You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

186 lines
8.1 KiB

/*
* Copyright (C) 2019 The Android Open Source Project
*
* 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.
*/
#define LOG_TAG "Operations"
#include <vector>
#include "OperationResolver.h"
#include "Tracing.h"
#ifdef NN_INCLUDE_CPU_IMPLEMENTATION
#include <tensorflow/lite/kernels/internal/optimized/legacy_optimized_ops.h>
#include <tensorflow/lite/kernels/internal/reference/reference_ops.h>
#include "CpuOperationUtils.h"
#endif // NN_INCLUDE_CPU_IMPLEMENTATION
namespace android {
namespace nn {
namespace transpose {
constexpr char kOperationName[] = "TRANSPOSE";
constexpr uint32_t kNumInputs = 2;
constexpr uint32_t kInputTensor = 0;
constexpr uint32_t kPermTensor = 1;
constexpr uint32_t kNumOutputs = 1;
constexpr uint32_t kOutputTensor = 0;
#ifdef NN_INCLUDE_CPU_IMPLEMENTATION
namespace {
template <typename T>
bool transposeGeneric(const T* inputData, const Shape& inputShape, const int32_t* perm,
const Shape& permShape, T* outputData, const Shape& outputShape) {
NNTRACE_TRANS("transposeGeneric");
// Reverse the permuted axes and convert to 4D due to the way Dims are
// constructed.
const int32_t kOutputDimensionNum = 4;
// permData can be NO_VALUE representing a regular 2D matrix transpose
int32_t permSize = perm == nullptr ? 2 : static_cast<int32_t>(getSizeOfDimension(permShape, 0));
int32_t perm_tmp[2] = {1, 0};
if (perm == nullptr) {
perm = perm_tmp;
}
int32_t reversed_perm[kOutputDimensionNum];
for (int32_t output_k = 0, input_k = permSize - 1; output_k < permSize; ++output_k, --input_k) {
reversed_perm[output_k] = permSize - perm[input_k] - 1;
}
for (int32_t k = permSize; k < kOutputDimensionNum; ++k) {
reversed_perm[k] = k;
}
NNTRACE_COMP_SWITCH("reference_ops::Transpose");
tflite::reference_ops::Transpose(inputData, convertShapeToDims(inputShape), outputData,
convertShapeToDims(outputShape), reversed_perm);
return true;
}
} // namespace
#endif // NN_INCLUDE_CPU_IMPLEMENTATION
Result<Version> validate(const IOperationValidationContext* context) {
NN_RET_CHECK_EQ(context->getNumInputs(), kNumInputs);
NN_RET_CHECK_EQ(context->getNumOutputs(), kNumOutputs);
const OperandType inputType = context->getInputType(kInputTensor);
auto minSupportedVersion = Version::ANDROID_OC_MR1;
if (inputType == OperandType::TENSOR_FLOAT32 || inputType == OperandType::TENSOR_QUANT8_ASYMM) {
minSupportedVersion = Version::ANDROID_P;
} else if (inputType == OperandType::TENSOR_FLOAT16) {
minSupportedVersion = Version::ANDROID_Q;
} else if (inputType == OperandType::TENSOR_QUANT8_ASYMM_SIGNED) {
minSupportedVersion = Version::ANDROID_R;
} else {
NN_RET_CHECK_FAIL() << "Unsupported tensor type for operation " << kOperationName;
}
const Shape& input = context->getInputShape(kInputTensor);
if (hasKnownRank(input)) {
NN_RET_CHECK_LE(getNumberOfDimensions(input), 4);
}
NN_RET_CHECK(validateInputTypes(context, {inputType, OperandType::TENSOR_INT32}));
NN_RET_CHECK(validateOutputTypes(context, {inputType}));
return minSupportedVersion;
}
#ifdef NN_INCLUDE_CPU_IMPLEMENTATION
bool prepare(IOperationExecutionContext* context) {
// Only the permutation tensor can be omitted.
NN_RET_CHECK(!context->isOmittedInput(kInputTensor));
NN_RET_CHECK(!context->isOmittedOutput(kOutputTensor));
const Shape& input = context->getInputShape(kInputTensor);
uint32_t numInputDims = getNumberOfDimensions(input);
Shape output = context->getOutputShape(kOutputTensor);
output.type = input.type;
output.offset = input.offset;
output.scale = input.scale;
// permData can be NO_VALUE representing a regular 2D matrix transpose
if (context->isOmittedInput(kPermTensor)) {
NN_RET_CHECK_EQ(numInputDims, 2);
output.dimensions = {getSizeOfDimension(input, 1), getSizeOfDimension(input, 0)};
} else {
const Shape& permShape = context->getInputShape(kPermTensor);
const int32_t* permData = context->getInputBuffer<int32_t>(kPermTensor);
// Transpose op only supports 1D-4D input arrays.
NN_RET_CHECK_LE(numInputDims, 4);
// perm need to be provided as a 1-D int32 tensor.
NN_RET_CHECK(permShape.type == OperandType::TENSOR_INT32);
NN_RET_CHECK_EQ(getNumberOfDimensions(permShape), 1);
NN_RET_CHECK_EQ(numInputDims, getSizeOfDimension(permShape, 0));
std::vector<uint32_t> outDims(numInputDims);
for (int32_t idx = 0; idx < static_cast<int32_t>(numInputDims); ++idx) {
NN_RET_CHECK(permData[idx] >= 0 && permData[idx] < static_cast<int32_t>(numInputDims));
outDims[idx] = getSizeOfDimension(input, permData[idx]);
}
output.dimensions = outDims;
}
return context->setOutputShape(kOutputTensor, output);
}
bool execute(IOperationExecutionContext* context) {
// Bypass execution in the case of zero-sized input.
if (getNumberOfElements(context->getOutputShape(kOutputTensor)) == 0) return true;
switch (context->getInputType(kInputTensor)) {
case OperandType::TENSOR_FLOAT32:
return transposeGeneric(context->getInputBuffer<float>(kInputTensor),
context->getInputShape(kInputTensor),
context->getInputBuffer<int32_t>(kPermTensor),
context->getInputShape(kPermTensor),
context->getOutputBuffer<float>(kOutputTensor),
context->getOutputShape(kOutputTensor));
case OperandType::TENSOR_FLOAT16:
return transposeGeneric(context->getInputBuffer<_Float16>(kInputTensor),
context->getInputShape(kInputTensor),
context->getInputBuffer<int32_t>(kPermTensor),
context->getInputShape(kPermTensor),
context->getOutputBuffer<_Float16>(kOutputTensor),
context->getOutputShape(kOutputTensor));
case OperandType::TENSOR_QUANT8_ASYMM:
return transposeGeneric(context->getInputBuffer<uint8_t>(kInputTensor),
context->getInputShape(kInputTensor),
context->getInputBuffer<int32_t>(kPermTensor),
context->getInputShape(kPermTensor),
context->getOutputBuffer<uint8_t>(kOutputTensor),
context->getOutputShape(kOutputTensor));
case OperandType::TENSOR_QUANT8_ASYMM_SIGNED:
return transposeGeneric(context->getInputBuffer<int8_t>(kInputTensor),
context->getInputShape(kInputTensor),
context->getInputBuffer<int32_t>(kPermTensor),
context->getInputShape(kPermTensor),
context->getOutputBuffer<int8_t>(kOutputTensor),
context->getOutputShape(kOutputTensor));
default:
NN_RET_CHECK_FAIL() << "Unsupported tensor type for operation " << kOperationName;
}
}
#endif // NN_INCLUDE_CPU_IMPLEMENTATION
} // namespace transpose
NN_REGISTER_OPERATION(TRANSPOSE, transpose::kOperationName, transpose::validate, transpose::prepare,
transpose::execute, .allowOmittedOperand = true, .allowZeroSizedInput = true);
} // namespace nn
} // namespace android