diff options
author | Sang-Hoon Park <sang-hoon.park@arm.com> | 2019-11-11 17:33:28 +0000 |
---|---|---|
committer | Sang-Hoon Park <sang-hoon.park@arm.com> | 2019-11-12 11:23:42 +0000 |
commit | eaa01ab593428bc7267ebbe107b2d813a11b64b5 (patch) | |
tree | 80e0af6b7ed0a82387080d6abb6c0a04d9f1d509 /src/core/CL/kernels/CLReductionOperationKernel.cpp | |
parent | 75041a1cb81c59a5a5ddd9b708476c0142362d9e (diff) | |
download | ComputeLibrary-eaa01ab593428bc7267ebbe107b2d813a11b64b5.tar.gz |
COMPMID-2671 use Signed32 for default output data type of ArgMinMax
Signed32 is used as data types before and after reshaping of ArgMinMax.
Change-Id: I230af43a931d4e106de6c72f716ced1dab511084
Signed-off-by: Sang-Hoon Park <sang-hoon.park@arm.com>
Reviewed-on: https://review.mlplatform.org/c/2262
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Giorgio Arena <giorgio.arena@arm.com>
Diffstat (limited to 'src/core/CL/kernels/CLReductionOperationKernel.cpp')
-rw-r--r-- | src/core/CL/kernels/CLReductionOperationKernel.cpp | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/src/core/CL/kernels/CLReductionOperationKernel.cpp b/src/core/CL/kernels/CLReductionOperationKernel.cpp index a085ab1683..cbf3923243 100644 --- a/src/core/CL/kernels/CLReductionOperationKernel.cpp +++ b/src/core/CL/kernels/CLReductionOperationKernel.cpp @@ -83,7 +83,7 @@ std::tuple<Status, Window> validate_and_configure_window(ITensorInfo *input, ITe // Output tensor auto initialization if not yet initialized const bool is_arg_min_max = (op == ReductionOperation::ARG_IDX_MIN || op == ReductionOperation::ARG_IDX_MAX); const TensorShape output_shape = arm_compute::misc::shape_calculator::compute_reduced_shape(input->tensor_shape(), axis, !is_arg_min_max); - const DataType output_data_type = is_arg_min_max ? DataType::U32 : input->data_type(); + const DataType output_data_type = is_arg_min_max ? DataType::S32 : input->data_type(); auto_init_if_empty(*output, input->clone()->set_tensor_shape(output_shape).set_data_type(output_data_type).reset_padding().set_is_resizable(true)); const unsigned int num_elems_processed_per_iteration = (is_data_type_quantized(input->data_type()) && (axis == 0)) ? 1 : 16; |