Prepared generator for batched routines; added batched AXPY routine interface

5 files changed, 194 insertions, 103 deletions

diff --git a/scripts/generator/generator.py b/scripts/generator/generator.py
index 09c743bb..521afe28 100755
--- a/scripts/generator/generator.py
+++ b/scripts/generator/generator.py
@@ -41,7 +41,7 @@ FILES = [
     "/include/clblast_netlib_c.h",
     "/src/clblast_netlib_c.cpp",
 ]
-HEADER_LINES = [121, 75, 125, 23, 29, 41, 65, 32]
+HEADER_LINES = [121, 76, 125, 23, 29, 41, 65, 32]
 FOOTER_LINES = [25, 138, 27, 38, 6, 6, 9, 2]
 HEADER_LINES_DOC = 0
 FOOTER_LINES_DOC = 63
@@ -101,65 +101,68 @@ bmnn = size_helper("layout == CLBlastLayoutRowMajor", "((side == CLBlastSideLeft
 # Populates a list of routines
 ROUTINES = [
 [  # Level 1: vector-vector
-  Routine(False, True,  "1", "rotg",  T, [S,D],            [],                  [],                                                     [],         ["sa","sb","sc","ss"],        ["1","1","1","1"], [],       "",    "Generate givens plane rotation", "", []),
-  Routine(False, True,  "1", "rotmg", T, [S,D],            [],                  [],                                                     ["sy1"],    ["sd1","sd2","sx1","sparam"], ["1","1","1","1","1"], [],   "",    "Generate modified givens plane rotation", "", []),
-  Routine(False, True,  "1", "rot",   T, [S,D],            ["n"],               [],                                                     [],         ["x","y"],                    [xn,yn],       ["cos","sin"],"",    "Apply givens plane rotation", "", []),
-  Routine(False, True,  "1", "rotm",  T, [S,D],            ["n"],               [],                                                     [],         ["x","y","sparam"],           [xn,yn,"1"],   [],           "",    "Apply modified givens plane rotation", "", []),
-  Routine(True,  True,  "1", "swap",  T, [S,D,C,Z,H],      ["n"],               [],                                                     [],         ["x","y"],                    [xn,yn],       [],           "",    "Swap two vectors", "Interchanges _n_ elements of vectors _x_ and _y_.", []),
-  Routine(True,  True,  "1", "scal",  T, [S,D,C,Z,H],      ["n"],               [],                                                     [],         ["x"],                        [xn],          ["alpha"],    "",    "Vector scaling", "Multiplies _n_ elements of vector _x_ by a scalar constant _alpha_.", []),
-  Routine(True,  True,  "1", "copy",  T, [S,D,C,Z,H],      ["n"],               [],                                                     ["x"],      ["y"],                        [xn,yn],       [],           "",    "Vector copy", "Copies the contents of vector _x_ into vector _y_.", []),
-  Routine(True,  True,  "1", "axpy",  T, [S,D,C,Z,H],      ["n"],               [],                                                     ["x"],      ["y"],                        [xn,yn],       ["alpha"],    "",    "Vector-times-constant plus vector", "Performs the operation _y = alpha * x + y_, in which _x_ and _y_ are vectors and _alpha_ is a scalar constant.", []),
-  Routine(True,  True,  "1", "dot",   T, [S,D,H],          ["n"],               [],                                                     ["x","y"],  ["dot"],                      [xn,yn,"1"],   [],           "n",   "Dot product of two vectors", "Multiplies _n_ elements of the vectors _x_ and _y_ element-wise and accumulates the results. The sum is stored in the _dot_ buffer.", []),
-  Routine(True,  True,  "1", "dotu",  T, [C,Z],            ["n"],               [],                                                     ["x","y"],  ["dot"],                      [xn,yn,"1"],   [],           "n",   "Dot product of two complex vectors", "See the regular xDOT routine.", []),
-  Routine(True,  True,  "1", "dotc",  T, [C,Z],            ["n"],               [],                                                     ["x","y"],  ["dot"],                      [xn,yn,"1"],   [],           "n",   "Dot product of two complex vectors, one conjugated", "See the regular xDOT routine.", []),
-  Routine(True,  True,  "1", "nrm2",  T, [S,D,Sc,Dz,H],    ["n"],               [],                                                     ["x"],      ["nrm2"],                     [xn,"1"],      [],           "2*n", "Euclidian norm of a vector", "Accumulates the square of _n_ elements in the _x_ vector and takes the square root. The resulting L2 norm is stored in the _nrm2_ buffer.", []),
-  Routine(True,  True,  "1", "asum",  T, [S,D,Sc,Dz,H],    ["n"],               [],                                                     ["x"],      ["asum"],                     [xn,"1"],      [],           "n",   "Absolute sum of values in a vector", "Accumulates the absolute value of _n_ elements in the _x_ vector. The results are stored in the _asum_ buffer.", []),
-  Routine(True,  False, "1", "sum",   T, [S,D,Sc,Dz,H],    ["n"],               [],                                                     ["x"],      ["sum"],                      [xn,"1"],      [],           "n",   "Sum of values in a vector (non-BLAS function)", "Accumulates the values of _n_ elements in the _x_ vector. The results are stored in the _sum_ buffer. This routine is the non-absolute version of the xASUM BLAS routine.", []),
-  Routine(True,  True,  "1", "amax",  T, [iS,iD,iC,iZ,iH], ["n"],               [],                                                     ["x"],      ["imax"],                     [xn,"1"],      [],           "2*n", "Index of absolute maximum value in a vector", "Finds the index of the maximum of the absolute values in the _x_ vector. The resulting integer index is stored in the _imax_ buffer.", []),
-  Routine(True,  False, "1", "max",   T, [iS,iD,iC,iZ,iH], ["n"],               [],                                                     ["x"],      ["imax"],                     [xn,"1"],      [],           "2*n", "Index of maximum value in a vector (non-BLAS function)", "Finds the index of the maximum of the values in the _x_ vector. The resulting integer index is stored in the _imax_ buffer. This routine is the non-absolute version of the IxAMAX BLAS routine.", []),
-  Routine(True,  False, "1", "min",   T, [iS,iD,iC,iZ,iH], ["n"],               [],                                                     ["x"],      ["imin"],                     [xn,"1"],      [],           "2*n", "Index of minimum value in a vector (non-BLAS function)", "Finds the index of the minimum of the values in the _x_ vector. The resulting integer index is stored in the _imin_ buffer. This routine is the non-absolute minimum version of the IxAMAX BLAS routine.", []),
+  Routine(False, True,  False, "1", "rotg",  T, [S,D],            [],                  [],                                                     [],         ["sa","sb","sc","ss"],        ["1","1","1","1"], [],       "",    "Generate givens plane rotation", "", []),
+  Routine(False, True,  False, "1", "rotmg", T, [S,D],            [],                  [],                                                     ["sy1"],    ["sd1","sd2","sx1","sparam"], ["1","1","1","1","1"], [],   "",    "Generate modified givens plane rotation", "", []),
+  Routine(False, True,  False, "1", "rot",   T, [S,D],            ["n"],               [],                                                     [],         ["x","y"],                    [xn,yn],       ["cos","sin"],"",    "Apply givens plane rotation", "", []),
+  Routine(False, True,  False, "1", "rotm",  T, [S,D],            ["n"],               [],                                                     [],         ["x","y","sparam"],           [xn,yn,"1"],   [],           "",    "Apply modified givens plane rotation", "", []),
+  Routine(True,  True,  False, "1", "swap",  T, [S,D,C,Z,H],      ["n"],               [],                                                     [],         ["x","y"],                    [xn,yn],       [],           "",    "Swap two vectors", "Interchanges _n_ elements of vectors _x_ and _y_.", []),
+  Routine(True,  True,  False, "1", "scal",  T, [S,D,C,Z,H],      ["n"],               [],                                                     [],         ["x"],                        [xn],          ["alpha"],    "",    "Vector scaling", "Multiplies _n_ elements of vector _x_ by a scalar constant _alpha_.", []),
+  Routine(True,  True,  False, "1", "copy",  T, [S,D,C,Z,H],      ["n"],               [],                                                     ["x"],      ["y"],                        [xn,yn],       [],           "",    "Vector copy", "Copies the contents of vector _x_ into vector _y_.", []),
+  Routine(True,  True,  False, "1", "axpy",  T, [S,D,C,Z,H],      ["n"],               [],                                                     ["x"],      ["y"],                        [xn,yn],       ["alpha"],    "",    "Vector-times-constant plus vector", "Performs the operation _y = alpha * x + y_, in which _x_ and _y_ are vectors and _alpha_ is a scalar constant.", []),
+  Routine(True,  True,  False, "1", "dot",   T, [S,D,H],          ["n"],               [],                                                     ["x","y"],  ["dot"],                      [xn,yn,"1"],   [],           "n",   "Dot product of two vectors", "Multiplies _n_ elements of the vectors _x_ and _y_ element-wise and accumulates the results. The sum is stored in the _dot_ buffer.", []),
+  Routine(True,  True,  False, "1", "dotu",  T, [C,Z],            ["n"],               [],                                                     ["x","y"],  ["dot"],                      [xn,yn,"1"],   [],           "n",   "Dot product of two complex vectors", "See the regular xDOT routine.", []),
+  Routine(True,  True,  False, "1", "dotc",  T, [C,Z],            ["n"],               [],                                                     ["x","y"],  ["dot"],                      [xn,yn,"1"],   [],           "n",   "Dot product of two complex vectors, one conjugated", "See the regular xDOT routine.", []),
+  Routine(True,  True,  False, "1", "nrm2",  T, [S,D,Sc,Dz,H],    ["n"],               [],                                                     ["x"],      ["nrm2"],                     [xn,"1"],      [],           "2*n", "Euclidian norm of a vector", "Accumulates the square of _n_ elements in the _x_ vector and takes the square root. The resulting L2 norm is stored in the _nrm2_ buffer.", []),
+  Routine(True,  True,  False, "1", "asum",  T, [S,D,Sc,Dz,H],    ["n"],               [],                                                     ["x"],      ["asum"],                     [xn,"1"],      [],           "n",   "Absolute sum of values in a vector", "Accumulates the absolute value of _n_ elements in the _x_ vector. The results are stored in the _asum_ buffer.", []),
+  Routine(True,  False, False, "1", "sum",   T, [S,D,Sc,Dz,H],    ["n"],               [],                                                     ["x"],      ["sum"],                      [xn,"1"],      [],           "n",   "Sum of values in a vector (non-BLAS function)", "Accumulates the values of _n_ elements in the _x_ vector. The results are stored in the _sum_ buffer. This routine is the non-absolute version of the xASUM BLAS routine.", []),
+  Routine(True,  True,  False, "1", "amax",  T, [iS,iD,iC,iZ,iH], ["n"],               [],                                                     ["x"],      ["imax"],                     [xn,"1"],      [],           "2*n", "Index of absolute maximum value in a vector", "Finds the index of the maximum of the absolute values in the _x_ vector. The resulting integer index is stored in the _imax_ buffer.", []),
+  Routine(True,  False, False, "1", "max",   T, [iS,iD,iC,iZ,iH], ["n"],               [],                                                     ["x"],      ["imax"],                     [xn,"1"],      [],           "2*n", "Index of maximum value in a vector (non-BLAS function)", "Finds the index of the maximum of the values in the _x_ vector. The resulting integer index is stored in the _imax_ buffer. This routine is the non-absolute version of the IxAMAX BLAS routine.", []),
+  Routine(True,  False, False, "1", "min",   T, [iS,iD,iC,iZ,iH], ["n"],               [],                                                     ["x"],      ["imin"],                     [xn,"1"],      [],           "2*n", "Index of minimum value in a vector (non-BLAS function)", "Finds the index of the minimum of the values in the _x_ vector. The resulting integer index is stored in the _imin_ buffer. This routine is the non-absolute minimum version of the IxAMAX BLAS routine.", []),
 ],
 [  # Level 2: matrix-vector
-  Routine(True,  True,  "2a", "gemv",  T,  [S,D,C,Z,H],    ["m","n"],           ["layout","a_transpose"],                               ["a","x"],  ["y"],                        [amn,xmn,ynm], ["alpha","beta"], "",    "General matrix-vector multiplication", "Performs the operation _y = alpha * A * x + beta * y_, in which _x_ is an input vector, _y_ is an input and output vector, _A_ is an input matrix, and _alpha_ and _beta_ are scalars. The matrix _A_ can optionally be transposed before performing the operation.", [ald_m]),
-  Routine(True,  True,  "2a", "gbmv",  T,  [S,D,C,Z,H],    ["m","n","kl","ku"], ["layout","a_transpose"],                               ["a","x"],  ["y"],                        [amn,xmn,ynm], ["alpha","beta"], "",    "General banded matrix-vector multiplication", "Same operation as xGEMV, but matrix _A_ is banded instead.", [ald_kl_ku_one]),
-  Routine(True,  True,  "2a", "hemv",  T,  [C,Z],          ["n"],               ["layout","triangle"],                                  ["a","x"],  ["y"],                        [an,xn,yn],    ["alpha","beta"], "",    "Hermitian matrix-vector multiplication", "Same operation as xGEMV, but matrix _A_ is an Hermitian matrix instead.", [ald_n]),
-  Routine(True,  True,  "2a", "hbmv",  T,  [C,Z],          ["n","k"],           ["layout","triangle"],                                  ["a","x"],  ["y"],                        [an,xn,yn],    ["alpha","beta"], "",    "Hermitian banded matrix-vector multiplication", "Same operation as xGEMV, but matrix _A_ is an Hermitian banded matrix instead.", [ald_k_one]),
-  Routine(True,  True,  "2a", "hpmv",  T,  [C,Z],          ["n"],               ["layout","triangle"],                                  ["ap","x"], ["y"],                        [apn,xn,yn],   ["alpha","beta"], "",    "Hermitian packed matrix-vector multiplication", "Same operation as xGEMV, but matrix _A_ is an Hermitian packed matrix instead and represented as _AP_.", []),
-  Routine(True,  True,  "2a", "symv",  T,  [S,D,H],        ["n"],               ["layout","triangle"],                                  ["a","x"],  ["y"],                        [an,xn,yn],    ["alpha","beta"], "",    "Symmetric matrix-vector multiplication", "Same operation as xGEMV, but matrix _A_ is symmetric instead.", [ald_n]),
-  Routine(True,  True,  "2a", "sbmv",  T,  [S,D,H],        ["n","k"],           ["layout","triangle"],                                  ["a","x"],  ["y"],                        [an,xn,yn],    ["alpha","beta"], "",    "Symmetric banded matrix-vector multiplication", "Same operation as xGEMV, but matrix _A_ is symmetric and banded instead.", [ald_k_one]),
-  Routine(True,  True,  "2a", "spmv",  T,  [S,D,H],        ["n"],               ["layout","triangle"],                                  ["ap","x"], ["y"],                        [apn,xn,yn],   ["alpha","beta"], "",    "Symmetric packed matrix-vector multiplication", "Same operation as xGEMV, but matrix _A_ is a symmetric packed matrix instead and represented as _AP_.", []),
-  Routine(True,  True,  "2a", "trmv",  T,  [S,D,C,Z,H],    ["n"],               ["layout","triangle","a_transpose","diagonal"],         ["a"],      ["x"],                        [an,xn],       [],               "n",   "Triangular matrix-vector multiplication", "Same operation as xGEMV, but matrix _A_ is triangular instead.", [ald_n]),
-  Routine(True,  True,  "2a", "tbmv",  T,  [S,D,C,Z,H],    ["n","k"],           ["layout","triangle","a_transpose","diagonal"],         ["a"],      ["x"],                        [an,xn],       [],               "n",   "Triangular banded matrix-vector multiplication", "Same operation as xGEMV, but matrix _A_ is triangular and banded instead.", [ald_k_one]),
-  Routine(True,  True,  "2a", "tpmv",  T,  [S,D,C,Z,H],    ["n"],               ["layout","triangle","a_transpose","diagonal"],         ["ap"],     ["x"],                        [apn,xn],      [],               "n",   "Triangular packed matrix-vector multiplication", "Same operation as xGEMV, but matrix _A_ is a triangular packed matrix instead and repreented as _AP_.", []),
-  Routine(True,  True,  "2a", "trsv",  T,  [S,D,C,Z],      ["n"],               ["layout","triangle","a_transpose","diagonal"],         ["a"],      ["x"],                        [an,xn],       [],               "",    "Solves a triangular system of equations", "", []),
-  Routine(False, True,  "2a", "tbsv",  T,  [S,D,C,Z],      ["n","k"],           ["layout","triangle","a_transpose","diagonal"],         ["a"],      ["x"],                        [an,xn],       [],               "",    "Solves a banded triangular system of equations", "", [ald_k_one]),
-  Routine(False, True,  "2a", "tpsv",  T,  [S,D,C,Z],      ["n"],               ["layout","triangle","a_transpose","diagonal"],         ["ap"],     ["x"],                        [apn,xn],      [],               "",    "Solves a packed triangular system of equations", "", []),
+  Routine(True,  True,  False, "2a", "gemv",  T,  [S,D,C,Z,H],    ["m","n"],           ["layout","a_transpose"],                               ["a","x"],  ["y"],                        [amn,xmn,ynm], ["alpha","beta"], "",    "General matrix-vector multiplication", "Performs the operation _y = alpha * A * x + beta * y_, in which _x_ is an input vector, _y_ is an input and output vector, _A_ is an input matrix, and _alpha_ and _beta_ are scalars. The matrix _A_ can optionally be transposed before performing the operation.", [ald_m]),
+  Routine(True,  True,  False, "2a", "gbmv",  T,  [S,D,C,Z,H],    ["m","n","kl","ku"], ["layout","a_transpose"],                               ["a","x"],  ["y"],                        [amn,xmn,ynm], ["alpha","beta"], "",    "General banded matrix-vector multiplication", "Same operation as xGEMV, but matrix _A_ is banded instead.", [ald_kl_ku_one]),
+  Routine(True,  True,  False, "2a", "hemv",  T,  [C,Z],          ["n"],               ["layout","triangle"],                                  ["a","x"],  ["y"],                        [an,xn,yn],    ["alpha","beta"], "",    "Hermitian matrix-vector multiplication", "Same operation as xGEMV, but matrix _A_ is an Hermitian matrix instead.", [ald_n]),
+  Routine(True,  True,  False, "2a", "hbmv",  T,  [C,Z],          ["n","k"],           ["layout","triangle"],                                  ["a","x"],  ["y"],                        [an,xn,yn],    ["alpha","beta"], "",    "Hermitian banded matrix-vector multiplication", "Same operation as xGEMV, but matrix _A_ is an Hermitian banded matrix instead.", [ald_k_one]),
+  Routine(True,  True,  False, "2a", "hpmv",  T,  [C,Z],          ["n"],               ["layout","triangle"],                                  ["ap","x"], ["y"],                        [apn,xn,yn],   ["alpha","beta"], "",    "Hermitian packed matrix-vector multiplication", "Same operation as xGEMV, but matrix _A_ is an Hermitian packed matrix instead and represented as _AP_.", []),
+  Routine(True,  True,  False, "2a", "symv",  T,  [S,D,H],        ["n"],               ["layout","triangle"],                                  ["a","x"],  ["y"],                        [an,xn,yn],    ["alpha","beta"], "",    "Symmetric matrix-vector multiplication", "Same operation as xGEMV, but matrix _A_ is symmetric instead.", [ald_n]),
+  Routine(True,  True,  False, "2a", "sbmv",  T,  [S,D,H],        ["n","k"],           ["layout","triangle"],                                  ["a","x"],  ["y"],                        [an,xn,yn],    ["alpha","beta"], "",    "Symmetric banded matrix-vector multiplication", "Same operation as xGEMV, but matrix _A_ is symmetric and banded instead.", [ald_k_one]),
+  Routine(True,  True,  False, "2a", "spmv",  T,  [S,D,H],        ["n"],               ["layout","triangle"],                                  ["ap","x"], ["y"],                        [apn,xn,yn],   ["alpha","beta"], "",    "Symmetric packed matrix-vector multiplication", "Same operation as xGEMV, but matrix _A_ is a symmetric packed matrix instead and represented as _AP_.", []),
+  Routine(True,  True,  False, "2a", "trmv",  T,  [S,D,C,Z,H],    ["n"],               ["layout","triangle","a_transpose","diagonal"],         ["a"],      ["x"],                        [an,xn],       [],               "n",   "Triangular matrix-vector multiplication", "Same operation as xGEMV, but matrix _A_ is triangular instead.", [ald_n]),
+  Routine(True,  True,  False, "2a", "tbmv",  T,  [S,D,C,Z,H],    ["n","k"],           ["layout","triangle","a_transpose","diagonal"],         ["a"],      ["x"],                        [an,xn],       [],               "n",   "Triangular banded matrix-vector multiplication", "Same operation as xGEMV, but matrix _A_ is triangular and banded instead.", [ald_k_one]),
+  Routine(True,  True,  False, "2a", "tpmv",  T,  [S,D,C,Z,H],    ["n"],               ["layout","triangle","a_transpose","diagonal"],         ["ap"],     ["x"],                        [apn,xn],      [],               "n",   "Triangular packed matrix-vector multiplication", "Same operation as xGEMV, but matrix _A_ is a triangular packed matrix instead and repreented as _AP_.", []),
+  Routine(True,  True,  False, "2a", "trsv",  T,  [S,D,C,Z],      ["n"],               ["layout","triangle","a_transpose","diagonal"],         ["a"],      ["x"],                        [an,xn],       [],               "",    "Solves a triangular system of equations", "", []),
+  Routine(False, True,  False, "2a", "tbsv",  T,  [S,D,C,Z],      ["n","k"],           ["layout","triangle","a_transpose","diagonal"],         ["a"],      ["x"],                        [an,xn],       [],               "",    "Solves a banded triangular system of equations", "", [ald_k_one]),
+  Routine(False, True,  False, "2a", "tpsv",  T,  [S,D,C,Z],      ["n"],               ["layout","triangle","a_transpose","diagonal"],         ["ap"],     ["x"],                        [apn,xn],      [],               "",    "Solves a packed triangular system of equations", "", []),
   # Level 2: matrix update
-  Routine(True,  True,  "2b", "ger",   T,  [S,D,H],        ["m","n"],           ["layout"],                                             ["x","y"],  ["a"],                        [xm,yn,amn],   ["alpha"],        "",    "General rank-1 matrix update", "Performs the operation _A = alpha * x * y^T + A_, in which _x_ is an input vector, _y^T_ is the transpose of the input vector _y_, _A_ is the matrix to be updated, and _alpha_ is a scalar value.", [ald_m]),
-  Routine(True,  True,  "2b", "geru",  T,  [C,Z],          ["m","n"],           ["layout"],                                             ["x","y"],  ["a"],                        [xm,yn,amn],   ["alpha"],        "",    "General rank-1 complex matrix update", "Same operation as xGER, but with complex data-types.", [ald_m]),
-  Routine(True,  True,  "2b", "gerc",  T,  [C,Z],          ["m","n"],           ["layout"],                                             ["x","y"],  ["a"],                        [xm,yn,amn],   ["alpha"],        "",    "General rank-1 complex conjugated matrix update", "Same operation as xGERU, but the update is done based on the complex conjugate of the input vectors.", [ald_m]),
-  Routine(True,  True,  "2b", "her",   Tc, [Css,Zdd],      ["n"],               ["layout","triangle"],                                  ["x"],      ["a"],                        [xn,an],       ["alpha"],        "",    "Hermitian rank-1 matrix update", "Performs the operation _A = alpha * x * x^T + A_, in which x is an input vector, x^T is the transpose of this vector, _A_ is the triangular Hermetian matrix to be updated, and alpha is a scalar value.", [ald_n]),
-  Routine(True,  True,  "2b", "hpr",   Tc, [Css,Zdd],      ["n"],               ["layout","triangle"],                                  ["x"],      ["ap"],                       [xn,apn],      ["alpha"],        "",    "Hermitian packed rank-1 matrix update", "Same operation as xHER, but matrix _A_ is an Hermitian packed matrix instead and represented as _AP_.", []),
-  Routine(True,  True,  "2b", "her2",  T,  [C,Z],          ["n"],               ["layout","triangle"],                                  ["x","y"],  ["a"],                        [xn,yn,an],    ["alpha"],        "",    "Hermitian rank-2 matrix update", "Performs the operation _A = alpha * x * y^T + conj(alpha) * y * x^T + A_, in which _x_ is an input vector and _x^T_ its transpose, _y_ is an input vector and _y^T_ its transpose, _A_ is the triangular Hermetian matrix to be updated, _alpha_ is a scalar value and _conj(alpha)_ its complex conjugate.", [ald_n]),
-  Routine(True,  True,  "2b", "hpr2",  T,  [C,Z],          ["n"],               ["layout","triangle"],                                  ["x","y"],  ["ap"],                       [xn,yn,apn],   ["alpha"],        "",    "Hermitian packed rank-2 matrix update", "Same operation as xHER2, but matrix _A_ is an Hermitian packed matrix instead and represented as _AP_.", []),
-  Routine(True,  True,  "2b", "syr",   T,  [S,D,H],        ["n"],               ["layout","triangle"],                                  ["x"],      ["a"],                        [xn,an],       ["alpha"],        "",    "Symmetric rank-1 matrix update", "Same operation as xHER, but matrix A is a symmetric matrix instead.", [ald_n]),
-  Routine(True,  True,  "2b", "spr",   T,  [S,D,H],        ["n"],               ["layout","triangle"],                                  ["x"],      ["ap"],                       [xn,apn],      ["alpha"],        "",    "Symmetric packed rank-1 matrix update", "Same operation as xSPR, but matrix _A_ is a symmetric packed matrix instead and represented as _AP_.", []),
-  Routine(True,  True,  "2b", "syr2",  T,  [S,D,H],        ["n"],               ["layout","triangle"],                                  ["x","y"],  ["a"],                        [xn,yn,an],    ["alpha"],        "",    "Symmetric rank-2 matrix update", "Same operation as xHER2, but matrix _A_ is a symmetric matrix instead.", [ald_n]),
-  Routine(True,  True,  "2b", "spr2",  T,  [S,D,H],        ["n"],               ["layout","triangle"],                                  ["x","y"],  ["ap"],                       [xn,yn,apn],   ["alpha"],        "",    "Symmetric packed rank-2 matrix update", "Same operation as xSPR2, but matrix _A_ is a symmetric packed matrix instead and represented as _AP_.", []),
+  Routine(True,  True,  False, "2b", "ger",   T,  [S,D,H],        ["m","n"],           ["layout"],                                             ["x","y"],  ["a"],                        [xm,yn,amn],   ["alpha"],        "",    "General rank-1 matrix update", "Performs the operation _A = alpha * x * y^T + A_, in which _x_ is an input vector, _y^T_ is the transpose of the input vector _y_, _A_ is the matrix to be updated, and _alpha_ is a scalar value.", [ald_m]),
+  Routine(True,  True,  False, "2b", "geru",  T,  [C,Z],          ["m","n"],           ["layout"],                                             ["x","y"],  ["a"],                        [xm,yn,amn],   ["alpha"],        "",    "General rank-1 complex matrix update", "Same operation as xGER, but with complex data-types.", [ald_m]),
+  Routine(True,  True,  False, "2b", "gerc",  T,  [C,Z],          ["m","n"],           ["layout"],                                             ["x","y"],  ["a"],                        [xm,yn,amn],   ["alpha"],        "",    "General rank-1 complex conjugated matrix update", "Same operation as xGERU, but the update is done based on the complex conjugate of the input vectors.", [ald_m]),
+  Routine(True,  True,  False, "2b", "her",   Tc, [Css,Zdd],      ["n"],               ["layout","triangle"],                                  ["x"],      ["a"],                        [xn,an],       ["alpha"],        "",    "Hermitian rank-1 matrix update", "Performs the operation _A = alpha * x * x^T + A_, in which x is an input vector, x^T is the transpose of this vector, _A_ is the triangular Hermetian matrix to be updated, and alpha is a scalar value.", [ald_n]),
+  Routine(True,  True,  False, "2b", "hpr",   Tc, [Css,Zdd],      ["n"],               ["layout","triangle"],                                  ["x"],      ["ap"],                       [xn,apn],      ["alpha"],        "",    "Hermitian packed rank-1 matrix update", "Same operation as xHER, but matrix _A_ is an Hermitian packed matrix instead and represented as _AP_.", []),
+  Routine(True,  True,  False, "2b", "her2",  T,  [C,Z],          ["n"],               ["layout","triangle"],                                  ["x","y"],  ["a"],                        [xn,yn,an],    ["alpha"],        "",    "Hermitian rank-2 matrix update", "Performs the operation _A = alpha * x * y^T + conj(alpha) * y * x^T + A_, in which _x_ is an input vector and _x^T_ its transpose, _y_ is an input vector and _y^T_ its transpose, _A_ is the triangular Hermetian matrix to be updated, _alpha_ is a scalar value and _conj(alpha)_ its complex conjugate.", [ald_n]),
+  Routine(True,  True,  False, "2b", "hpr2",  T,  [C,Z],          ["n"],               ["layout","triangle"],                                  ["x","y"],  ["ap"],                       [xn,yn,apn],   ["alpha"],        "",    "Hermitian packed rank-2 matrix update", "Same operation as xHER2, but matrix _A_ is an Hermitian packed matrix instead and represented as _AP_.", []),
+  Routine(True,  True,  False, "2b", "syr",   T,  [S,D,H],        ["n"],               ["layout","triangle"],                                  ["x"],      ["a"],                        [xn,an],       ["alpha"],        "",    "Symmetric rank-1 matrix update", "Same operation as xHER, but matrix A is a symmetric matrix instead.", [ald_n]),
+  Routine(True,  True,  False, "2b", "spr",   T,  [S,D,H],        ["n"],               ["layout","triangle"],                                  ["x"],      ["ap"],                       [xn,apn],      ["alpha"],        "",    "Symmetric packed rank-1 matrix update", "Same operation as xSPR, but matrix _A_ is a symmetric packed matrix instead and represented as _AP_.", []),
+  Routine(True,  True,  False, "2b", "syr2",  T,  [S,D,H],        ["n"],               ["layout","triangle"],                                  ["x","y"],  ["a"],                        [xn,yn,an],    ["alpha"],        "",    "Symmetric rank-2 matrix update", "Same operation as xHER2, but matrix _A_ is a symmetric matrix instead.", [ald_n]),
+  Routine(True,  True,  False, "2b", "spr2",  T,  [S,D,H],        ["n"],               ["layout","triangle"],                                  ["x","y"],  ["ap"],                       [xn,yn,apn],   ["alpha"],        "",    "Symmetric packed rank-2 matrix update", "Same operation as xSPR2, but matrix _A_ is a symmetric packed matrix instead and represented as _AP_.", []),
 ],
 [  # Level 3: matrix-matrix
-  Routine(True,  True,  "3", "gemm",  T,  [S,D,C,Z,H],     ["m","n","k"],        ["layout","a_transpose","b_transpose"],                ["a","b"],  ["c"],                        [amk,bkn,cmn],   ["alpha","beta"], "",    "General matrix-matrix multiplication", "Performs the matrix product _C = alpha * A * B + beta * C_, in which _A_ (_m_ by _k_) and _B_ (_k_ by _n_) are two general rectangular input matrices, _C_ (_m_ by _n_) is the matrix to be updated, and _alpha_ and _beta_ are scalar values. The matrices _A_ and/or _B_ can optionally be transposed before performing the operation.", [ald_transa_m_k, bld_transb_k_n, cld_m]),
-  Routine(True,  True,  "3", "symm",  T,  [S,D,C,Z,H],     ["m","n"],            ["layout","side","triangle"],                          ["a","b"],  ["c"],                        [ammn,bmnn,cmn], ["alpha","beta"], "",    "Symmetric matrix-matrix multiplication", "Same operation as xGEMM, but _A_ is symmetric instead. In case of `side == kLeft`, _A_ is a symmetric _m_ by _m_ matrix and _C = alpha * A * B + beta * C_ is performed. Otherwise, in case of `side == kRight`, _A_ is a symmtric _n_ by _n_ matrix and _C = alpha * B * A + beta * C_ is performed.", [ald_side_m_n, bld_m, cld_m]),
-  Routine(True,  True,  "3", "hemm",  T,  [C,Z],           ["m","n"],            ["layout","side","triangle"],                          ["a","b"],  ["c"],                        [ammn,bmnn,cmn], ["alpha","beta"], "",    "Hermitian matrix-matrix multiplication", "Same operation as xSYMM, but _A_ is an Hermitian matrix instead.", [ald_side_m_n, bld_m, cld_m]),
-  Routine(True,  True,  "3", "syrk",  T,  [S,D,C,Z,H],     ["n","k"],            ["layout","triangle","a_transpose"],                   ["a"],      ["c"],                        [ank,cn],        ["alpha","beta"], "",    "Rank-K update of a symmetric matrix", "Performs the matrix product _C = alpha * A * A^T + beta * C_ or _C = alpha * A^T * A + beta * C_, in which _A_ is a general matrix and _A^T_ is its transpose, _C_ (_n_ by _n_) is the symmetric matrix to be updated, and _alpha_ and _beta_ are scalar values.", [ald_trans_n_k, cld_m]),
-  Routine(True,  True,  "3", "herk",  Tc, [Css,Zdd],       ["n","k"],            ["layout","triangle","a_transpose"],                   ["a"],      ["c"],                        [ank,cn],        ["alpha","beta"], "",    "Rank-K update of a hermitian matrix", "Same operation as xSYRK, but _C_ is an Hermitian matrix instead.", [ald_trans_n_k, cld_m]),
-  Routine(True,  True,  "3", "syr2k", T,  [S,D,C,Z,H],     ["n","k"],            ["layout","triangle","ab_transpose"],                  ["a","b"],  ["c"],                        [ankab,bnkab,cn],["alpha","beta"], "",    "Rank-2K update of a symmetric matrix", "Performs the matrix product _C = alpha * A * B^T + alpha * B * A^T + beta * C_ or _C = alpha * A^T * B + alpha * B^T * A + beta * C_, in which _A_ and _B_ are general matrices and _A^T_ and _B^T_ are their transposed versions, _C_ (_n_ by _n_) is the symmetric matrix to be updated, and _alpha_ and _beta_ are scalar values.", [ald_trans_n_k, bld_trans_n_k, cld_n]),
-  Routine(True,  True,  "3", "her2k", TU, [Ccs,Zzd],       ["n","k"],            ["layout","triangle","ab_transpose"],                  ["a","b"],  ["c"],                        [ankab,bnkab,cn],["alpha","beta"], "",    "Rank-2K update of a hermitian matrix", "Same operation as xSYR2K, but _C_ is an Hermitian matrix instead.", [ald_trans_n_k, bld_trans_n_k, cld_n]),
-  Routine(True,  True,  "3", "trmm",  T,  [S,D,C,Z,H],     ["m","n"],            ["layout","side","triangle","a_transpose","diagonal"], ["a"],      ["b"],                        [amns,bmn],      ["alpha"],        "",    "Triangular matrix-matrix multiplication", "Performs the matrix product _B = alpha * A * B_ or _B = alpha * B * A_, in which _A_ is a unit or non-unit triangular matrix, _B_ (_m_ by _n_) is the general matrix to be updated, and _alpha_ is a scalar value.", [ald_side_m_n, bld_m]),
-  Routine(True,  True,  "3", "trsm",  T,  [S,D,C,Z],       ["m","n"],            ["layout","side","triangle","a_transpose","diagonal"], ["a"],      ["b"],                        [amns,bmn],      ["alpha"],        "",    "Solves a triangular system of equations", "Solves the equation _A * X = alpha * B_ for the unknown _m_ by _n_ matrix X, in which _A_ is an _n_ by _n_ unit or non-unit triangular matrix and B is an _m_ by _n_ matrix. The matrix _B_ is overwritten by the solution _X_.", []),
+  Routine(True,  True,  False, "3", "gemm",  T,  [S,D,C,Z,H],     ["m","n","k"],        ["layout","a_transpose","b_transpose"],                ["a","b"],  ["c"],                        [amk,bkn,cmn],   ["alpha","beta"], "",    "General matrix-matrix multiplication", "Performs the matrix product _C = alpha * A * B + beta * C_, in which _A_ (_m_ by _k_) and _B_ (_k_ by _n_) are two general rectangular input matrices, _C_ (_m_ by _n_) is the matrix to be updated, and _alpha_ and _beta_ are scalar values. The matrices _A_ and/or _B_ can optionally be transposed before performing the operation.", [ald_transa_m_k, bld_transb_k_n, cld_m]),
+  Routine(True,  True,  False, "3", "symm",  T,  [S,D,C,Z,H],     ["m","n"],            ["layout","side","triangle"],                          ["a","b"],  ["c"],                        [ammn,bmnn,cmn], ["alpha","beta"], "",    "Symmetric matrix-matrix multiplication", "Same operation as xGEMM, but _A_ is symmetric instead. In case of `side == kLeft`, _A_ is a symmetric _m_ by _m_ matrix and _C = alpha * A * B + beta * C_ is performed. Otherwise, in case of `side == kRight`, _A_ is a symmtric _n_ by _n_ matrix and _C = alpha * B * A + beta * C_ is performed.", [ald_side_m_n, bld_m, cld_m]),
+  Routine(True,  True,  False, "3", "hemm",  T,  [C,Z],           ["m","n"],            ["layout","side","triangle"],                          ["a","b"],  ["c"],                        [ammn,bmnn,cmn], ["alpha","beta"], "",    "Hermitian matrix-matrix multiplication", "Same operation as xSYMM, but _A_ is an Hermitian matrix instead.", [ald_side_m_n, bld_m, cld_m]),
+  Routine(True,  True,  False, "3", "syrk",  T,  [S,D,C,Z,H],     ["n","k"],            ["layout","triangle","a_transpose"],                   ["a"],      ["c"],                        [ank,cn],        ["alpha","beta"], "",    "Rank-K update of a symmetric matrix", "Performs the matrix product _C = alpha * A * A^T + beta * C_ or _C = alpha * A^T * A + beta * C_, in which _A_ is a general matrix and _A^T_ is its transpose, _C_ (_n_ by _n_) is the symmetric matrix to be updated, and _alpha_ and _beta_ are scalar values.", [ald_trans_n_k, cld_m]),
+  Routine(True,  True,  False, "3", "herk",  Tc, [Css,Zdd],       ["n","k"],            ["layout","triangle","a_transpose"],                   ["a"],      ["c"],                        [ank,cn],        ["alpha","beta"], "",    "Rank-K update of a hermitian matrix", "Same operation as xSYRK, but _C_ is an Hermitian matrix instead.", [ald_trans_n_k, cld_m]),
+  Routine(True,  True,  False, "3", "syr2k", T,  [S,D,C,Z,H],     ["n","k"],            ["layout","triangle","ab_transpose"],                  ["a","b"],  ["c"],                        [ankab,bnkab,cn],["alpha","beta"], "",    "Rank-2K update of a symmetric matrix", "Performs the matrix product _C = alpha * A * B^T + alpha * B * A^T + beta * C_ or _C = alpha * A^T * B + alpha * B^T * A + beta * C_, in which _A_ and _B_ are general matrices and _A^T_ and _B^T_ are their transposed versions, _C_ (_n_ by _n_) is the symmetric matrix to be updated, and _alpha_ and _beta_ are scalar values.", [ald_trans_n_k, bld_trans_n_k, cld_n]),
+  Routine(True,  True,  False, "3", "her2k", TU, [Ccs,Zzd],       ["n","k"],            ["layout","triangle","ab_transpose"],                  ["a","b"],  ["c"],                        [ankab,bnkab,cn],["alpha","beta"], "",    "Rank-2K update of a hermitian matrix", "Same operation as xSYR2K, but _C_ is an Hermitian matrix instead.", [ald_trans_n_k, bld_trans_n_k, cld_n]),
+  Routine(True,  True,  False, "3", "trmm",  T,  [S,D,C,Z,H],     ["m","n"],            ["layout","side","triangle","a_transpose","diagonal"], ["a"],      ["b"],                        [amns,bmn],      ["alpha"],        "",    "Triangular matrix-matrix multiplication", "Performs the matrix product _B = alpha * A * B_ or _B = alpha * B * A_, in which _A_ is a unit or non-unit triangular matrix, _B_ (_m_ by _n_) is the general matrix to be updated, and _alpha_ is a scalar value.", [ald_side_m_n, bld_m]),
+  Routine(True,  True,  False, "3", "trsm",  T,  [S,D,C,Z],       ["m","n"],            ["layout","side","triangle","a_transpose","diagonal"], ["a"],      ["b"],                        [amns,bmn],      ["alpha"],        "",    "Solves a triangular system of equations", "Solves the equation _A * X = alpha * B_ for the unknown _m_ by _n_ matrix X, in which _A_ is an _n_ by _n_ unit or non-unit triangular matrix and B is an _m_ by _n_ matrix. The matrix _B_ is overwritten by the solution _X_.", []),
 ],
 [  # Level X: extra routines (not part of BLAS)
-  Routine(True,  True,  "x", "omatcopy", T, [S,D,C,Z,H],   ["m","n"],            ["layout","a_transpose"],                              ["a"],      ["b"],                        [amn,bnma],      ["alpha"],        "",    "Scaling and out-place transpose/copy (non-BLAS function)", "Performs scaling and out-of-place transposition/copying of matrices according to _B = alpha*op(A)_, in which _A_ is an input matrix (_m_ rows by _n_ columns), _B_ an output matrix, and _alpha_ a scalar value. The operation _op_ can be a normal matrix copy, a transposition or a conjugate transposition.", [ald_m, bld_n]),
+  # Special routines:
+  Routine(True,  True,  False, "x", "omatcopy", T, [S,D,C,Z,H],   ["m","n"],            ["layout","a_transpose"],                              ["a"],      ["b"],                        [amn,bnma],      ["alpha"],        "",    "Scaling and out-place transpose/copy (non-BLAS function)", "Performs scaling and out-of-place transposition/copying of matrices according to _B = alpha*op(A)_, in which _A_ is an input matrix (_m_ rows by _n_ columns), _B_ an output matrix, and _alpha_ a scalar value. The operation _op_ can be a normal matrix copy, a transposition or a conjugate transposition.", [ald_m, bld_n]),
+  # Batched routines:
+  Routine(True,  True,  True,  "x", "axpy",     T, [S,D,C,Z,H],   ["n"],                [],                                                    ["x"],      ["y"],                        [xn,yn],         ["alpha"],        "",    "Batched version of AXPY", "As AXPY, but multiple operations are batched together for better performance.", []),
 ]]
 
 
@@ -207,9 +210,11 @@ def main(argv):
                     if i == 5:
                         body += cpp.wrapper_cblas(routine)
                     if i == 6:
-                        body += cpp.clblast_netlib_c_h(routine)
+                        if not routine.batched:
+                            body += cpp.clblast_netlib_c_h(routine)
                     if i == 7:
-                        body += cpp.clblast_netlib_c_cc(routine)
+                        if not routine.batched:
+                            body += cpp.clblast_netlib_c_cc(routine)
             f.write("".join(file_header))
             f.write(body)
             f.write("".join(file_footer))
@@ -219,7 +224,7 @@ def main(argv):
         for routine in ROUTINES[level - 1]:
             if routine.has_tests:
                 level_string = cpp.LEVEL_NAMES[level - 1]
-                routine_suffix = "level" + level_string + "/x" + routine.name + ".cpp"
+                routine_suffix = "level" + level_string + "/x" + routine.lowercase_name() + ".cpp"
 
                 # Correctness tests
                 filename = library_root + "/test/correctness/routines/" + routine_suffix
diff --git a/scripts/generator/generator/cpp.py b/scripts/generator/generator/cpp.py
index c14d00a1..91fdf458 100644
--- a/scripts/generator/generator/cpp.py
+++ b/scripts/generator/generator/cpp.py
@@ -51,8 +51,10 @@ def clblast_cc(routine):
         result += routine.routine_header_cpp(12, "") + " {" + NL
         result += "  try {" + NL
         result += "    auto queue_cpp = Queue(*queue);" + NL
-        result += "    auto routine = X" + routine.name + "<" + routine.template.template + ">(queue_cpp, event);" + NL
-        result += "    routine.Do" + routine.name.capitalize() + "("
+        result += "    auto routine = X" + routine.plain_name() + "<" + routine.template.template + ">(queue_cpp, event);" + NL
+        if routine.batched:
+            result += "    " + (NL + "    ").join(routine.batched_transform_to_cpp()) + NL
+        result += "    routine.Do" + routine.capitalized_name() + "("
         result += ("," + NL + indent1).join([a for a in routine.arguments_clcudaapi()])
         result += ");" + NL
         result += "    return StatusCode::kSuccess;" + NL
@@ -63,7 +65,7 @@ def clblast_cc(routine):
     result += "}" + NL
     for flavour in routine.flavours:
         indent2 = " " * (34 + routine.length() + len(flavour.template))
-        result += "template StatusCode PUBLIC_API " + routine.name.capitalize() + "<" + flavour.template + ">("
+        result += "template StatusCode PUBLIC_API " + routine.capitalized_name() + "<" + flavour.template + ">("
         result += ("," + NL + indent2).join([a for a in routine.arguments_type(flavour)])
         result += "," + NL + indent2 + "cl_command_queue*, cl_event*);" + NL
     return result
@@ -84,9 +86,11 @@ def clblast_c_cc(routine):
         template = "<" + flavour.template + ">" if routine.no_scalars() else ""
         indent = " " * (16 + routine.length() + len(template))
         result += routine.routine_header_c(flavour, 27, "") + " {" + NL
+        if routine.batched:
+            result += "  " + (NL + "  ").join(routine.batched_transform_to_complex(flavour)) + NL
         result += "  try {" + NL
         result += "    return static_cast<CLBlastStatusCode>(" + NL
-        result += "      clblast::" + routine.name.capitalize() + template + "("
+        result += "      clblast::" + routine.capitalized_name() + template + "("
         result += ("," + NL + indent).join([a for a in routine.arguments_cast(flavour, indent)])
         result += "," + NL + indent + "queue, event)" + NL
         result += "    );" + NL
@@ -290,7 +294,7 @@ def performance_test(routine, level_string):
     """Generates the body of a performance test for a specific routine"""
     result = ""
     result += "#include \"test/performance/client.hpp\"" + NL
-    result += "#include \"test/routines/level" + level_string + "/x" + routine.name + ".hpp\"" + NL + NL
+    result += "#include \"test/routines/level" + level_string + "/x" + routine.lowercase_name() + ".hpp\"" + NL + NL
     result += "// Shortcuts to the clblast namespace" + NL
     result += "using float2 = clblast::float2;" + NL
     result += "using double2 = clblast::double2;" + NL + NL
@@ -304,7 +308,7 @@ def performance_test(routine, level_string):
         found = False
         for flavour in routine.flavours:
             if flavour.precision_name == precision:
-                result += NL + "      clblast::RunClient<clblast::TestX" + routine.name + flavour.test_template()
+                result += NL + "      clblast::RunClient<clblast::TestX" + routine.plain_name() + flavour.test_template()
                 result += ">(argc, argv); break;" + NL
                 found = True
         if not found:
@@ -319,7 +323,7 @@ def correctness_test(routine, level_string):
     """Generates the body of a correctness test for a specific routine"""
     result = ""
     result += "#include \"test/correctness/testblas.hpp\"" + NL
-    result += "#include \"test/routines/level" + level_string + "/x" + routine.name + ".hpp\"" + NL + NL
+    result += "#include \"test/routines/level" + level_string + "/x" + routine.lowercase_name() + ".hpp\"" + NL + NL
     result += "// Shortcuts to the clblast namespace" + NL
     result += "using float2 = clblast::float2;" + NL
     result += "using double2 = clblast::double2;" + NL + NL
@@ -328,8 +332,8 @@ def correctness_test(routine, level_string):
     result += "  auto errors = size_t{0};" + NL
     not_first = "false"
     for flavour in routine.flavours:
-        result += "  errors += clblast::RunTests<clblast::TestX" + routine.name + flavour.test_template()
-        result += ">(argc, argv, " + not_first + ", \"" + flavour.name + routine.name.upper() + "\");" + NL
+        result += "  errors += clblast::RunTests<clblast::TestX" + routine.plain_name() + flavour.test_template()
+        result += ">(argc, argv, " + not_first + ", \"" + flavour.name + routine.upper_name() + "\");" + NL
         not_first = "true"
     result += "  if (errors > 0) { return 1; } else { return 0; }" + NL
     result += "}" + NL
diff --git a/scripts/generator/generator/datatype.py b/scripts/generator/generator/datatype.py
index 98874174..cfdbf748 100644
--- a/scripts/generator/generator/datatype.py
+++ b/scripts/generator/generator/datatype.py
@@ -30,17 +30,17 @@ class DataType:
         self.beta_cl = scalars[3]
         self.buffer_type = buffer_type
 
-    def use_alpha(self):
+    def use_alpha(self, postfix=""):
         """Outputs the name of the data-type (alpha/beta), possibly transforming into the right type"""
         if self.alpha_cpp in [D_FLOAT2, D_DOUBLE2]:
-            return self.alpha_cpp + "{alpha.s[0], alpha.s[1]}"
-        return "alpha"
+            return self.alpha_cpp + "{alpha" + postfix + ".s[0], alpha" + postfix + ".s[1]}"
+        return "alpha" + postfix
 
-    def use_beta(self):
+    def use_beta(self, postfix=""):
         """As above, but for beta instead of alpha"""
         if self.beta_cpp in [D_FLOAT2, D_DOUBLE2]:
-            return self.beta_cpp + "{beta.s[0], beta.s[1]}"
-        return "beta"
+            return self.beta_cpp + "{beta" + postfix + ".s[0], beta" + postfix + ".s[1]}"
+        return "beta" + postfix
 
     def use_alpha_opencl(self):
         """As above, but the transformation is in the opposite direction"""
diff --git a/scripts/generator/generator/doc.py b/scripts/generator/generator/doc.py
index c77ec1a0..9c73ffbc 100644
--- a/scripts/generator/generator/doc.py
+++ b/scripts/generator/generator/doc.py
@@ -20,7 +20,7 @@ def generate(routine):
     result = ""
 
     # Routine header
-    result += "x" + routine.name.upper() + ": " + routine.description + NL
+    result += "x" + routine.upper_name() + ": " + routine.description + NL
     result += "-------------" + NL + NL
     result += routine.details + NL + NL
 
@@ -36,7 +36,7 @@ def generate(routine):
     result += "```" + NL + NL
 
     # Routine arguments
-    result += "Arguments to " + routine.name.upper() + ":" + NL + NL
+    result += "Arguments to " + routine.upper_name() + ":" + NL + NL
     for argument in routine.arguments_doc():
         result += "* " + argument + NL
     result += "* `cl_command_queue* queue`: "
@@ -47,7 +47,7 @@ def generate(routine):
 
     # Routine requirements
     if len(routine.requirements_doc()) > 0:
-        result += "Requirements for " + routine.name.upper() + ":" + NL + NL
+        result += "Requirements for " + routine.upper_name() + ":" + NL + NL
         for requirement in routine.requirements_doc():
             result += "* " + requirement + NL
         result += NL
diff --git a/scripts/generator/generator/routine.py b/scripts/generator/generator/routine.py
index 6fcce23b..9b07e671 100644
--- a/scripts/generator/generator/routine.py
+++ b/scripts/generator/generator/routine.py
@@ -12,11 +12,12 @@ import generator.convert as convert
 
 class Routine:
     """Class holding routine-specific information (e.g. name, which arguments, which precisions)"""
-    def __init__(self, implemented, has_tests, level, name, template, flavours, sizes, options,
+    def __init__(self, implemented, has_tests, batched, level, name, template, flavours, sizes, options,
                  inputs, outputs, buffer_sizes, scalars, scratch,
                  description, details, requirements):
         self.implemented = implemented
         self.has_tests = has_tests
+        self.batched = batched
         self.level = level
         self.name = name
         self.template = template
@@ -32,6 +33,69 @@ class Routine:
         self.details = details
         self.requirements = requirements
 
+    def lowercase_name(self):
+        postfix = "batched" if self.batched else ""
+        return self.name + postfix
+
+    def plain_name(self):
+        postfix = "Batched" if self.batched else ""
+        return self.name + postfix
+
+    def capitalized_name(self):
+        postfix = "Batched" if self.batched else ""
+        return self.name.capitalize() + postfix
+
+    def upper_name(self):
+        postfix = "BATCHED" if self.batched else ""
+        return self.name.upper() + postfix
+
+    def b_star(self):
+        return "*" if self.batched else ""
+
+    def b_s(self):
+        return "s" if self.batched else ""
+
+    def batch_count_def(self):
+        return ["const size_t batch_count"] if self.batched else []
+
+    def batch_count_list(self):
+        return ["batch_count"] if self.batched else []
+
+    def batch_count_type(self):
+        return ["const size_t"] if self.batched else []
+
+    def batch_count_doc(self):
+        return ["`const size_t batch_count`: Number of batches. This value must be positive."] if self.batched else []
+
+    def batched_transform_to_cpp(self):
+        result = []
+        for scalar in self.scalars:
+            result.append("auto " + scalar + "s_cpp = std::vector<T>();")
+        for buffer_name in self.inputs + self.outputs:
+            result.append("auto " + buffer_name + "_buffers_cpp = std::vector<Buffer<T>>();")
+        result.append("for (auto batch = size_t{0}; batch < batch_count; ++batch) {")
+        for scalar in self.scalars:
+            result.append("  " + scalar + "s_cpp.push_back(" + scalar + "s[batch]);")
+        for buffer_name in self.inputs + self.outputs:
+            result.append("  " + buffer_name + "_buffers_cpp.push_back(Buffer<T>(" + buffer_name + "_buffers[batch]));")
+        result.append("}")
+        return result
+
+    def batched_transform_to_complex(self, flavour):
+        result = []
+        for scalar in self.scalars:
+            result.append("auto " + scalar + "s_cpp = std::vector<" + flavour.buffer_type + ">();")
+        result.append("for (auto batch = size_t{0}; batch < batch_count; ++batch) {")
+        for scalar in self.scalars:
+            content = scalar
+            if scalar == "alpha":
+                content = flavour.use_alpha(postfix="s[batch]")
+            elif scalar == "beta":
+                content = flavour.use_beta(postfix="s[batch]")
+            result.append("  " + scalar + "s_cpp.push_back(" + content + ");")
+        result.append("}")
+        return result
+
     @staticmethod
     def scalar_buffers_first():
         """List of scalar buffers"""
@@ -127,7 +191,7 @@ class Routine:
 
     def length(self):
         """Retrieves the number of characters in the routine's name"""
-        return len(self.name)
+        return len(self.capitalized_name())
 
     def no_scalars(self):
         """Determines whether or not this routine has scalar arguments (alpha/beta)"""
@@ -135,13 +199,13 @@ class Routine:
 
     def short_names(self):
         """Returns the upper-case names of these routines (all flavours)"""
-        return "/".join([f.name + self.name.upper() for f in self.flavours])
+        return "/".join([f.name + self.upper_name() for f in self.flavours])
 
     def short_names_tested(self):
         """As above, but excludes some"""
-        names = [f.name + self.name.upper() for f in self.flavours]
-        if "H" + self.name.upper() in names:
-            names.remove("H" + self.name.upper())
+        names = [f.name + self.upper_name() for f in self.flavours]
+        if "H" + self.upper_name() in names:
+            names.remove("H" + self.upper_name())
         return "/".join(names)
 
     def buffers_first(self):
@@ -158,7 +222,7 @@ class Routine:
     def buffer(self, name):
         """Retrieves a variable name for a specific input/output vector/matrix (e.g. 'x')"""
         if name in self.inputs or name in self.outputs:
-            a = [name + "_buffer"]
+            a = [name + "_buffer" + self.b_s()]
             b = [name + "_offset"]
             c = [name + "_" + self.postfix(name)] if (name not in self.buffers_without_ld_inc()) else []
             return [", ".join(a + b + c)]
@@ -186,7 +250,7 @@ class Routine:
         """As above but with data-types"""
         prefix = "const " if name in self.inputs else ""
         if name in self.inputs or name in self.outputs:
-            a = [prefix + "cl_mem " + name + "_buffer"]
+            a = [prefix + "cl_mem " + self.b_star() + name + "_buffer" + self.b_s()]
             b = ["const size_t " + name + "_offset"]
             c = ["const size_t " + name + "_" + self.postfix(name)] if name not in self.buffers_without_ld_inc() else []
             return [", ".join(a + b + c)]
@@ -227,7 +291,10 @@ class Routine:
         """As above but with CLCudaAPI buffers"""
         if name in self.inputs or name in self.outputs:
             buffer_type = "unsigned int" if (name in self.index_buffers()) else self.template.buffer_type
-            a = ["Buffer<" + buffer_type + ">(" + name + "_buffer)"]
+            if self.batched:
+                a = [name + "_buffers_cpp"]
+            else:
+                a = ["Buffer<" + buffer_type + ">(" + name + "_buffer)"]
             b = [name + "_offset"]
             c = [name + "_" + self.postfix(name)] if (name not in self.buffers_without_ld_inc()) else []
             return [", ".join(a + b + c)]
@@ -269,7 +336,7 @@ class Routine:
         """As above, but only data-types"""
         prefix = "const " if (name in self.inputs) else ""
         if (name in self.inputs) or (name in self.outputs):
-            a = [prefix + "cl_mem"]
+            a = [prefix + "cl_mem" + self.b_star()]
             b = ["const size_t"]
             c = ["const size_t"] if (name not in self.buffers_without_ld_inc()) else []
             return [", ".join(a + b + c)]
@@ -280,9 +347,9 @@ class Routine:
         prefix = "const " if (name in self.inputs) else ""
         inout = "input" if (name in self.inputs) else "output"
         if (name in self.inputs) or (name in self.outputs):
-            math_name = name.upper() + " matrix" if (name in self.buffers_matrix()) else name + " vector"
+            math_name = name.upper() + " matrix" + self.b_s() if (name in self.buffers_matrix()) else name + " vector" + self.b_s()
             inc_ld_description = "Leading dimension " if (name in self.buffers_matrix()) else "Stride/increment "
-            a = ["`" + prefix + "cl_mem " + name + "_buffer`: OpenCL buffer to store the " + inout + " " + math_name + "."]
+            a = ["`" + prefix + "cl_mem " + self.b_star() + name + "_buffer" + self.b_s() + "`: OpenCL buffer" + self.b_s() + " to store the " + inout + " " + math_name + "."]
             b = ["`const size_t " + name + "_offset`: The offset in elements from the start of the " + inout + " " + math_name + "."]
             if name not in self.buffers_without_ld_inc():
                 c = ["`const size_t " + name + "_" + self.postfix(name) + "`: " +
@@ -295,6 +362,8 @@ class Routine:
     def scalar(self, name):
         """Retrieves the name of a scalar (alpha/beta)"""
         if name in self.scalars:
+            if self.batched:
+                return [name + "s_cpp"]
             return [name]
         return []
 
@@ -314,8 +383,12 @@ class Routine:
         """Retrieves the use of a scalar (alpha/beta)"""
         if name in self.scalars:
             if name == "alpha":
+                if self.batched:
+                    return ["alphas_cpp.data()"]
                 return [flavour.use_alpha()]
             elif name == "beta":
+                if self.batched:
+                    return ["betas_cpp.data()"]
                 return [flavour.use_beta()]
             return [name]
         return []
@@ -342,16 +415,16 @@ class Routine:
         """Retrieves the definition of a scalar (alpha/beta)"""
         if name in self.scalars:
             if name == "alpha":
-                return ["const " + flavour.alpha_cl + " " + name]
-            return ["const " + flavour.beta_cl + " " + name]
+                return ["const " + flavour.alpha_cl + " " + self.b_star() + name + self.b_s()]
+            return ["const " + flavour.beta_cl + " " + self.b_star() + name + self.b_s()]
         return []
 
     def scalar_def_plain(self, name, flavour):
         """As above, but without 'cl_' prefix"""
         if name in self.scalars:
             if name == "alpha":
-                return ["const " + flavour.alpha_cpp + " " + name]
-            return ["const " + flavour.beta_cpp + " " + name]
+                return ["const " + flavour.alpha_cpp + " " + self.b_star() + name + self.b_s()]
+            return ["const " + flavour.beta_cpp + " " + self.b_star() + name + self.b_s()]
         return []
 
     def scalar_def_void(self, name, flavour):
@@ -368,16 +441,16 @@ class Routine:
         """Retrieves the type of a scalar (alpha/beta)"""
         if name in self.scalars:
             if name == "alpha":
-                return ["const " + flavour.alpha_cpp]
-            return ["const " + flavour.beta_cpp]
+                return ["const " + flavour.alpha_cpp + self.b_star()]
+            return ["const " + flavour.beta_cpp + self.b_star()]
         return []
 
     def scalar_doc(self, name):
         """Retrieves the documentation of a scalar"""
         if name in self.scalars:
             if name == "alpha":
-                return ["`const " + self.template.alpha_cpp + " " + name + "`: Input scalar constant."]
-            return ["`const " + self.template.beta_cpp + " " + name + "`: Input scalar constant."]
+                return ["`const " + self.template.alpha_cpp + " " + self.b_star() + name + self.b_s() + "`: Input scalar constant" + self.b_s() + "."]
+            return ["`const " + self.template.beta_cpp + " " + self.b_star() + name + self.b_s() + "`: Input scalar constant" + self.b_s() + "."]
         return []
 
     def scalar_create_cpp(self, flavour):
@@ -507,7 +580,8 @@ class Routine:
                 self.scalar("beta") +
                 list(chain(*[self.buffer_clcudaapi(b) for b in self.buffers_second()])) +
                 list(chain(*[self.buffer_clcudaapi(b) for b in self.scalar_buffers_second()])) +
-                list(chain(*[self.scalar(s) for s in self.other_scalars()])))
+                list(chain(*[self.scalar(s) for s in self.other_scalars()])) +
+                self.batch_count_list())
 
     def arguments_cast(self, flavour, indent):
         """As above, but with CLBlast casts"""
@@ -518,7 +592,8 @@ class Routine:
                 self.scalar_use("beta", flavour) +
                 list(chain(*[self.buffer(b) for b in self.buffers_second()])) +
                 list(chain(*[self.buffer(b) for b in self.scalar_buffers_second()])) +
-                list(chain(*[self.scalar_use(s, flavour) for s in self.other_scalars()])))
+                list(chain(*[self.scalar_use(s, flavour) for s in self.other_scalars()])) +
+                self.batch_count_list())
 
     def arguments_netlib(self, flavour, indent):
         """As above, but for the Netlib CBLAS API"""
@@ -561,7 +636,8 @@ class Routine:
                 self.scalar_def("beta", flavour) +
                 list(chain(*[self.buffer_def(b) for b in self.buffers_second()])) +
                 list(chain(*[self.buffer_def(b) for b in self.scalar_buffers_second()])) +
-                list(chain(*[self.scalar_def(s, flavour) for s in self.other_scalars()])))
+                list(chain(*[self.scalar_def(s, flavour) for s in self.other_scalars()])) +
+                self.batch_count_def())
 
     def arguments_def_netlib(self, flavour):
         """As above, but for the Netlib CBLAS API"""
@@ -574,6 +650,7 @@ class Routine:
                 list(chain(*[self.scalar_def(s, flavour) for s in self.other_scalars()])))
         if self.name in self.routines_scalar_no_return():
             result += list(chain(*[self.buffer_def_pointer(b, flavour) for b in self.scalar_buffers_first()]))
+        result += self.batch_count_def()
         return result
 
     def arguments_def_c(self, flavour):
@@ -585,7 +662,8 @@ class Routine:
                 self.scalar_def("beta", flavour) +
                 list(chain(*[self.buffer_def(b) for b in self.buffers_second()])) +
                 list(chain(*[self.buffer_def(b) for b in self.scalar_buffers_second()])) +
-                list(chain(*[self.scalar_def(s, flavour) for s in self.other_scalars()])))
+                list(chain(*[self.scalar_def(s, flavour) for s in self.other_scalars()])) +
+                self.batch_count_def())
 
     def arguments_def_wrapper_clblas(self, flavour):
         """As above, but clBLAS wrapper plain data-types"""
@@ -618,7 +696,8 @@ class Routine:
                 self.scalar_type("beta", flavour) +
                 list(chain(*[self.buffer_type(b) for b in self.buffers_second()])) +
                 list(chain(*[self.buffer_type(b) for b in self.scalar_buffers_second()])) +
-                list(chain(*[self.scalar_type(s, flavour) for s in self.other_scalars()])))
+                list(chain(*[self.scalar_type(s, flavour) for s in self.other_scalars()])) +
+                self.batch_count_type())
 
     def arguments_doc(self):
         """Retrieves a combination of all the argument types"""
@@ -630,7 +709,8 @@ class Routine:
                 self.scalar_doc("beta") +
                 list(chain(*[self.buffer_doc(b) for b in self.buffers_second()])) +
                 list(chain(*[self.buffer_doc(b) for b in self.scalar_buffers_second()])) +
-                list(chain(*[self.scalar_doc(s) for s in self.other_scalars()])))
+                list(chain(*[self.scalar_doc(s) for s in self.other_scalars()])) +
+                self.batch_count_doc())
 
     def requirements_doc(self):
         """Retrieves a list of routine requirements for documentation"""
@@ -640,7 +720,7 @@ class Routine:
         """Retrieves the C++ templated definition for a routine"""
         indent = " " * (spaces + self.length())
         result = "template <" + self.template.name + ">\n"
-        result += "StatusCode " + self.name.capitalize() + "("
+        result += "StatusCode " + self.capitalized_name() + "("
         result += (",\n" + indent).join([a for a in self.arguments_def(self.template)])
         result += ",\n" + indent + "cl_command_queue* queue, cl_event* event" + default_event + ")"
         return result
@@ -649,7 +729,7 @@ class Routine:
         """As above, but now without variable names"""
         indent = " " * (spaces + self.length())
         result = "template <" + self.template.name + ">\n"
-        result += "StatusCode " + self.name.capitalize() + "("
+        result += "StatusCode " + self.capitalized_name() + "("
         result += (",\n" + indent).join([a for a in self.arguments_type(self.template)])
         result += ",\n" + indent + "cl_command_queue*, cl_event*)"
         return result
@@ -657,7 +737,7 @@ class Routine:
     def routine_header_c(self, flavour, spaces, extra_qualifier):
         """As above, but now for C"""
         indent = " " * (spaces + self.length())
-        result = "CLBlastStatusCode" + extra_qualifier + " CLBlast" + flavour.name + self.name + "("
+        result = "CLBlastStatusCode" + extra_qualifier + " CLBlast" + flavour.name + self.plain_name() + "("
         result += (",\n" + indent).join([a for a in self.arguments_def_c(flavour)])
         result += ",\n" + indent + "cl_command_queue* queue, cl_event* event)"
         return result
@@ -677,6 +757,8 @@ class Routine:
         if self.name in self.routines_scalar_no_return():
             routine_name += "_sub"
             indent += "    "
+        if self.batched:
+            routine_name += "batched"
         result = return_type + extra_qualifier + " cblas_" + flavour.name.lower() + routine_name + "("
         result += (",\n" + indent).join([a for a in self.arguments_def_netlib(flavour)]) + ")"
         return result