[onert-micro] Add SVDF kernel

This pr adds SVDF kernel.

ONE-DCO-1.0-Signed-off-by: Artem Balyshev <[email protected]>

onert-micro/onert-micro/include/pal/common/PALSVDFCommon.h

@@ -0,0 +1,195 @@
+/*
+ * Copyright (c) 2024 Samsung Electronics Co., Ltd. All Rights Reserved
+ * Copyright 2019 The TensorFlow Authors. All Rights Reserved.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *    http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef ONERT_MICRO_EXECUTE_PAL_SVDF_COMMON_H
+#define ONERT_MICRO_EXECUTE_PAL_SVDF_COMMON_H
+
+#include "PALUtils.h"
+
+#include "core/OMKernelData.h"
+
+#include <cmath>
+
+namespace onert_micro
+{
+namespace execute
+{
+namespace pal
+{
+
+namespace
+{
+// Returns the floating point value for a fused activation:
+inline float activationValFloat(const circle::ActivationFunctionType act, float a)
+{
+  switch (act)
+  {
+    case circle::ActivationFunctionType_NONE:
+      return a;
+    case circle::ActivationFunctionType_RELU:
+      return std::max(0.0f, a);
+    case circle::ActivationFunctionType_RELU_N1_TO_1:
+      return std::max(-1.0f, std::min(a, 1.0f));
+    case circle::ActivationFunctionType_RELU6:
+      return std::max(0.0f, std::min(a, 6.0f));
+    case circle::ActivationFunctionType_TANH:
+      return std::tanh(a);
+    case circle::ActivationFunctionType_SIGN_BIT:
+      return std::signbit(a);
+    default:
+      assert(false && "Not supported");
+  }
+  return 0.0f; // To indicate an unsupported activation (i.e. when a new fused
+  // activation is added to the enum and not handled here).
+}
+
+static inline void
+applyTimeWeightsBiasAndActivation(int batch_size, int memory_size, int num_filters, int num_units,
+                                  int rank, const float *const weights_time_ptr,
+                                  const float *const bias_ptr,
+                                  circle::ActivationFunctionType activation, float *const state_ptr,
+                                  float *const scratch_ptr, float *const output_ptr)
+{
+  // Compute matmul(activation_state, weights_time).
+  for (int b = 0; b < batch_size; ++b)
+  {
+    // Perform batched vector dot product:
+    float *scratch_ptr_batch = scratch_ptr + b * num_filters;
+    const float *vector1_ptr = weights_time_ptr;
+    const float *vector2_ptr = state_ptr + b * memory_size * num_filters;
+    for (int i = 0; i < num_filters; ++i)
+    {
+      *scratch_ptr_batch = 0.f;
+      for (int j = 0; j < memory_size; ++j)
+      {
+        *scratch_ptr_batch += *vector1_ptr++ * *vector2_ptr++;
+      }
+      scratch_ptr_batch++;
+    }
+  }
+
+  // Initialize output with bias if provided.
+  if (bias_ptr)
+  {
+    // VectorBatchVectorAssign
+    for (int i = 0; i < batch_size; ++i)
+    {
+      float *output_data = output_ptr + i * num_units;
+      const float *bias_data = bias_ptr;
+      for (int j = 0; j < num_units; ++j)
+      {
+        *output_data++ = *bias_data++;
+      }
+    }
+  }
+  else
+  {
+    float *output_data = output_ptr;
+    for (int i = 0; i < batch_size * num_units; ++i)
+    {
+      *output_data++ = 0.0f;
+    }
+  }
+
+  // Reduction sum.
+  for (int b = 0; b < batch_size; ++b)
+  {
+    float *output_ptr_batch = output_ptr + b * num_units;
+    float *scratch_ptr_batch = scratch_ptr + b * num_filters;
+
+    // Reduction sum vector
+    for (int i = 0; i < num_units; ++i)
+    {
+      for (int j = 0; j < rank; j++)
+      {
+        output_ptr_batch[i] += *scratch_ptr_batch++;
+      }
+    }
+  }
+
+  // Apply activation.
+  for (int b = 0; b < batch_size; ++b)
+  {
+    float *output_ptr_batch = output_ptr + b * num_units;
+    for (int i = 0; i < num_units; ++i)
+    {
+      *output_ptr_batch = activationValFloat(activation, *output_ptr_batch);
+      ++output_ptr_batch;
+    }
+  }
+}
+
+} // namespace
+
+OMStatus SVDF(const float *input_data, const float *weights_feature_data,
+              const float *weights_time_data, const float *bias_data, float *state_data,
+              float *scratch_data, float *output_data, const int rank, const int input_size,
+              const int batch_size, const int num_filters, const int num_units,
+              const int memory_size, const circle::ActivationFunctionType activation)
+{
+  // Left shift the activation_state.
+  {
+    float *new_state_start = state_data;
+    const float *old_state_start = state_data + 1;
+    const float *old_state_end = state_data + batch_size * num_filters * memory_size;
+    while (old_state_start != old_state_end)
+    {
+      *new_state_start++ = *old_state_start++;
+    }
+  }
+
+  // Note: no need to clear the latest activation, matmul is not accumulative.
+
+  // Compute conv1d(inputs, weights_feature).
+  // The activation_state's rightmost column is used to save current cycle
+  // activation. This is achieved by starting at state_ptr[memory_size - 1] and
+  // having the stride equal to memory_size.
+
+  // Perform batched matrix vector multiply operation:
+  {
+    const float *matrix = weights_feature_data;
+    const float *vector = input_data;
+    float *result = &state_data[memory_size - 1];
+    float *result_in_batch = result;
+    for (int i = 0; i < batch_size; ++i)
+    {
+      const float *matrix_ptr = matrix;
+      for (int j = 0; j < num_filters; ++j)
+      {
+        float dot_prod = 0.0f;
+        const float *vector_in_batch = vector + i * input_size;
+        for (int k = 0; k < input_size; ++k)
+        {
+          dot_prod += *matrix_ptr++ * *vector_in_batch++;
+        }
+        *result_in_batch = dot_prod;
+        result_in_batch += memory_size;
+      }
+    }
+  }
+
+  applyTimeWeightsBiasAndActivation(batch_size, memory_size, num_filters, num_units, rank,
+                                    weights_time_data, bias_data, activation, state_data,
+                                    scratch_data, output_data);
+  return Ok;
+}
+
+} // namespace pal
+} // namespace execute
+} // namespace onert_micro
+
+#endif // ONERT_MICRO_EXECUTE_PAL_SVDF_COMMON_H

onert-micro/onert-micro/include/pal/mcu/KernelsToBuild.lst

@@ -75,7 +75,7 @@ REGISTER_KERNEL(TRANSPOSE_CONV, TransposeConv)
 REGISTER_KERNEL(SOFTMAX, Softmax)
 #/*REGISTER_KERNEL(SUM, Sum)*/
 #/*REGISTER_KERNEL(SELECT_V2, SelectV2)*/
-#/*REGISTER_KERNEL(SVDF, SVDF)*/
+REGISTER_KERNEL(SVDF, SVDF)
 REGISTER_KERNEL(WHILE, While)
 #/*REGISTER_KERNEL(UNIDIRECTIONAL_SEQUENCE_LSTM, UnidirectionalSequenceLSTM)*/
 #/*REGISTER_KERNEL(RESIZE_BILINEAR, ResizeBilinear)*/

onert-micro/onert-micro/include/pal/mcu/PALSVDF.h

@@ -0,0 +1,23 @@
+/*
+ * Copyright (c) 2024 Samsung Electronics Co., Ltd. All Rights Reserved
+ * Copyright 2019 The TensorFlow Authors. All Rights Reserved.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *    http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef ONERT_MICRO_EXECUTE_PAL_SVDF_H
+#define ONERT_MICRO_EXECUTE_PAL_SVDF_H
+
+#include "PALSVDFCommon.h"
+
+#endif // ONERT_MICRO_EXECUTE_PAL_SVDF_H