diff options
| author | Pablo Tello <pablo.tello@arm.com> | 2017-06-28 17:27:56 +0100 |
|---|---|---|
| committer | Anthony Barbier <anthony.barbier@arm.com> | 2018-09-17 14:15:39 +0100 |
| commit | 221f38176b0d4dbc212441779d9bbac3cc0eecfa (patch) | |
| tree | e838d673b35c5b40d4b484a3645cc7ae3c9d3ecc /tests/validation/Reference.cpp | |
| parent | 6410fb2a14427713251f5d97144ac5d4f17c988c (diff) | |
| download | ComputeLibrary-221f38176b0d4dbc212441779d9bbac3cc0eecfa.tar.gz | |
COMPMID-421: Fixed FP16 support in Neon GEMM.
Fixed GEMM FP16 problem with matrices that are not multiple of 32.
Added a new test suite NEON/GEMM/Float16/SmallGEMM.
Implemented FP16 function to multiply vector by a matrix.
Change-Id: Ie6c692885a48d0206bd6fe748332fa83bc286d67
Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79118
Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com>
Diffstat (limited to 'tests/validation/Reference.cpp')
| -rw-r--r-- | tests/validation/Reference.cpp | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/tests/validation/Reference.cpp b/tests/validation/Reference.cpp index 0518819173..62dfcba37e 100644 --- a/tests/validation/Reference.cpp +++ b/tests/validation/Reference.cpp @@ -335,7 +335,7 @@ RawTensor Reference::compute_reference_gemm(const TensorShape &src_shape1, const RawTensor dst = library->get(dst_shape, dt, 1, fixed_point_position); // Fill reference - if(dt == DataType::F32) + if(dt == DataType::F16 || dt == DataType::F32) { std::uniform_real_distribution<> distribution(-1.0f, 1.0f); library->fill(src1, distribution, 0); |