTIN(type)) { \
            (type) = MPIR_Internal_types[(type) & 0xff]; \
        } \
    } while (0)

Discussion

• For communication routines, such as MPI_Send, MPI_Bcast, we can replace builtin datatypes with MPIR_FIXED#, since all we care is the data size.
• For reduction routines, such as MPI_Reduce, MPI_Accumulate, we can replace builtin datatypes with internal types (but not MPIR_FIXED#)
• For datatype creation routines, I think we can replace all builtin "oldtype" with MPIR_FIXED#. We don't need worry about reduction op because we'll always rely on user op for them.
• NOTE: once we convert to internal types, we'll not be able to perform strict type matching validation, such as matching MPI_INT to MPI_FLOAT. But we don't perform such validation today anyway. It is an extra overhead that we can't afford. In principle, we could perform strict type matching under e.g. --enable-error-checking=2

[skip warnings]

Author Checklist

• Provide Description
  Particularly focus on why, not what. Reference background, issues, test failures, xfail entries, etc.
• Commits Follow Good Practice
  Commits are self-contained and do not do two things at once.
  Commit message is of the form: module: short description
  Commit message explains what's in the commit.
• Passes All Tests
  Whitespace checker. Warnings test. Additional tests via comments.
• Contribution Agreement
  For non-Argonne authors, check contribution agreement.
  If necessary, request an explicit comment from your companies PR approval manager.