aboutsummaryrefslogtreecommitdiff
path: root/src/core/CL/kernels/CLGEMMMatrixMultiplyKernel.cpp
diff options
context:
space:
mode:
authorIsabella Gottardi <isabella.gottardi@arm.com>2018-10-01 12:26:28 +0100
committerAnthony Barbier <anthony.barbier@arm.com>2018-11-02 16:55:19 +0000
commitf02e52796c3e2bd4a88b696cbe8415cd36884c12 (patch)
tree42c29e2d49ba364f03d429108eb5bde9ee085b9c /src/core/CL/kernels/CLGEMMMatrixMultiplyKernel.cpp
parenteddff5a6f43638a27b60564037324f040339faf5 (diff)
downloadComputeLibrary-f02e52796c3e2bd4a88b696cbe8415cd36884c12.tar.gz
COMPMID-1607 - (Nightly) CLGEMMLowpMatrixMultiplyCore errors and mismatches
Change-Id: I5f2e6843526cb154176a5b113627d4f36c3a8edd Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/150967 Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Tested-by: bsgcomp <bsgcomp@arm.com>
Diffstat (limited to 'src/core/CL/kernels/CLGEMMMatrixMultiplyKernel.cpp')
-rw-r--r--src/core/CL/kernels/CLGEMMMatrixMultiplyKernel.cpp8
1 files changed, 6 insertions, 2 deletions
diff --git a/src/core/CL/kernels/CLGEMMMatrixMultiplyKernel.cpp b/src/core/CL/kernels/CLGEMMMatrixMultiplyKernel.cpp
index 715edae606..c8bcb37b9c 100644
--- a/src/core/CL/kernels/CLGEMMMatrixMultiplyKernel.cpp
+++ b/src/core/CL/kernels/CLGEMMMatrixMultiplyKernel.cpp
@@ -120,8 +120,12 @@ inline std::pair<Status, Window> validate_and_configure_window(ITensorInfo *inpu
reinterpret_output_as_3d = false;
}
+ const GEMMReshapeInfo reshape_info_to_use(reshape_info.m(), reshape_info.n(), reshape_info.k(), reshape_info.mult_transpose1xW_width(),
+ reshape_info.mult_interleave4x4_height(), reinterpret_output_as_3d ? reshape_info.depth_output_gemm3d() : 1, reinterpret_input_as_3d);
+
// Output tensor auto inizialitation if not yet initialized
- auto_init_if_empty(*output, input0->clone()->set_tensor_shape(compute_mm_shape(*input0, *input1, is_interleaved_transposed, reshape_info)));
+ auto_init_if_empty(*output, input0->clone()->set_tensor_shape(compute_mm_shape(*input0, *input1, is_interleaved_transposed,
+ reshape_info_to_use)));
TensorInfo tmp_info(*output);
@@ -137,7 +141,7 @@ inline std::pair<Status, Window> validate_and_configure_window(ITensorInfo *inpu
if(is_interleaved_transposed)
{
// reinterpret_input_as_3d is not supported if is_interleaved_transposed is set
- ARM_COMPUTE_ERROR_ON(reshape_info.reinterpret_input_as_3d());
+ ARM_COMPUTE_ERROR_ON(reinterpret_input_as_3d);
// Configure kernel window
num_elems_processed_per_iteration_x = max_cl_vector_width / data_size_from_type(data_type);