diff options
| author | Cedric Nugteren <web@cedricnugteren.nl> | 2017-03-10 21:24:35 +0100 |
|---|---|---|
| committer | Cedric Nugteren <web@cedricnugteren.nl> | 2017-03-10 21:24:35 +0100 |
| commit | 49e04c7fce8fed45559e143137cef3a1a36328cc (patch) | |
| tree | f73a5c280f12cc5e38f6d4fd4e853b8b8e1aa432 /scripts/generator/generator.py | |
| parent | de3500ed18ddb39261ffa270f460909571276462 (diff) | |
Added API and test infrastructure for the batched GEMM routine
Diffstat (limited to 'scripts/generator/generator.py')
| -rwxr-xr-x | scripts/generator/generator.py | 3 |
1 files changed, 2 insertions, 1 deletions
diff --git a/scripts/generator/generator.py b/scripts/generator/generator.py index 8dd5fc0c..086b27d3 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 = [122, 76, 126, 23, 29, 41, 65, 32] +HEADER_LINES = [123, 76, 126, 23, 29, 41, 65, 32] FOOTER_LINES = [25, 138, 27, 38, 6, 6, 9, 2] HEADER_LINES_DOC = 0 FOOTER_LINES_DOC = 63 @@ -163,6 +163,7 @@ 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.", []), + Routine(True, True, True, "x", "gemm", T, [S,D,C,Z,H], ["m","n","k"], ["layout","a_transpose","b_transpose"], ["a","b"], ["c"], [amk,bkn,cmn], ["alpha","beta"], "", "Batched version of GEMM", "As GEMM, but multiple operations are batched together for better performance.", [ald_transa_m_k, bld_transb_k_n, cld_m]), ]] |