[COMPMID-1353] Add support for 4D Softmax layer on OpenCL

Change-Id: I4342d4240fe5b1aab234c015684a1216c3990a5f
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/145631
Tested-by: Jenkins <bsgcomp@arm.com>
Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>

1 files changed, 88 insertions, 11 deletions

diff --git a/src/runtime/CL/functions/CLSoftmaxLayer.cpp b/src/runtime/CL/functions/CLSoftmaxLayer.cpp
index 7a20d9f94b..3a7d6c770b 100644
--- a/src/runtime/CL/functions/CLSoftmaxLayer.cpp
+++ b/src/runtime/CL/functions/CLSoftmaxLayer.cpp
@@ -29,14 +29,32 @@
 #include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/Types.h"
 #include "arm_compute/core/Utils.h"
+#include "arm_compute/core/utils/misc/ShapeCalculator.h"
 #include "arm_compute/runtime/CL/CLMemoryGroup.h"
 #include "arm_compute/runtime/CL/CLScheduler.h"
 
-using namespace arm_compute;
-
+namespace arm_compute
+{
 CLSoftmaxLayer::CLSoftmaxLayer(std::shared_ptr<IMemoryManager> memory_manager)
-    : _memory_group(std::move(memory_manager)), _max_shift_exp_sum_kernel(), _norm_kernel(), _max(), _sum(), _tmp()
+    : _memory_group(std::move(memory_manager)), _max_shift_exp_sum_kernel(), _norm_kernel(), _flatten_kernel(), _reshape_kernel(), _max(), _sum(), _tmp(), _input_flat(), _output_flat(),
+      _needs_flattening(false)
+{
+}
+
+void CLSoftmaxLayer::configure_flatten_kernel(const ICLTensor *input, const ICLTensor *output)
 {
+    // Flatten the input
+    const TensorShape shape_flatten = misc::shape_calculator::compute_flatten_shape(input->info());
+
+    // Initialize the flat input
+    _input_flat.allocator()->init(input->info()->clone()->set_is_resizable(true).reset_padding().set_tensor_shape(shape_flatten));
+
+    // Configure the flatten_kernel
+    _flatten_kernel.configure(input, &_input_flat);
+
+    // We need to init the output tensor here. Indeed, the reshape kernel expects
+    // both tensors to be already initialized
+    auto_init_if_empty(*output->info(), *input->info()->clone());
 }
 
 void CLSoftmaxLayer::configure(const ICLTensor *input, ICLTensor *output, float beta)
@@ -45,13 +63,32 @@ void CLSoftmaxLayer::configure(const ICLTensor *input, ICLTensor *output, float
     ARM_COMPUTE_ERROR_ON_NULLPTR(input, output);
     ARM_COMPUTE_ERROR_THROW_ON(CLSoftmaxLayer::validate(input->info(), output->info()));
 
+    _needs_flattening = input->info()->num_dimensions() > 2;
+
+    // If we are dealing with a 4D tensor, we will:
+    // - Flatten the input, so that we end up with a [width*height*depth] * batches 2D tensor
+    // - Execute all the pipeline (reduction + normalization) on the flattened tensor
+    // - Reshape the flattened output into the real output
+    if(_needs_flattening)
+    {
+        // Add to the memory manager _input_flat
+        _memory_group.manage(&_input_flat);
+
+        // Cofigure  _flatten_kernel and _input_flat
+        configure_flatten_kernel(input, output);
+    }
+
+    // We want to deal with a 2D input. Either it is the flattened version of the original input (4D case)
+    // or it is the original input case (2D case)
+    const ICLTensor *input_2D = (_needs_flattening ? &_input_flat : input);
+
     // Create intermediate tensors shapes
-    const TensorInfo input_info    = input->info()->clone()->reset_padding().set_is_resizable(true);
-    DataType         tmp_data_type = is_data_type_quantized_asymmetric(input->info()->data_type()) ? DataType::S32 : input->info()->data_type();
-    TensorInfo       tensor_info_tmp(input_info.clone()->set_data_type(tmp_data_type));
+    TensorInfo input_info    = input_2D->info()->clone()->reset_padding().set_is_resizable(true);
+    DataType   tmp_data_type = is_data_type_quantized_asymmetric(input_2D->info()->data_type()) ? DataType::S32 : input_2D->info()->data_type();
+    TensorInfo tensor_info_tmp(input_info.clone()->set_data_type(tmp_data_type));
     _tmp.allocator()->init(tensor_info_tmp);
 
-    TensorShape max_sum_shape = input->info()->tensor_shape();
+    TensorShape max_sum_shape = input_2D->info()->tensor_shape();
     max_sum_shape.set(0, 1);
     _max.allocator()->init(input_info.clone()->set_tensor_shape(max_sum_shape));
     _sum.allocator()->init(input_info.clone()->set_tensor_shape(max_sum_shape).set_data_type(tmp_data_type));
@@ -65,8 +102,28 @@ void CLSoftmaxLayer::configure(const ICLTensor *input, ICLTensor *output, float
     _memory_group.manage(&_sum);
 
     // Configure kernels
-    _max_shift_exp_sum_kernel.configure(input, &_max, &_tmp, &_sum, beta);
-    _norm_kernel.configure(&_tmp, &_sum, output, beta);
+    _max_shift_exp_sum_kernel.configure(input_2D, &_max, &_tmp, &_sum, beta);
+
+    if(_needs_flattening)
+    {
+        // Add to the memory manager _output_flat
+        _memory_group.manage(&_output_flat);
+
+        // The normalization kernel stores the result in a flat output tensor
+        _norm_kernel.configure(&_tmp, &_sum, &_output_flat, beta);
+
+        // Reshape the flat output into a the requested (4D) output
+        _reshape_kernel.configure(&_output_flat, output);
+
+        // Allocate the intermediate flat tensors
+        _input_flat.allocator()->allocate();
+        _output_flat.allocator()->allocate();
+    }
+    else
+    {
+        // Softmax 2D case
+        _norm_kernel.configure(&_tmp, &_sum, output, beta);
+    }
 
     // Allocate intermediate buffers
     _tmp.allocator()->allocate();
@@ -77,7 +134,7 @@ void CLSoftmaxLayer::configure(const ICLTensor *input, ICLTensor *output, float
 Status CLSoftmaxLayer::validate(const ITensorInfo *input, const ITensorInfo *output)
 {
     ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(input, output);
-    ARM_COMPUTE_RETURN_ERROR_ON_MSG(input->num_dimensions() > 2, "Only 2D inputs are supported");
+    ARM_COMPUTE_RETURN_ERROR_ON_MSG(input->num_dimensions() > 4, "Only up to 4 dimensions are supported");
 
     // Create intermediate tensor info
     DataType   tmp_data_type = is_data_type_quantized_asymmetric(input->data_type()) ? DataType::S32 : input->data_type();
@@ -88,6 +145,14 @@ Status CLSoftmaxLayer::validate(const ITensorInfo *input, const ITensorInfo *out
     TensorInfo tensor_info_max(input->clone()->set_tensor_shape(max_sum_shape).set_is_resizable(true));
     TensorInfo tensor_info_sum(input->clone()->set_tensor_shape(max_sum_shape).set_data_type(tmp_data_type).set_quantization_info(QuantizationInfo()).set_is_resizable(true));
 
+    const TensorShape shape_flatten = misc::shape_calculator::compute_flatten_shape(input);
+    TensorInfo        tensor_info_flat(input->clone()->set_tensor_shape(shape_flatten).set_is_resizable(true));
+
+    if(input->num_dimensions() > 2) // needs flattening
+    {
+        ARM_COMPUTE_RETURN_ON_ERROR(CLFlattenLayerKernel::validate(input, &tensor_info_flat));
+    }
+
     ARM_COMPUTE_RETURN_ON_ERROR(CLLogits1DMaxShiftExpSumKernel::validate(input, &tensor_info_max, &tensor_info_tmp, &tensor_info_sum));
     ARM_COMPUTE_RETURN_ON_ERROR(CLLogits1DNormKernel::validate(&tensor_info_tmp, &tensor_info_sum, output));
 
@@ -97,9 +162,21 @@ Status CLSoftmaxLayer::validate(const ITensorInfo *input, const ITensorInfo *out
 void CLSoftmaxLayer::run()
 {
     _memory_group.acquire();
+    if(_needs_flattening)
+    {
+        CLScheduler::get().enqueue(_flatten_kernel, false);
+    }
 
     CLScheduler::get().enqueue(_max_shift_exp_sum_kernel, false);
-    CLScheduler::get().enqueue(_norm_kernel);
+    CLScheduler::get().enqueue(_norm_kernel, !_needs_flattening);
 
+    if(_needs_flattening)
+    {
+        CLScheduler::get().enqueue(_reshape_kernel, true);
+    }
+
+    // Relase intermediate buffers
     _memory_group.release();
 }
+
+} // namespace arm_compute