[hipblaslt] Docs: Add new FP6/FP4 data types to reference materials

projects/hipblaslt/docs/reference/api-reference.rst

@@ -165,6 +165,24 @@ To use special data ordering for ``HIPBLASLT_ORDER_COL16_4R8`` and ``HIPBLASLT_O
 |  BF16 | BF16  |  BF16 | BF16  |  T   |   N   |  HIPBLASLT_ORDER_COL16_4R8  |  HIPBLASLT_ORDER_COL        | HIPBLASLT_ORDER_COL | HIPBLASLT_ORDER_COL |
 +-------+-------+-------+-------+------+-------+-----------------------------+-----------------------------+---------------------+---------------------+
 
+There are restrictions on the supported problem sizes for the ``HIP_R_4F_E2M1``, ``HIP_R_6F_E2M3``, ``HIP_R_6F_E3M2``,
+``HIP_R_8F_E4M3``, and ``HIP_R_8F_E5M2`` data types.
+When ``HIPBLASLT_MATMUL_DESC_A_SCALE_MODE`` and ``HIPBLASLT_MATMUL_DESC_B_SCALE_MODE`` are both set
+to ``HIPBLASLT_MATMUL_MATRIX_SCALE_VEC32_UE8M0``, the following restrictions apply:
+
+*  Atype and Btype can be any combination of: ``HIP_R_8F_E4M3``, ``HIP_R_8F_E5M2``, ``HIP_R_6F_E2M3``, ``HIP_R_6F_E3M2``, and ``HIP_R_4F_E2M1``.
+*  Ctype must be the same as Dtype.
+*  Dtype can be any of the following: ``HIP_R_32F``, ``HIP_R_16F``, or ``HIP_R_16BF``.
+*  ``m % 16`` must be equal to ``0``.
+*  ``n % 16`` must be equal to ``0``.
+*  ``K % 128`` must be equal to ``0``.
+*  ``B`` must be equal to ``1``.
+*  ``opA`` must be equal to ``T``.
+*  ``opB`` must be equal to ``N``.
+*  Epilogues are not supported.
+*  The scaling data pointed to by ``HIPBLASLT_MATMUL_DESC_A_SCALE_POINTER`` must be stored in the same order as ``A``.
+*  The scaling data pointed to by ``HIPBLASLT_MATMUL_DESC_B_SCALE_POINTER`` must be stored in the same order as ``B``.
+
 hipblasLtMatrixTransformDescCreate()
 ------------------------------------------
 .. doxygenfunction:: hipblasLtMatrixTransformDescCreate

projects/hipblaslt/docs/reference/data-type-support.rst

@@ -35,6 +35,21 @@ The hipBLASLt input and output types are listed in the following table.
       - ``hipblasLtInt32``
       - 32-bit real signed integer.
 
+    * 
+      - ``HIP_R_4F_E2M1``
+      - N/A
+      - 4-bit real float4 precision floating-point
+
+    * 
+      - ``HIP_R_6F_E2M3``
+      - N/A
+      - 6-bit real float6 precision floating-point
+
+    * 
+      - ``HIP_R_6F_E3M2``
+      - N/A
+      - 6-bit real bfloat6 precision floating-point
+
     * 
       - ``HIP_R_8F_E4M3_FNUZ``
       - ``hipblaslt_f8_fnuz``
@@ -45,6 +60,16 @@ The hipBLASLt input and output types are listed in the following table.
       - ``hipblaslt_bf8_fnuz``
       - 8-bit real bfloat8 precision floating-point
 
+    * 
+      - ``HIP_R_8F_E4M3``
+      - ``hipblaslt_f8``
+      - 8-bit real float8 precision floating-point
+
+    * 
+      - ``HIP_R_8F_E5M2``
+      - ``hipblaslt_bf8``
+      - 8-bit real bfloat8 precision floating-point
+
     * 
       - ``HIP_R_16F``
       - ``hipblasLtHalf``
@@ -60,6 +85,12 @@ The hipBLASLt input and output types are listed in the following table.
       - ``hipblasLtFloat``
       - 32-bit real single precision floating-point
 
+.. note:: 
+
+   The ``hipblaslt_f8_fnuz`` and ``hipblaslt_bf8_fnuz`` data types are only
+   supported on the gfx942 platform. The ``hipblaslt_f8`` and ``hipblaslt_bf8`` data types are only
+   supported on the gfx950 and gfx12 platforms.
+
 The hipBLASLt compute modes are listed in the following table.
 
 .. list-table::
@@ -107,11 +138,6 @@ hipBLASLt GEMM operation equation:
 Where :math:`op( )` refers to in-place operations, such as transpose and
 non-transpose, and :math:`alpha` and :math:`beta` are scalars.
 
-.. note:: 
-
-  The ``hipblaslt_f8_fnuz`` and ``hipblaslt_bf8_fnuz`` data types are only
-  supported on the gfx94x platform.
-
 
 .. list-table:: Supported data types
   :header-rows: 1

projects/hipblaslt/docs/reference/datatypes.rst

@@ -48,6 +48,10 @@ hipblasLtMatmulPreferenceAttributes_t
 -------------------------------------
 .. doxygenenum:: hipblasLtMatmulPreferenceAttributes_t
 
+hipblasLtMatmulMatrixScale_t
+--------------------------------
+.. doxygenenum:: hipblasLtMatmulMatrixScale_t
+
 hipblasLtMatrixLayout_t
 -----------------------
 .. doxygentypedef:: hipblasLtMatrixLayout_t
@@ -59,3 +63,4 @@ hipblasLtMatrixLayoutAttribute_t
 hipblasLtMatrixTransformDesc_t
 --------------------------------
 .. doxygentypedef:: hipblasLtMatrixTransformDesc_t
+

projects/hipblaslt/library/include/hipblaslt/hipblaslt.h

@@ -161,12 +161,12 @@ typedef enum {
  *  \brief Block scale mode for A and B.
  */
 typedef enum {
-    HIPBLASLT_MATMUL_MATRIX_SCALE_SCALAR_32F = 0,    /** Scaling factors are single-precision scalars applied to the whole tensors (this mode is the default for fp8). */
-    HIPBLASLT_MATMUL_MATRIX_SCALE_VEC16_UE4M3 = 1,   /** Not supported yet. Scaling factors are tensors that contain a dedicated scaling factor stored as an 8-bit HIP_R_8F_E4M3 value for each 16-element block in the innermost dimension of the corresponding data tensor. */
-    HIPBLASLT_MATMUL_MATRIX_SCALE_VEC32_UE8M0 = 2,   /** Scaling factors are tensors that contain a dedicated scaling factor stored as an 8-bit R_8F_UE8M0 value for each 32-element block in the innermost dimension of the corresponding data tensor. */
-    HIPBLASLT_MATMUL_MATRIX_SCALE_OUTER_VEC_32F = 3, /** Scaling factors are single-precision vectors. This mode is only applicable to matrices A and B, in which case the vectors are expected to have M and N elements respectively, and each (i, j)-th element of product of A and B is multiplied by i-th element of A scale and j-th element of B scale. */
-    HIPBLASLT_MATMUL_MATRIX_SCALE_VEC128_32F = 4,    /** Not supported yet. Scaling factors are tensors that contain a dedicated FP32 scaling factor for each 128-element block in the innermost dimension of the corresponding data tensor */
-    HIPBLASLT_MATMUL_MATRIX_SCALE_BLK128x128_32F = 5,/** Not supported yet. Scaling factors are tensors that contain a dedicated FP32 scaling factor for each 128x128-element block in the corresponding data tensor */
+    HIPBLASLT_MATMUL_MATRIX_SCALE_SCALAR_32F = 0,    /**<Scaling factors are single-precision scalars applied to the whole tensors (this mode is the default for fp8). */
+    HIPBLASLT_MATMUL_MATRIX_SCALE_VEC16_UE4M3 = 1,   /**<Not supported yet. Scaling factors are tensors that contain a dedicated scaling factor stored as an 8-bit HIP_R_8F_E4M3 value for each 16-element block in the innermost dimension of the corresponding data tensor. */
+    HIPBLASLT_MATMUL_MATRIX_SCALE_VEC32_UE8M0 = 2,   /**<Scaling factors are tensors that contain a dedicated scaling factor stored as an 8-bit R_8F_UE8M0 value for each 32-element block in the innermost dimension of the corresponding data tensor. */
+    HIPBLASLT_MATMUL_MATRIX_SCALE_OUTER_VEC_32F = 3, /**<Scaling factors are single-precision vectors. This mode is only applicable to matrices A and B, in which case the vectors are expected to have M and N elements respectively, and each (i, j)-th element of product of A and B is multiplied by i-th element of A scale and j-th element of B scale. */
+    HIPBLASLT_MATMUL_MATRIX_SCALE_VEC128_32F = 4,    /**<Not supported yet. Scaling factors are tensors that contain a dedicated FP32 scaling factor for each 128-element block in the innermost dimension of the corresponding data tensor */
+    HIPBLASLT_MATMUL_MATRIX_SCALE_BLK128x128_32F = 5, /**<Not supported yet. Scaling factors are tensors that contain a dedicated FP32 scaling factor for each 128x128-element block in the corresponding data tensor */
     HIPBLASLT_MATMUL_MATRIX_SCALE_END
 } hipblasLtMatmulMatrixScale_t;