Overview - 2024.2 English

matrix_mult performs a General Matrix Multiply (GEMM), taking two input matrices of configurable dimensions *and data type.

These are the templates to configure the Matrix Multiply graph class.

Parameters:

TT_DATA_A	describes the type of individual data samples input of Matrix A to the gemm function. This is a typename and must be one of the following: int16, cint16, int32, cint32, float, cfloat.
TT_DATA_B	describes the type of individual data samples input of Matrix B to the gemm function. This is a typename and must be one of the following: int16, cint16, int32, cint32, float, cfloat. The following rules apply: must be an integer type if TT_DATA_A is an integer type must be a float type if TT_DATA_A is a float type.
TP_DIM_A	is an unsigned integer which describes the number of elements along the unique dimension (rows) of Matrix A.
TP_DIM_AB	is an unsigned integer which describes the number of elements along the common dimension of Matrix A (columns) and Matrix B (rows).
TP_DIM_B	is an unsigned integer which describes the number of elements along the unique dimension (columns) of Matrix B.
TP_SHIFT	describes power of 2 shift down applied to the accumulation of product terms before each output. `TP_SHIFT` must be in the range 0 to 59 (61 for AIE1).
TP_RND	describes the selection of rounding to be applied during the shift down stage of processing. Although, TP_RND accepts unsigned integer values descriptive macros are recommended where rnd_floor = Truncate LSB, always round down (towards negative infinity). rnd_ceil = Always round up (towards positive infinity). rnd_sym_floor = Truncate LSB, always round towards 0. rnd_sym_ceil = Always round up towards infinity. rnd_pos_inf = Round halfway towards positive infinity. rnd_neg_inf = Round halfway towards negative infinity. rnd_sym_inf = Round halfway towards infinity (away from zero). rnd_sym_zero = Round halfway towards zero (away from infinity). rnd_conv_even = Round halfway towards nearest even number. rnd_conv_odd = Round halfway towards nearest odd number. No rounding is performed on ceil or floor mode variants. Other modes round to the nearest integer. They differ only in how they round for values of 0.5. Note: Rounding modes `rnd_sym_floor` and `rnd_sym_ceil` are only supported on AIE-ML device.
TP_DIM_A_LEADING	describes the scheme in which the data should be stored in memory. ROW_MAJOR = 0, COL_MAJOR = 1. Note, a COL_MAJOR matrix can be transposed to become a ROW_MAJOR matrix.
TP_DIM_B_LEADING	describes the scheme in which the data should be stored in memory. ROW_MAJOR = 0, COL_MAJOR = 1.
TP_DIM_OUT_LEADING	describes the scheme in which the data should be stored in memory. ROW_MAJOR = 0, COL_MAJOR = 1.
TP_ADD_TILING_A	describes wether or not to add an additional kernel to rearrange the matrix samples into their required position. Setting this option to 0 indicates that the re-arrangement will be done externally to the AIE matrix multiply graph.
TP_ADD_TILING_B	describes wether or not to add an additional kernel to rearrange the matrix samples into their required position. Setting this option to 0 indicates that the re-arrangement will be done externally to the AIE matrix multiply graph.
TP_ADD_DETILING_OUT	describes wether or not to add an additional kernel to rearrange the matrix samples into their required position. Setting this option to 0 indicates that the re-arrangement will be done externally to the AIE matrix multiply graph.
TP_INPUT_WINDOW_VSIZE_A	describes the number of samples in the window API used for input to Matrix A. It must be of size TP_DIM_A*TP_DIM_AB.
TP_INPUT_WINDOW_VSIZE_B	describes the number of samples in the window API used for input to Matrix B. It must be of size TP_DIM_BTP_DIM_AB Note, the output window will be of size TP_DIM_A TP_DIM_B.
TP_CASC_LEN	describes the number of AIE kernels the matrix multiplication will be divided into in series. TP_CASC_LEN splits the operation over shared dimension TP_DIM_AB, where each kernel utilizes the cascade stream to pass partial accumulation results to the next kernel. In effect, dot(A,B) + C.
TP_SAT	describes the selection of saturation to be applied during the shift down stage of processing. TP_SAT accepts unsigned integer values, where: 0: none = No saturation is performed and the value is truncated on the MSB side. 1: saturate = Default. Saturation rounds an n-bit signed value in the range [- ( 2^(n-1) ) : +2^(n-1) - 1 ]. 3: symmetric = Controls symmetric saturation. Symmetric saturation rounds an n-bit signed value in the range [- ( 2^(n-1) -1 ) : +2^(n-1) - 1 ].
TP_SSR	describes the number of kernels (or cascaded kernel chains) that will compute the matrix multiplication in parallel. Each SSR rank will receive an equal sized split (along the unique dimension) of Matrix A data. There is no splitting of the Matrix B data when TP_SSR > 1 (only split when TP_CASC_LEN > 1). The Matrix B inputs across a chain of cascaded kernels will be the same across all SSR ranks

template <
    typename TT_DATA_A,
    typename TT_DATA_B,
    unsigned int TP_DIM_A,
    unsigned int TP_DIM_AB,
    unsigned int TP_DIM_B,
    unsigned int TP_SHIFT,
    unsigned int TP_RND,
    unsigned int TP_DIM_A_LEADING = ROW_MAJOR,
    unsigned int TP_DIM_B_LEADING = COL_MAJOR,
    unsigned int TP_DIM_OUT_LEADING = ROW_MAJOR,
    unsigned int TP_ADD_TILING_A = 1,
    unsigned int TP_ADD_TILING_B = 1,
    unsigned int TP_ADD_DETILING_OUT = 1,
    unsigned int TP_INPUT_WINDOW_VSIZE_A = TP_DIM_A* TP_DIM_AB,
    unsigned int TP_INPUT_WINDOW_VSIZE_B = TP_DIM_B* TP_DIM_AB,
    unsigned int TP_CASC_LEN = 1,
    unsigned int TP_SAT = 1,
    unsigned int TP_SSR = 1
    >
class matrix_mult_graph: public graph

// fields

port <input> inA[TP_CASC_LEN *TP_SSR]
port <input> inB[TP_CASC_LEN *TP_SSR]
port <output> out[TP_SSR]
kernel m_MatmultKernels[TP_CASC_LEN *TP_SSR]
kernel untiler[TP_SSR]
kernel tilerA[TP_CASC_LEN *TP_SSR]
kernel tilerB[TP_CASC_LEN *TP_SSR]