232 lines
10 KiB
Plaintext
232 lines
10 KiB
Plaintext
/*******************************************************************************
|
|
* Copyright (c) 2015-2018 Skymind, Inc.
|
|
*
|
|
* This program and the accompanying materials are made available under the
|
|
* terms of the Apache License, Version 2.0 which is available at
|
|
* https://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.
|
|
*
|
|
* SPDX-License-Identifier: Apache-2.0
|
|
******************************************************************************/
|
|
|
|
//
|
|
// @author Yurii Shyrma, created on 16.04.2018
|
|
//
|
|
|
|
#include <ops/declarable/helpers/reverse.h>
|
|
#include <helpers/ShapeUtils.h>
|
|
#include <array/ResultSet.h>
|
|
#include <TAD.h>
|
|
#include <PointersManager.h>
|
|
#include <ConstantTadHelper.h>
|
|
|
|
|
|
namespace nd4j {
|
|
namespace ops {
|
|
namespace helpers {
|
|
|
|
template <typename T>
|
|
inline void __device__ indexSwap(T* arr, Nd4jLong idx1, Nd4jLong idx2) {
|
|
T tmp = arr[idx1];
|
|
arr[idx1] = arr[idx2];
|
|
arr[idx2] = tmp;
|
|
}
|
|
// template <typename T>
|
|
// void reverseArray(nd4j::LaunchContext * context, void* inArr, Nd4jLong *inShapeBuffer, void *result, Nd4jLong *zShapeBuffer, int numOfElemsToReverse = 0);
|
|
|
|
/////////////////////////////////////////////////////////////////////////////////////
|
|
template <typename T>
|
|
static __global__ void reverseArrayInplaceKernel(void *input, Nd4jLong *inputShape, Nd4jLong numOfElemsToReverse) {
|
|
const auto tid = blockIdx.x * gridDim.x + threadIdx.x;
|
|
const auto step = gridDim.x * blockDim.x;
|
|
__shared__ Nd4jLong length;
|
|
__shared__ int linearStatus;
|
|
__shared__ T* inputArr;
|
|
if (threadIdx.x == 0) {
|
|
length = shape::length(inputShape);
|
|
linearStatus = shape::elementWiseStride(inputShape);
|
|
inputArr = reinterpret_cast<T*>(input);
|
|
}
|
|
__syncthreads();
|
|
|
|
for (Nd4jLong e = tid; e < numOfElemsToReverse / 2; e += step) {
|
|
if (linearStatus == 1) {
|
|
auto idx = numOfElemsToReverse - e - 1;
|
|
indexSwap(inputArr, e, idx);
|
|
}
|
|
else if (linearStatus > 1) {
|
|
auto idx1 = (numOfElemsToReverse - e - 1) * linearStatus;
|
|
Nd4jLong idx2 = e * linearStatus;
|
|
indexSwap(inputArr, idx1, idx2);
|
|
}
|
|
else {
|
|
auto inOffset = shape::getIndexOffset(e, inputShape, length);
|
|
auto outOffset = shape::getIndexOffset(numOfElemsToReverse - e - 1, inputShape, length);
|
|
indexSwap(inputArr, inOffset, outOffset);
|
|
}
|
|
}
|
|
}
|
|
|
|
template <typename T>
|
|
static __global__ void reverseArrayKernel(void* input, Nd4jLong *inputShape, void* output, Nd4jLong *outputShape, Nd4jLong numOfElemsToReverse) {
|
|
const auto tid = blockIdx.x * gridDim.x + threadIdx.x;
|
|
const auto step = gridDim.x * blockDim.x;
|
|
__shared__ Nd4jLong length;
|
|
__shared__ int linearStatus;
|
|
__shared__ T* inputArr;
|
|
__shared__ T* outputArr;
|
|
__shared__ char inputOrder, outputOrder;
|
|
|
|
if (threadIdx.x == 0) {
|
|
length = shape::length(inputShape);
|
|
linearStatus = (shape::elementWiseStride(inputShape) == shape::elementWiseStride(outputShape)) && (inputOrder == outputOrder)? shape::elementWiseStride(inputShape):0;
|
|
|
|
char inputOrder = shape::order(inputShape);
|
|
char outputOrder = shape::order(outputShape);
|
|
inputArr = reinterpret_cast<T*>(input);
|
|
outputArr = reinterpret_cast<T*>(output);
|
|
}
|
|
__syncthreads();
|
|
|
|
for (Nd4jLong e = tid; e < length; e += step) {
|
|
if (e < numOfElemsToReverse ) {
|
|
if (linearStatus == 1) {
|
|
auto idx = numOfElemsToReverse - e - 1;
|
|
outputArr[idx] = inputArr[e];
|
|
} else if (linearStatus > 1) {
|
|
auto idx1 = (numOfElemsToReverse - e - 1) * linearStatus;
|
|
Nd4jLong idx2 = e * linearStatus;
|
|
outputArr[idx1] = inputArr[idx2];
|
|
} else {
|
|
auto inOffset = shape::getIndexOffset(e, inputShape, length);
|
|
auto outOffset = shape::getIndexOffset(numOfElemsToReverse - e - 1, outputShape, length);
|
|
outputArr[outOffset] = inputArr[inOffset];
|
|
}
|
|
}
|
|
else {
|
|
if (linearStatus == 1) {
|
|
outputArr[e] = inputArr[e];
|
|
} else if (linearStatus > 1) {
|
|
auto idx1 = e * linearStatus;
|
|
Nd4jLong idx2 = e * linearStatus;
|
|
outputArr[idx1] = inputArr[idx2];
|
|
} else {
|
|
auto inOffset = shape::getIndexOffset(e, inputShape, length);
|
|
auto outOffset = shape::getIndexOffset(e, outputShape, length);
|
|
outputArr[outOffset] = inputArr[inOffset];
|
|
}
|
|
}
|
|
}
|
|
|
|
//printf("\n");
|
|
}
|
|
|
|
template<typename T>
|
|
static void reverseArray(nd4j::LaunchContext * context, NDArray* input, NDArray* output, int numOfElemsToReverse) {
|
|
auto stream = context->getCudaStream();
|
|
Nd4jLong numOfReverse = numOfElemsToReverse;
|
|
if (numOfElemsToReverse == 0)
|
|
numOfReverse = input->lengthOf();
|
|
if (input == output) {
|
|
reverseArrayInplaceKernel<T><<<256, 512, 8192, *stream>>>(input->specialBuffer(), input->specialShapeInfo(), numOfReverse);
|
|
}
|
|
else {
|
|
reverseArrayKernel<T><<<256, 512, 8192, *stream>>>(input->specialBuffer(), input->specialShapeInfo(), output->specialBuffer(), output->specialShapeInfo(), numOfReverse);
|
|
}
|
|
}
|
|
|
|
|
|
///////////////////////////////////////////////////////////////////
|
|
template <typename T>
|
|
static void _reverseSequence(nd4j::LaunchContext * context, const NDArray* input, const NDArray* seqLengths, NDArray* output, int seqDim, const int batchDim){
|
|
int posOfNonUnityDim = -1;
|
|
seqLengths->syncToHost();
|
|
auto stream = context->getCudaStream();
|
|
if (!input->isActualOnDeviceSide())
|
|
input->syncToDevice();
|
|
|
|
if(input->isVector() || shape::isLikeVector(input->getShapeInfo(), posOfNonUnityDim) || seqLengths->lengthOf() == 1) {
|
|
int numOfElemsToReverse = seqLengths->e<int>(0);
|
|
// printf("Length %d\n", numOfElemsToReverse);
|
|
// input->printBuffer("INPUT");
|
|
if((seqDim == 0 && input->sizeAt(0) == 1) || (batchDim == posOfNonUnityDim))
|
|
output->assign(input);
|
|
else
|
|
reverseArrayKernel<T><<<256, 512, 8192, *stream>>>(input->getSpecialBuffer(), input->getSpecialShapeInfo(), output->specialBuffer(), output->specialShapeInfo(), numOfElemsToReverse);//helpers::reverseArray<T>(context, const_cast<NDArray*>(input), output, numOfElemsToReverse);
|
|
}
|
|
else {
|
|
|
|
if(seqDim > batchDim)
|
|
--seqDim;
|
|
|
|
std::vector<int> dimensions = ShapeUtils::evalDimsToExclude(input->rankOf(), {batchDim});
|
|
|
|
auto inSubArrsSet = input->allTensorsAlongDimension(dimensions);
|
|
auto outSubArrsSet = output->allTensorsAlongDimension(dimensions);
|
|
|
|
// #pragma omp parallel for schedule(guided) if(inSubArrsSet->size() > Environment::getInstance()->elementwiseThreshold())
|
|
for(int i = 0; i < inSubArrsSet->size(); ++i) {
|
|
|
|
int numOfElemsToReverse = seqLengths->e<int>(i);
|
|
|
|
if(numOfElemsToReverse == 0 || numOfElemsToReverse == 1) {
|
|
outSubArrsSet->at(i)->assign(inSubArrsSet->at(i));
|
|
}
|
|
else {
|
|
auto inInnerSet = inSubArrsSet->at(i)->allTensorsAlongDimension({seqDim});
|
|
auto outInnerSet = outSubArrsSet->at(i)->allTensorsAlongDimension({seqDim});
|
|
for(int j = 0; j < inInnerSet->size(); ++j)
|
|
reverseArray<T>(context, inInnerSet->at(j), outInnerSet->at(j), numOfElemsToReverse);
|
|
|
|
delete inInnerSet;
|
|
delete outInnerSet;
|
|
}
|
|
}
|
|
delete inSubArrsSet;
|
|
delete outSubArrsSet;
|
|
}
|
|
input->tickReadDevice();
|
|
output->tickWriteDevice();
|
|
}
|
|
|
|
void reverseSequence(nd4j::LaunchContext * context, const NDArray* input, const NDArray* seqLengths, NDArray* output, int seqDim, const int batchDim) {
|
|
BUILD_SINGLE_SELECTOR(input->dataType(), _reverseSequence, (context, input, seqLengths, output, seqDim, batchDim), LIBND4J_TYPES);
|
|
}
|
|
|
|
//////////////////////////////////////////////////////////////////////////
|
|
void reverse(nd4j::LaunchContext * context, const NDArray* input, NDArray* output, const std::vector<int>* intArgs, bool isBackProp) {
|
|
// we need to reverse axis only if that's new op
|
|
std::vector<int> dimensions = isBackProp ? ShapeUtils::evalDimsToExclude(input->rankOf(), *intArgs) : *intArgs;
|
|
std::vector<int> axis = ShapeUtils::evalDimsToExclude(input->rankOf(), dimensions);
|
|
auto packX = nd4j::ConstantTadHelper::getInstance()->tadForDimensions(input->getShapeInfo(), axis);
|
|
auto packZ = nd4j::ConstantTadHelper::getInstance()->tadForDimensions(output->getShapeInfo(), axis);
|
|
|
|
auto listOut = output->allTensorsAlongDimension(dimensions);
|
|
auto listIn = input->allTensorsAlongDimension(dimensions);
|
|
|
|
NDArray *subArrIn, *subArrOut;
|
|
|
|
for(int i = 0; i < listIn->size(); ++i) { // listIn->size() = listOut->size()
|
|
subArrIn = listIn->at(i);
|
|
subArrOut = listOut->at(i);
|
|
BUILD_SINGLE_SELECTOR(input->dataType(), reverseArray, (context, subArrIn, subArrOut, 0), LIBND4J_TYPES);
|
|
}
|
|
//BUILD_SINGLE_SELECTOR(input->dataType(), reverseArray, (context, const_cast<NDArray*>(input), output, (int)0), LIBND4J_TYPES);
|
|
input->tickReadDevice();
|
|
output->tickWriteDevice();
|
|
delete listOut;
|
|
delete listIn;
|
|
}
|
|
|
|
BUILD_SINGLE_TEMPLATE(template void reverseArray, (nd4j::LaunchContext * context, NDArray *inArr, NDArray *outArr, int numOfElemsToReverse), LIBND4J_TYPES);
|
|
|
|
}
|
|
}
|
|
}
|
|
|