cavis/libnd4j/include/legacy/cuda/NativeOps.cu

/*******************************************************************************
 * 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
 ******************************************************************************/


#include <legacy/NativeOpExecutioner.h>
#include <legacy/NativeOps.h>
#include <cuda.h>

#include <system/buffer.h>


#include <loops/transform_any.h>
#include <loops/reduce_bool.h>
#include <loops/reduce_long.h>
#include <loops/scalar.h>
#include <helpers/threshold.h>
#include <ops/specials_cuda.h>
#include <helpers/DebugHelper.h>
#include <execution/AffinityManager.h>

#include <exceptions/datatype_exception.h>
#include <exceptions/cuda_exception.h>
#include <helpers/CudaLaunchHelper.h>
#include <graph/GraphExecutioner.h>
#include <helpers/BlasHelper.h>
#include <graph/GraphHolder.h>
#include <ops/declarable/CustomOperations.h>
#include <helpers/PointersManager.h>


//#include <sys/time.h>

#include <curand.h>
#include <graph/Status.h>
#include <helpers/DebugHelper.h>

using namespace sd;

#include <loops/special_kernels.h>
#include <performance/benchmarking/FullBenchmarkSuit.h>
#include <performance/benchmarking/LightBenchmarkSuit.h>

cudaDeviceProp *deviceProperties;
cudaFuncAttributes *funcAttributes = new cudaFuncAttributes[64];
int blockLimit = 128;
int maxThreads = 512;
bool allowedP2P = false;
bool supportedP2P = false;
#ifdef __ND4J_EXPERIMENTAL__
bool experimentalSupport = true;
#else
bool experimentalSupport = false;
#endif

int minThreads = 32;

__constant__ char deviceConstantMemory[49152];


// this method just does type conversion in fancy way
int getDeviceId(Nd4jPointer ptrToDeviceId) {
    return (int)(Nd4jLong)ptrToDeviceId;
}

/*
 * Basic CUDA constants here: number of blocks per MP
 */
int getDeviceBlockThreshold(int deviceId) {
	int ccMinor = deviceProperties[deviceId].minor;
	int ccMajor = deviceProperties[deviceId].major;

	int blockThreshold = 8;

	if (ccMajor >= 5)
		blockThreshold = 32;
	else if (ccMajor == 3)
		blockThreshold = 16;
	else if (ccMajor < 3)
		blockThreshold = 8;

	return blockThreshold;
}


/*
 * This message returns shared memory threshold value. default overflow ratio is 0.3
 */
int getDeviceSharedThreshold(int deviceId) {
	int ccMinor = deviceProperties[deviceId].minor;
	int ccMajor = deviceProperties[deviceId].major;

	// please note threshold isn't multiple of 32, and that's NOT a mistake

	int shmemThreshold;
	if (ccMajor == 6 && ccMinor == 0)
		shmemThreshold = 65536;
	else if (ccMajor == 6 && ccMinor == 1)
		shmemThreshold = 49152;
	else if (ccMajor == 5 && ccMinor == 2)
		shmemThreshold = 98304;
	else if (ccMajor == 5)
		shmemThreshold = 65536;
	else if (ccMajor == 3 && ccMinor == 7)
		shmemThreshold = 114688;
	else shmemThreshold = 49152;

	return shmemThreshold / 0.3;
}



sd::buffer::Buffer<Nd4jLong> * createScalarBuffer(cudaStream_t stream) {
	Nd4jLong *scalarShapeInfo = shape::createScalarShapeInfo();
	sd::buffer::Buffer<Nd4jLong> *buff = sd::buffer::createBuffer(scalarShapeInfo,shape::shapeInfoLength(2), stream);
	sd::buffer::copyDataToGpu(&buff, stream);
	return buff;
}


class ScalarShapeInformation {
private:
	sd::buffer::Buffer<Nd4jLong> *scalarDimension;
	sd::buffer::Buffer<Nd4jLong> *scalarShapeInfo;
//	std::thread::id threadId;

public:
	ScalarShapeInformation(cudaStream_t stream) {
		auto scalarDimensionBuff = reinterpret_cast<Nd4jLong *>(malloc(sizeof(Nd4jLong)));

		CHECK_ALLOC(scalarDimensionBuff, "Failed to allocate ShapeInfoBuffer", sizeof(Nd4jLong));

		scalarDimensionBuff[0] = MAX_DIMENSION;
		scalarDimension = sd::buffer::createBuffer(scalarDimensionBuff,1, stream);
		scalarShapeInfo = createScalarBuffer(stream);
//		threadId = std::this_thread::get_id();

	}
	~ScalarShapeInformation() {
		sd::buffer::freeBuffer(&scalarShapeInfo);
		sd::buffer::freeBuffer(&scalarDimension);
	}


	Nd4jLong *getShapeInfoHostPointer() {
		return scalarShapeInfo->data;
	}

	Nd4jLong * getShapeInfoGpuPointer() {
		return scalarShapeInfo->gData;
	}

	Nd4jLong * getDimensionHostPointer() {
		return scalarDimension->data;
	}

	Nd4jLong  * getDimensionGpuPointer() {
		return scalarDimension->gData;
	}

};





template <typename T>
class ScalarInfo {
	sd::buffer::Buffer<T> *scalarData;
	ScalarShapeInformation *shapeInfo;
	T finalResult;
	cudaStream_t streamRef;
public:
	ScalarInfo(cudaStream_t stream) {
		T *scalarResult = reinterpret_cast<T*>(malloc(sizeof(T)));

		CHECK_ALLOC(scalarResult, "Failed to allocate new scalar buffer", sizeof(T));

		shapeInfo = new ScalarShapeInformation(stream);
		scalarData = sd::buffer::createBuffer(scalarResult,1, stream);
		streamRef = stream;
		sd::buffer::copyDataToGpu(&scalarData, stream);
	}

	T getFinalResultFromDevice() {
		sd::buffer::copyDataFromGpu(&scalarData, streamRef);
		return scalarData->data[0];
	}

	/**
	 * Get the device shape information
	 * representing a scalar
	 */
	 Nd4jLong *getDeviceShapeInfo() {
		return shapeInfo->getShapeInfoGpuPointer();
	}

	/**
	 * Get the dZ pointers
	 */
	 T *getDevicePointer() {
		 return scalarData->gData;
	 }

	 /**
	  * Get the infinite dimension device pointer
	  */
	  Nd4jLong *getDimensionDevicePointer() {
		 return shapeInfo->getDimensionGpuPointer();
	 }

	 ~ScalarInfo() {
		 sd::buffer::freeBuffer(&scalarData);
		 delete shapeInfo;
	 }
};

void execPairwiseTransform( Nd4jPointer *extraPointers,
        								int opNum,
                                        OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
                                        OpaqueDataBuffer *dbY, Nd4jLong *hYShapeInfo, Nd4jLong *dYShapeInfo,
                                        OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo,
        								void *extraParams) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX, dbY});

        LaunchContext lc(extraPointers[1], extraPointers[4], extraPointers[5], extraPointers[3]);
        NativeOpExecutioner::execPairwiseTransform(&lc, opNum, dbX->primary(), hXShapeInfo, dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(),
                                                        dbY->primary(), hYShapeInfo, dbY->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hYShapeInfo).specialAsT<Nd4jLong>(),
                                                        dbZ->primary(), hZShapeInfo, dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>(), extraParams);

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX, dbY});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

////////////////////////////////////////////////////////////////////////
void execPairwiseTransformBool(Nd4jPointer *extraPointers,
        								int opNum,
                                        OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
                                        OpaqueDataBuffer *dbY, Nd4jLong *hYShapeInfo, Nd4jLong *dYShapeInfo,
        								OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo,
        								void *extraParams) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX, dbY});

        LaunchContext lc(extraPointers[1], extraPointers[4], extraPointers[5], extraPointers[3]);
        NativeOpExecutioner::execPairwiseBoolTransform(&lc, opNum,
                                                       dbX->primary(), hXShapeInfo, dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(),
                                                       dbY->primary(), hYShapeInfo, dbY->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hYShapeInfo).specialAsT<Nd4jLong>(),
                                                       dbZ->primary(), hZShapeInfo, dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>(),
                                                       extraParams);

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX, dbY});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

////////////////////////////////////////////////////////////////////////
void execSummaryStatsScalar(Nd4jPointer *extraPointers,
                                       int opNum,
                                       OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
                                       void *extraParams,
                                       OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo,
                                       bool biasCorrected) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX});

        LaunchContext lc(extraPointers[1], extraPointers[4], extraPointers[5], extraPointers[3]);
        NativeOpExecutioner::execSummaryStatsScalar(&lc, opNum,
                                                         dbX->primary(), hXShapeInfo, dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(),
                                                         extraParams,
                                                         dbZ->primary(), hZShapeInfo, dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>(),
                                                         biasCorrected);

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

////////////////////////////////////////////////////////////////////////
void execBroadcastBool(Nd4jPointer *extraPointers,
        						int opNum,
                                OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
                                OpaqueDataBuffer *dbY, Nd4jLong *hYShapeInfo, Nd4jLong *dYShapeInfo,
                                OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo,
                                void *extraParams,
                                OpaqueDataBuffer *dbDimension, Nd4jLong *hDimensionShape, Nd4jLong *dDimensionShape) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX, dbY});
        InteropDataBuffer::preparePrimaryUse({}, {dbDimension});

        auto dimension = reinterpret_cast<int *>(dbDimension->primary());
        int dimensionLength = static_cast<int>(shape::length(hDimensionShape));

        auto hTADShapeInfo = reinterpret_cast<Nd4jLong *>(extraPointers[9]);
        auto tadOnlyShapeInfo = reinterpret_cast<Nd4jLong *>(extraPointers[10]);
        auto tadOffsets = reinterpret_cast<Nd4jLong *>(extraPointers[11]);
        auto tadOnlyShapeInfoZ = reinterpret_cast<Nd4jLong *>(extraPointers[12]);
        auto tadOffsetsZ = reinterpret_cast<Nd4jLong *>(extraPointers[13]);

        LaunchContext lc(extraPointers[1], extraPointers[4], extraPointers[5], extraPointers[3]);
        NativeOpExecutioner::execBroadcastBool(&lc, opNum,
                dbX->primary(), hXShapeInfo, dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(),
                dbY->primary(), hYShapeInfo, dbY->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hYShapeInfo).specialAsT<Nd4jLong>(),
                dbZ->primary(), hZShapeInfo, dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>(),
                extraParams,
                dimension, dimensionLength,
                tadOnlyShapeInfo, tadOffsets, tadOnlyShapeInfoZ, tadOffsetsZ);

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX, dbY});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

/**
 *
 * @param opNum
 * @param dX
 * @param dXShapeInfo
 * @param dY
 * @param dYShapeInfo
 * @param dZ
 * @param dZShapeInfo
 * @param dimension
 * @param dimensionLength
 */
void   execBroadcast(
		Nd4jPointer *extraPointers,
		int opNum,
        OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
        OpaqueDataBuffer *dbY, Nd4jLong *hYShapeInfo, Nd4jLong *dYShapeInfo,
        OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo,
        OpaqueDataBuffer *dbDimension, Nd4jLong *hDimensionShape, Nd4jLong *dDimensionShape) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX, dbY});
        InteropDataBuffer::preparePrimaryUse({}, {dbDimension});

        auto dimension = reinterpret_cast<int *>(dbDimension->primary());
        int dimensionLength = static_cast<int>(shape::length(hDimensionShape));

        cudaStream_t *stream = reinterpret_cast<cudaStream_t *>(extraPointers[1]);

        auto hTADShapeInfo = reinterpret_cast<Nd4jLong *>(extraPointers[9]);
        auto tadOnlyShapeInfo = reinterpret_cast<Nd4jLong *>(extraPointers[10]);
        auto tadOffsets = reinterpret_cast<Nd4jLong *>(extraPointers[11]);
        auto tadOnlyShapeInfoZ = reinterpret_cast<Nd4jLong *>(extraPointers[12]);
        auto tadOffsetsZ = reinterpret_cast<Nd4jLong *>(extraPointers[13]);

        auto xType = sd::ArrayOptions::dataType(hXShapeInfo);
        auto yType = sd::ArrayOptions::dataType(hYShapeInfo);
        auto zType = sd::ArrayOptions::dataType(hZShapeInfo);

        LaunchContext lc(extraPointers[1], extraPointers[4], extraPointers[5], extraPointers[3]);
        NativeOpExecutioner::execBroadcast(&lc, opNum,
                dbX->primary(), hXShapeInfo, dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(),
                dbY->primary(), hYShapeInfo, dbY->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hYShapeInfo).specialAsT<Nd4jLong>(),
                dbZ->primary(), hZShapeInfo, dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>(),
                dimension, dimensionLength,
                tadOnlyShapeInfo, tadOffsets, tadOnlyShapeInfoZ, tadOffsetsZ);

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX, dbY});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}


/**
 *
 * @param opNum
 * @param dX
 * @param dXShapeInfo
 * @param extraParams
 * @param dZ
 * @param dZShapeInfo
 */
////////////////////////////////////////////////////////////////////////
void execReduceFloat(Nd4jPointer *extraPointers,
							int opNum,
                            OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
							void *extraParams,
                            OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX});

        LaunchContext lc(extraPointers[1], extraPointers[4], extraPointers[5], extraPointers[3]);
        NativeOpExecutioner::execReduceFloatScalar(&lc, opNum,
                dbX->primary(), hXShapeInfo, dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(),
                extraParams,
                dbZ->primary(), hZShapeInfo, dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>());

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

////////////////////////////////////////////////////////////////////////
void execReduceSame(Nd4jPointer *extraPointers,
                                int opNum,
                                OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
                                void *extraParams,
                                OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX});

        LaunchContext lc(extraPointers[1], extraPointers[4], extraPointers[5], extraPointers[3]);
        NativeOpExecutioner::execReduceSameScalar(&lc, opNum,
                dbX->primary(), hXShapeInfo, dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(),
                extraParams,
                dbZ->primary(), hZShapeInfo, dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>());

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

////////////////////////////////////////////////////////////////////////
void execReduceSame2(Nd4jPointer *extraPointers,
                            int opNum,
                            OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
                            void *extraParams,
                            OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo,
                            OpaqueDataBuffer *dbDimension, Nd4jLong *hDimensionShape, Nd4jLong *dDimensionShape) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX});
        InteropDataBuffer::preparePrimaryUse({}, {dbDimension});

        auto dimension = reinterpret_cast<int *>(dbDimension->primary());
        int dimensionLength = static_cast<int>(shape::length(hDimensionShape));

        auto tadPack = sd::ConstantTadHelper::getInstance()->tadForDimensions(hXShapeInfo,
                                                                                dimension,
                                                                                shape::length(hDimensionShape));

        LaunchContext lc(extraPointers[1], extraPointers[4], extraPointers[5], extraPointers[3]);
        NativeOpExecutioner::execReduceSame(&lc, opNum,
                dbX->primary(), hXShapeInfo, dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(),
                extraParams,
                dbZ->primary(), hZShapeInfo, dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>(),
                dimension, dimensionLength,
                tadPack.specialShapeInfo(), tadPack.specialOffsets());

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

////////////////////////////////////////////////////////////////////////
void execReduceLong2(Nd4jPointer *extraPointers,
                            int opNum,
                            OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
                            void *extraParams,
                            OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo,
                            OpaqueDataBuffer *dbDimension, Nd4jLong *hDimensionShape, Nd4jLong *dDimensionShape) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX});
        InteropDataBuffer::preparePrimaryUse({}, {dbDimension});

        auto dimension = reinterpret_cast<int *>(dbDimension->primary());
        int dimensionLength = static_cast<int>(shape::length(hDimensionShape));

        auto tadPack = sd::ConstantTadHelper::getInstance()->tadForDimensions(hXShapeInfo,
                                                                                dimension,
                                                                                shape::length(hDimensionShape));

        LaunchContext lc(extraPointers[1], extraPointers[4], extraPointers[5], extraPointers[3]);
        NativeOpExecutioner::execReduceLong(&lc, opNum,
                dbX->primary(), hXShapeInfo, dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(),
                extraParams,
                dbZ->primary(), hZShapeInfo, dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>(),
                dimension, dimensionLength,
                tadPack.specialShapeInfo(), tadPack.specialOffsets());

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

////////////////////////////////////////////////////////////////////////
void   execReduceLong(Nd4jPointer *extraPointers,
                                int opNum,
                                OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
                                void *extraParams,
                                OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX});

        auto stream = reinterpret_cast<cudaStream_t *>(extraPointers[1]);
        auto hTADShapeInfo = reinterpret_cast<Nd4jLong *>(extraPointers[9]);
        auto dTADShapeInfo = reinterpret_cast<Nd4jLong *>(extraPointers[10]);

        auto reductionPointer = reinterpret_cast<void *>(extraPointers[4]);

        auto xType = sd::ArrayOptions::dataType(hXShapeInfo);
        auto zType = sd::ArrayOptions::dataType(hZShapeInfo);

        if (zType != sd::DataType::INT64)
            throw datatype_exception::build("execReduceLong wrong Z data type", sd::DataType::INT64, zType);

        auto xLength = shape::length(hXShapeInfo);
        auto blockWidth = 256;
        auto numBlocks = CudaLaunchHelper::getReductionBlocks(xLength, blockWidth);
        dim3 launchDims(numBlocks, blockWidth, 32768);

        BUILD_DOUBLE_SELECTOR(xType, zType, functions::reduce::ReduceLongFunction,
                              ::execReduceScalar(launchDims, stream, opNum,
                                      dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(), hXShapeInfo,
                                      extraParams,
                                      dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>(), hXShapeInfo,
                                      nullptr, 0, reductionPointer, dTADShapeInfo), LIBND4J_TYPES, LONG_TYPES);

        sd::DebugHelper::checkErrorCode(stream, "execReduceLong(...) failed");

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

////////////////////////////////////////////////////////////////////////
void execReduceBool2(Nd4jPointer *extraPointers,
                            int opNum,
                            OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
                            void *extraParams,
                            OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo,
                            OpaqueDataBuffer *dbDimension, Nd4jLong *hDimensionShape, Nd4jLong *dDimensionShape) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX});
        InteropDataBuffer::preparePrimaryUse({}, {dbDimension});

        auto dimension = reinterpret_cast<int *>(dbDimension->primary());
        int dimensionLength = static_cast<int>(shape::length(hDimensionShape));

        auto tadPack = sd::ConstantTadHelper::getInstance()->tadForDimensions(hXShapeInfo,
                                                                                dimension,
                                                                                shape::length(hDimensionShape));

        LaunchContext lc(extraPointers[1], extraPointers[4], extraPointers[5], extraPointers[3]);
        NativeOpExecutioner::execReduceBool(&lc, opNum,
                dbX->primary(), hXShapeInfo, dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(),
                extraParams,
                dbZ->primary(), hZShapeInfo, dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>(),
                dimension, dimensionLength,
                tadPack.specialShapeInfo(), tadPack.specialOffsets());

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

////////////////////////////////////////////////////////////////////////
void   execReduceBool(Nd4jPointer *extraPointers,
                                int opNum,
                                OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
                                void *extraParams,
                                OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX});

        auto stream = reinterpret_cast<cudaStream_t *>(extraPointers[1]);
        auto hTADShapeInfo = reinterpret_cast<Nd4jLong *>(extraPointers[9]);
        auto dTADShapeInfo = reinterpret_cast<Nd4jLong *>(extraPointers[10]);

        auto reductionPointer = reinterpret_cast<void *>(extraPointers[4]);

        auto xType = sd::ArrayOptions::dataType(hXShapeInfo);
        auto zType = sd::ArrayOptions::dataType(hZShapeInfo);

        if (zType != sd::DataType::BOOL)
            throw std::runtime_error("execReduceBool requires Z operand to have BOOL type");

        auto xLength = shape::length(hXShapeInfo);
        auto blockWidth = 256;
        auto numBlocks = CudaLaunchHelper::getReductionBlocks(xLength, blockWidth);
        dim3 launchDims(numBlocks, blockWidth, 32768);

        BUILD_DOUBLE_SELECTOR(xType, zType, functions::reduce::ReduceBoolFunction,
                              ::execReduceScalar(launchDims, stream, opNum,
                                      dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(), hXShapeInfo,
                                      extraParams,
                                      dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>(), hZShapeInfo,
                                      nullptr, 0, reductionPointer, dTADShapeInfo), LIBND4J_TYPES, BOOL_TYPES);

        sd::DebugHelper::checkErrorCode(stream, "execReduceBool(...) failed");

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

/**
 *
 * @param opNum
 * @param dX
 * @param dXShapeInfo
 * @param extraParams
 * @param dZ
 * @param dZShapeInfo
 * @param dimension
 * @param dimensionLength
 */
////////////////////////////////////////////////////////////////////////
void execIndexReduce(Nd4jPointer *extraPointers,
								 int opNum,
								 OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
        						 void *extraParams,
        						 OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo,
        						 OpaqueDataBuffer *dbDimension, Nd4jLong *hDimensionShape, Nd4jLong *dDimensionShape) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX});
        InteropDataBuffer::preparePrimaryUse({}, {dbDimension});

        auto dimension = reinterpret_cast<int *>(dbDimension->primary());
        int dimensionLength = static_cast<int>(shape::length(hDimensionShape));

        auto tadPack = sd::ConstantTadHelper::getInstance()->tadForDimensions(hXShapeInfo,
                                                                                dimension,
                                                                                shape::length(hDimensionShape));

        LaunchContext lc(extraPointers[1], extraPointers[4], extraPointers[5], extraPointers[3]);
        NativeOpExecutioner::execIndexReduce(&lc, opNum,
                dbX->primary(), hXShapeInfo, dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(),
                extraParams,
                dbZ->primary(), hZShapeInfo, dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>(),
                (int *) dbDimension->special(), dimensionLength,
                tadPack.specialShapeInfo(), tadPack.specialOffsets());

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

/**
 *
 * @param opNum
 * @param dX
 * @param dXShapeInfo
 * @param extraParams
 * @param dZ
 * @param dZShapeInfo
 */
////////////////////////////////////////////////////////////////////////
void execReduceFloat2(Nd4jPointer *extraPointers,
								int opNum,
                                OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
        						void *extraParams,
                                OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo,
                                OpaqueDataBuffer *dbDimension, Nd4jLong *hDimensionShape, Nd4jLong *dDimensionShape) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX});
        InteropDataBuffer::preparePrimaryUse({}, {dbDimension});

        auto dimension = reinterpret_cast<int *>(dbDimension->primary());
        int dimensionLength = static_cast<int>(shape::length(hDimensionShape));

        auto tadPack = sd::ConstantTadHelper::getInstance()->tadForDimensions(hXShapeInfo,
                                                                                dimension,
                                                                                shape::length(hDimensionShape));

        LaunchContext lc(extraPointers[1], extraPointers[4], extraPointers[5], extraPointers[3]);
        NativeOpExecutioner::execReduceFloat(&lc, opNum,
                dbX->primary(), hXShapeInfo, dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(),
                extraParams,
                dbZ->primary(), hZShapeInfo, dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>(),
                dimension, dimensionLength,
                tadPack.specialShapeInfo(), tadPack.specialOffsets());

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

/**
 *
 * @param opNum
 * @param dX
 * @param dXShapeInfo
 * @param extraParams
 */
////////////////////////////////////////////////////////////////////////
void execIndexReduceScalar(
		Nd4jPointer *extraPointers,
		int opNum,
        OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
        void *extraParams,
        OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo){
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX});

        LaunchContext lc(extraPointers[1], extraPointers[4], extraPointers[5], extraPointers[3]);
        NativeOpExecutioner::execIndexReduceScalar(&lc, opNum,
                dbX->primary(), hXShapeInfo, dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(),
                extraParams,
                dbZ->primary(), hZShapeInfo, dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>());

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

////////////////////////////////////////////////////////////////////////
void execTransformSame(Nd4jPointer *extraPointers,int opNum,
                                   OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
                                   OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo,
                                   void *extraParams) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX});

        auto tadShapeInfo = reinterpret_cast<Nd4jLong *>(extraPointers != nullptr ? extraPointers[0] : nullptr);
        auto tadOffsets = reinterpret_cast<Nd4jLong *>(extraPointers != nullptr ? extraPointers[1] : nullptr);

        LaunchContext lc(extraPointers[1], extraPointers[4], extraPointers[5], extraPointers[3]);
        NativeOpExecutioner::execTransformSame(&lc, opNum,
                dbX->primary(), hXShapeInfo, dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(),
                dbZ->primary(), hZShapeInfo, dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>(),
                extraParams,
                tadShapeInfo, tadOffsets);

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

////////////////////////////////////////////////////////////////////////
void execTransformBool(Nd4jPointer *extraPointers,int opNum,
                                  OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
                                  OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo,
								  void *extraParams) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX});

        auto tadShapeInfo = reinterpret_cast<Nd4jLong *>(extraPointers != nullptr ? extraPointers[0] : nullptr);
        auto tadOffsets = reinterpret_cast<Nd4jLong *>(extraPointers != nullptr ? extraPointers[1] : nullptr);

        LaunchContext lc(extraPointers[1], extraPointers[4], extraPointers[5], extraPointers[3]);
        NativeOpExecutioner::execTransformBool(&lc, opNum,
                dbX->primary(), hXShapeInfo, dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(),
                dbZ->primary(), hZShapeInfo, dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>(),
                extraParams,
                tadShapeInfo, tadOffsets);

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

////////////////////////////////////////////////////////////////////////
void execTransformAny(Nd4jPointer *extraPointers,int opNum,
                                    OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
                                    OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo,
								    void *extraParams) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX});

        auto stream = reinterpret_cast<cudaStream_t *>(extraPointers[1]);
        auto streamSpecial = reinterpret_cast<cudaStream_t &>(extraPointers[4]);
        LaunchContext lc(stream, streamSpecial, extraPointers[5], extraPointers[3],
                         reinterpret_cast<int *>(extraPointers[6]));

        NativeOpExecutioner::execTransformAny(&lc, opNum,
                dbX->primary(), hXShapeInfo, dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(),
                dbZ->primary(), hZShapeInfo, dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>(),
                extraParams,
                nullptr, nullptr);

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

////////////////////////////////////////////////////////////////////////
void execTransformStrict(Nd4jPointer *extraPointers,int opNum,
                                    OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
                                    OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo,
                                    void *extraParams) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX});

        auto tadShapeInfo = reinterpret_cast<Nd4jLong *>(extraPointers != nullptr ? extraPointers[10] : nullptr);
        auto tadOffsets = reinterpret_cast<Nd4jLong *>(extraPointers != nullptr ? extraPointers[11] : nullptr);

        LaunchContext lc(extraPointers[1], extraPointers[4], extraPointers[5], extraPointers[3]);
        NativeOpExecutioner::execTransformStrict(&lc, opNum,
                dbX->primary(), hXShapeInfo, dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(),
                dbZ->primary(), hZShapeInfo, dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>(),
                extraParams,
                tadShapeInfo, tadOffsets);

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

////////////////////////////////////////////////////////////////////////
void execTransformFloat(Nd4jPointer *extraPointers,int opNum,
                                    OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
                                    OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo,
                                    void *extraParams) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX});

        auto tadShapeInfo = reinterpret_cast<Nd4jLong *>(extraPointers != nullptr ? extraPointers[10] : nullptr);
        auto tadOffsets = reinterpret_cast<Nd4jLong *>(extraPointers != nullptr ? extraPointers[11] : nullptr);

        LaunchContext lc(extraPointers[1], extraPointers[4], extraPointers[5], extraPointers[3]);
        NativeOpExecutioner::execTransformFloat(&lc, opNum,
                dbX->primary(), hXShapeInfo, dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(),
                dbZ->primary(), hZShapeInfo, dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>(),
                extraParams,
                tadShapeInfo, tadOffsets);

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

void checkP2P() {
	int curDevice = 0;

	cudaGetDevice(&curDevice);

	int devCnt = 0;
	cudaGetDeviceCount(&devCnt);

	if (curDevice < 0 && curDevice > devCnt)
		curDevice = 0;

	bool tempSupport = true;

	if (devCnt > 1) {
		for (int dX = 0; dX < devCnt; dX++) {

			for (int dY = 0; dY < devCnt; dY++) {
				if (dX == dY)
					continue;

				int canAccess = 0;
				cudaSetDevice(dX);

				cudaDeviceCanAccessPeer(&canAccess, dX , dY);

				if (!canAccess) {
                    tempSupport = false;
                    break;
                }
			}
		}

		supportedP2P = tempSupport;

		cudaSetDevice(curDevice);
	} else {
		// if we have only 1 device - we say that we support P2P, since all data will be on 1 device
		supportedP2P = true;
	}
}

void enableP2P(bool enable) {
    if (enable == allowedP2P)
        return;

    int curDevice = 0;

    cudaGetDevice(&curDevice);

    int devCnt = 0;
    cudaGetDeviceCount(&devCnt);

	if (curDevice < 0 && curDevice > devCnt)
		curDevice = 0;

    if (devCnt > 1) {
        for (int dX = 0; dX < devCnt; dX++) {

            for (int dY = 0; dY < devCnt; dY++) {
                if (dX == dY)
                    continue;

                int canAccess = 0;
                cudaSetDevice(dX);

                cudaDeviceCanAccessPeer(&canAccess, dX , dY);

                if (canAccess) {
                    if (enable) {
                        cudaDeviceEnablePeerAccess(dY, 0);
                    } else {
                        cudaDeviceDisablePeerAccess(dY);
                    }
                } else {
					if (sd::Environment::getInstance()->isVerbose()) printf("Peer access [%i] -> [%i] isn't possible\n", dX, dY);
				}
            }
        }

        cudaSetDevice(curDevice);
    }

    allowedP2P = enable;

    cudaSetDevice(curDevice);
}

bool isP2PAvailable() {
	return supportedP2P;
}


void initializeDevicesAndFunctions() {
    try {
        int devCnt = 0;
        cudaGetDeviceCount(&devCnt);
        deviceProperties = new cudaDeviceProp[devCnt];
        for (int i = 0; i < devCnt; i++) {
            cudaSetDevice(i);
            cudaGetDeviceProperties(&deviceProperties[i], i);

            cudaDeviceSetLimit(cudaLimitStackSize, 4096);
        }

        cudaSetDevice(0);

        checkP2P();

        // enabling p2p gpu access if it's supported
        if (supportedP2P && devCnt > 1)
            enableP2P(allowedP2P);
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

void initializeFunctions(Nd4jPointer *functions) {
    sd::BlasHelper::getInstance()->initializeDeviceFunctions(functions);
	/*
	cublasSgemv = (CublasSgemv)functions[0];
    cublasDgemv = (CublasDgemv)functions[1];
    cublasHgemm = (CublasHgemm)functions[2];
    cublasSgemm = (CublasSgemm)functions[3];
    cublasDgemm = (CublasDgemm)functions[4];
    cublasSgemmEx = (CublasSgemmEx)functions[5];
    cublasHgemmBatched = (CublasHgemmBatched)functions[6];
    cublasSgemmBatched = (CublasSgemmBatched)functions[7];
    cublasDgemmBatched = (CublasDgemmBatched)functions[8];
	*/
}


/**
 * This method acquires memory chunk of requested size on host side
 *
 * @param pointer pointer that'll be used for allocation
 * @param memorySize memory size, in bytes
 * @param flags optional parameter
 */
Nd4jPointer mallocHost(Nd4jLong memorySize, int flags) {
	Nd4jPointer pointer;
	// cudaHostAllocMapped |cudaHostAllocPortable
	auto res = cudaHostAlloc(reinterpret_cast<void **>(&pointer), memorySize + 8, cudaHostAllocDefault);
	if (res != 0) {
	    sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(res);
	    sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage("cudaHostAlloc failed");
    }

	return reinterpret_cast<int8_t*>(pointer);
}

/**
 * This method acquires memory chunk of requested size on specified device
 *
 * @param pointer pointer that'll be used for allocation
 * @param memorySize memory size, in bytes
 * @param ptrToDeviceId pointer to deviceId. For cuda that's just and int, for OpenCL that's pointer to device_id, etc
 * @param flags optional parameter
 */
Nd4jPointer mallocDevice(Nd4jLong memorySize, int deviceId, int flags) {
	Nd4jPointer pointer;
	auto res = cudaMalloc(reinterpret_cast<void **>(&pointer), memorySize + 8);
	if (res != 0) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(res);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage("cudaMalloc failed");
	}

	return reinterpret_cast<int8_t*>(pointer);
}

/**
 * This method releases previously allocated host memory space
 *
 * @param pointer pointer that'll be freed
 */
int freeHost(Nd4jPointer pointer) {
	auto res = cudaFreeHost(reinterpret_cast<void *>(pointer));
	if (res != 0) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(res);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage("cudaFreeHost failed");
    }

	return 1L;
}

/**
 * This method releases previously allocated memory space on device
 *
 * @param pointer pointer that'll be freed
 * @param ptrToDeviceId pointer to deviceId.
 */
int freeDevice(Nd4jPointer pointer, int deviceId) {
	auto res = cudaFree(reinterpret_cast<void *>(pointer));

	// we're intentionally skipping
	if (res != 0 && res != 1) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(res);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage("cudaFree failed");
	}

	return res == 0 ? 1L : 0L;
}


Nd4jPointer createContext() {
	return 0L;
}

Nd4jPointer createStream() {

    auto stream = new cudaStream_t();
    auto dZ = cudaStreamCreate(stream);
    if (dZ != 0) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(dZ);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage("cudaStreamCreate failed");
    }

    return stream;
}

Nd4jPointer createEvent() {
	Nd4jPointer nativeEvent= (Nd4jPointer) malloc(sizeof(cudaEvent_t));

	CHECK_ALLOC(nativeEvent, "Failed to allocate new CUDA event buffer", sizeof(cudaEvent_t));

	auto dZ = cudaEventCreateWithFlags(reinterpret_cast<cudaEvent_t *>(&nativeEvent), cudaEventDisableTiming);
	if (dZ != 0) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(dZ);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage("cudaEventCreateWithFlags failed");
	}

	return nativeEvent;
}

int registerEvent(Nd4jPointer event, Nd4jPointer stream) {
	auto pEvent = reinterpret_cast<cudaEvent_t *>(&event);
	auto pStream = reinterpret_cast<cudaStream_t *>(stream);

    auto dZ = cudaEventRecord(*pEvent, *pStream);
	if (dZ != 0) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(dZ);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage("cudaEventRecord failed");
	}

	return 1;
}

int setDevice(int deviceId) {
	AffinityManager::setCurrentDevice(deviceId);
	return 1;
}

Nd4jLong getDeviceFreeMemoryDefault() {
    size_t memFree = 0;
    size_t memTotal = 0;

    cudaMemGetInfo(&memFree, &memTotal);

    return (Nd4jLong) memFree;
}

Nd4jLong getDeviceFreeMemory(int device) {
	int orig = -1;

	cudaGetDevice(&orig);

	if (device >= 0 && device != orig) {
		cudaSetDevice(device);
	}

	size_t memFree = 0;
	size_t memTotal = 0;

	cudaMemGetInfo(&memFree, &memTotal);

	if (device >= 0 && device != orig) {
		cudaSetDevice(orig);
	}

	return (Nd4jLong) memFree;
}

Nd4jLong getDeviceTotalMemory(int device) {
	int orig = -1;

	cudaGetDevice(&orig);

	if (device >= 0 && device != orig) {
		cudaSetDevice(device);
	}
	size_t memFree = 0;
	size_t memTotal = 0;

	cudaMemGetInfo(&memFree, &memTotal);

	if (device >= 0 && device != orig) {
		cudaSetDevice(orig);
	}

	return (Nd4jLong) memTotal;
}

int memcpySync(Nd4jPointer dst, Nd4jPointer src, Nd4jLong size, int flags, Nd4jPointer reserved) {
    cudaMemcpyKind 	kind;

    switch (flags) {
        case 0: {
            kind = cudaMemcpyHostToHost;
        }
            break;
        case 1: {
            kind = cudaMemcpyHostToDevice;
        }
            break;
        case 2: {
            kind = cudaMemcpyDeviceToHost;
        }
            break;
        case 3: {
            kind = cudaMemcpyDeviceToDevice;
        }
            break;
        default: {
            sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
            sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage("UNDEFNED MEMCPY");
            return 0;
        }
    }

    auto dZ = cudaMemcpy(reinterpret_cast<void *>(dst), const_cast<const void *>(reinterpret_cast<void *>(src)), static_cast<size_t>(size), kind);
    if (dZ != 0) {
        printf("Failed on [%p] -> [%p], size: [%i], direction: [%i], dZ: [%i]\n", src, dst, size, flags, static_cast<int>(dZ));
        fflush(stdout);
        fflush(stderr);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(dZ);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage("cudaMemcpy failed");
        return 0;
    }

    return 1;
}

int memcpyAsync(Nd4jPointer dst, Nd4jPointer src, Nd4jLong size, int flags, Nd4jPointer reserved) {
	auto pStream = reinterpret_cast<cudaStream_t *>(reserved);

	cudaMemcpyKind 	kind;

	//sd::DebugHelper::checkErrorCode(pStream, "Preliminary sync failed");

	switch (flags) {
		case 0: {
				kind = cudaMemcpyHostToHost;
			}
			break;
		case 1: {
				kind = cudaMemcpyHostToDevice;
			}
			break;
		case 2: {
				kind = cudaMemcpyDeviceToHost;
			}
            break;
		case 3: {
			    kind = cudaMemcpyDeviceToDevice;
		    }
			break;
		default: {
            sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
            sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage("UNDEFNED MEMCPY");
            return 0;
		}
	}

	auto dZ = cudaMemcpyAsync(reinterpret_cast<void *>(dst), const_cast<const void *>(reinterpret_cast<void *>(src)), static_cast<size_t>(size), kind, *pStream);
    //auto dZ = cudaMemcpy(reinterpret_cast<void *>(dst), const_cast<const void *>(reinterpret_cast<void *>(src)), static_cast<size_t>(size), kind);
	if (dZ != 0) {
        printf("Failed on [%p] -> [%p], size: [%i], direction: [%i], dZ: [%i]\n", src, dst, size, flags, static_cast<int>(dZ));
        fflush(stdout);
        fflush(stderr);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(dZ);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage("cudaMemcpyAsync failed");
        return 0;
	}

	return 1;
}

int memsetSync(Nd4jPointer dst, int value, Nd4jLong size, int flags, Nd4jPointer reserved) {
	auto dZ = cudaMemset(reinterpret_cast<void *>(dst), value, static_cast<size_t>(size));
	if (dZ != 0) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(dZ);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage("cudaMemset failed");
	}

	return 1;
}

int memsetAsync(Nd4jPointer dst, int value, Nd4jLong size, int flags, Nd4jPointer reserved) {
	auto pStream = reinterpret_cast<cudaStream_t *>(reserved);

	auto dZ = cudaMemsetAsync(reinterpret_cast<void *>(dst), value, static_cast<size_t>(size), *pStream);
	if (dZ != 0) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(dZ);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage("cudaMemsetAsync failed");
	}

	return 1;
}

int destroyEvent(Nd4jPointer event) {
	auto pEvent = reinterpret_cast<cudaEvent_t *>(&event);
	auto dZ = cudaEventDestroy(*pEvent);
	if (dZ != 0) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(dZ);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage("cudaEventDestroy failed");
	}

	return 1;
}

int streamSynchronize(Nd4jPointer stream) {
	auto pStream = reinterpret_cast<cudaStream_t *>(stream);

	auto dZ = cudaStreamSynchronize(*pStream);
	if (dZ != 0) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(dZ);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage("cudaStreamSynchronize failed");
	}

	return 1L;
}

int eventSynchronize(Nd4jPointer event) {
	auto pEvent = reinterpret_cast<cudaEvent_t *>(&event);

	auto  dZ = cudaEventSynchronize(*pEvent);
	if (dZ != 0) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(dZ);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage("cudaEventSynchronize failed");
	}

	return 1L;
}

int getAvailableDevices() {
	int devCnt = 0;
	cudaGetDeviceCount(&devCnt);
	return devCnt;
}

void enableDebugMode(bool reallyEnable) {
	sd::Environment::getInstance()->setDebug(reallyEnable);
}

void setGridLimit(int gridSize) {
	if (gridSize > 8192)
		gridSize = 8192;
	if (gridSize < 1)
		gridSize = 1;
	blockLimit = gridSize;
}

int ompGetMaxThreads() {
	return maxThreads;
}

int ompGetNumThreads() {
	return maxThreads;
}

void setOmpNumThreads(int threads) {
	if (threads > 1024)
		threads = 1024;
	if (threads < 32)
		threads = 32;
	maxThreads = threads;
}

void enableVerboseMode(bool reallyEnable) {
	sd::Environment::getInstance()->setVerbose(reallyEnable);
}

int getDeviceMajor(int device) {
	return deviceProperties[device].major;
}

int getDeviceMinor(int device) {
	return deviceProperties[device].minor;
}


const char * getDeviceName(int device) {
    return deviceProperties[device].name;
}

void specialConcat(
        Nd4jPointer *extraPointers,
        int dimension,
        int numArrays,
        Nd4jPointer *data,
        Nd4jPointer *inputShapeInfo,
        void *dZ,
        Nd4jLong *dZShapeInfo, Nd4jPointer *tadPointers, Nd4jPointer *offsetPointers) {
    try {
        BUILD_SINGLE_SELECTOR(ArrayOptions::dataType(dZShapeInfo), sd::SpecialMethods,
                              ::concatCpuGeneric(dimension, numArrays, data, inputShapeInfo, dZ, dZShapeInfo),
                              LIBND4J_TYPES);
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}


/**
 * This method saves
 */
sd::TadPack* tadOnlyShapeInfo(Nd4jLong *dXShapeInfo, int *dimension, int dimensionLength) {
    try {
        auto pack = new TadPack();
        *pack = sd::ConstantTadHelper::getInstance()->tadForDimensions(dXShapeInfo, dimension, dimensionLength);
        return pack;
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
        return nullptr;
    }
}

Nd4jLong* getPrimaryShapeInfo(sd::TadPack* pack) {
    return pack->primaryShapeInfo();
}
Nd4jLong* getPrimaryOffsets(sd::TadPack* pack) {
    return pack->primaryOffsets();
}
Nd4jLong* getSpecialShapeInfo(sd::TadPack* pack) {
    return pack->specialShapeInfo();
}
Nd4jLong* getSpecialOffsets(sd::TadPack* pack) {
    return pack->specialOffsets();
}
Nd4jLong getNumberOfTads(sd::TadPack* pack) {
    return pack->numberOfTads();
}
int getShapeInfoLength(sd::TadPack* pack) {
    return pack->shapeInfoLength();
}

int memcpyConstantAsync(Nd4jLong dst, Nd4jPointer src, Nd4jLong size, int flags, Nd4jPointer reserved) {
	cudaStream_t *pStream = reinterpret_cast<cudaStream_t *>(reserved);

	cudaMemcpyKind 	kind;

	DEBUG_KERNEL(pStream, -1);

	switch (flags) {
		case 0: {
			kind = cudaMemcpyHostToHost;
		}
			break;
		case 1: {
			kind = cudaMemcpyHostToDevice;
		}
			break;
		case 2: {
			kind = cudaMemcpyDeviceToHost;
		}
		case 3: {
			kind = cudaMemcpyDeviceToDevice;
		}
			break;
	}
	auto dZ = cudaMemcpyToSymbolAsync(deviceConstantMemory, const_cast<const void *>(src), size, dst, kind, *pStream);
	if (dZ != 0) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(dZ);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage("cudaMemcpyToSymbolAsync failed");
	}

	return 1;
}

Nd4jPointer getConstantSpace() {
	Nd4jPointer dConstAddr;
	cudaError_t dZ = cudaGetSymbolAddress(reinterpret_cast<void **>(&dConstAddr), deviceConstantMemory);

	if (dZ != 0) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(dZ);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage("cudaGetSymbolAddress failed");
	}

	return dConstAddr;
}

void pullRows(Nd4jPointer *extraPointers,
                         OpaqueDataBuffer *dbX, Nd4jLong *xShapeInfo, Nd4jLong *dXShapeInfo,
                         OpaqueDataBuffer *dbZ, Nd4jLong *zShapeInfo, Nd4jLong *dZShapeInfo,
						 Nd4jLong n,
						 Nd4jLong *indexes,
						 Nd4jLong *tadShapeInfo,
						 Nd4jLong *tadOffsets,
						 Nd4jLong *zTadShapeInfo,
						 Nd4jLong *zTadOffsets) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX});

        cudaStream_t *stream = reinterpret_cast<cudaStream_t *>(extraPointers[1]);
        dim3 launchDims(64, 256, 1024);
        auto xType = sd::ArrayOptions::dataType(xShapeInfo);
        BUILD_SINGLE_SELECTOR(xType, pullRowsKernelGeneric,
                              (launchDims, stream, dbX->special(), dbZ->special(), n, indexes, tadShapeInfo, tadOffsets, zTadShapeInfo, zTadOffsets),
                              LIBND4J_TYPES);

        DEBUG_KERNEL(stream, -1);

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}


void average(Nd4jPointer *extras,
						Nd4jPointer *x, Nd4jLong *xShapeInfo,
						Nd4jPointer *dx, Nd4jLong *dXShapeInfo,
						void *z, Nd4jLong *zShapeInfo,
						void *dz, Nd4jLong *dzShapeInfo,
						int n,
						Nd4jLong length,
						bool propagate) {
    try {
        cudaStream_t *stream = reinterpret_cast<cudaStream_t *>(extras[1]);
        int mode = getDeviceId(extras[3]);

        auto dX = reinterpret_cast<void **>(dx);

        if (sd::Environment::getInstance()->isDebugAndVerbose())
            printf("averageFloat called\n");

        auto xType = sd::ArrayOptions::dataType(xShapeInfo);
        // launching on gpu
        if (mode == 0) {
            dim3 launchDims(256, 256, 4096);
            BUILD_SINGLE_SELECTOR(xType, averagingKernelGeneric, (launchDims, stream, dX, dz, n, length, propagate),
                                  LIBND4J_TYPES);
            sd::DebugHelper::checkErrorCode(stream, "AverageFloat(...) failed");
        } else {
            // launching on host memory
            BUILD_SINGLE_SELECTOR(xType, sd::SpecialMethods, ::averageGeneric(x, z, zShapeInfo, n, length, propagate),
                                  LIBND4J_TYPES);
        }
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

void accumulate(Nd4jPointer *extras,
						   Nd4jPointer *x, Nd4jLong *xShapeInfo,
						   Nd4jPointer *dx, Nd4jLong *dXShapeInfo,
						   void *z, Nd4jLong *zShapeInfo,
						   void *dz, Nd4jLong *dzShapeInfo,
						   int n,
						   Nd4jLong length) {
    try {
        auto stream = reinterpret_cast<cudaStream_t *>(extras[1]);
        int mode = getDeviceId(extras[3]);

        auto dX = reinterpret_cast<void **>(dx);

        if (sd::Environment::getInstance()->isDebugAndVerbose())
            printf("accumulateFloat called\n");
        auto xType = sd::ArrayOptions::dataType(xShapeInfo);

        // launching on gpu
        if (mode == 0) {
            dim3 launchDims(n, 256, 16384);
            BUILD_SINGLE_SELECTOR(xType, accumulateKernelGeneric, (launchDims, stream, dX, dz, n, length),
                                  LIBND4J_TYPES);
            sd::DebugHelper::checkErrorCode(stream, "AccumulateFloat(...) failed");
        } else {
            // launching on host memory
            BUILD_SINGLE_SELECTOR(xType, sd::SpecialMethods, ::accumulateGeneric(x, z, zShapeInfo, n, length),
                                  LIBND4J_TYPES);
        }
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}


void shuffle(Nd4jPointer *extras,
						Nd4jPointer *x, Nd4jPointer *xShapeInfo,
						Nd4jPointer *dx, Nd4jPointer *dXShapeInfo,
						Nd4jPointer *z, Nd4jPointer *zShapeInfo,
						Nd4jPointer *dz, Nd4jPointer *dZShapeInfo,
						int N,
						int *shuffleMap,
						Nd4jPointer *tadShapeInfo,
						Nd4jPointer *tadOffsets) {
    try {
        cudaStream_t *stream = reinterpret_cast<cudaStream_t *>(extras[1]);

        auto dX = reinterpret_cast<void **>(dx);
        auto dZ = reinterpret_cast<void **>(dz);
        auto xShape = reinterpret_cast<Nd4jLong **>(xShapeInfo);
        auto dxShape = reinterpret_cast<Nd4jLong **>(dXShapeInfo);
        auto tadOnlyShapeInfo = reinterpret_cast<Nd4jLong **>(tadShapeInfo);
        auto tadOffset = reinterpret_cast<Nd4jLong **>(tadOffsets);

        auto xType = sd::ArrayOptions::dataType(xShape[0]);
        dim3 launchDims(256, 512, 8192);
        BUILD_SINGLE_SELECTOR(xType, shuffleKernelGeneric,
                              (launchDims, stream, dX, dxShape, dZ, N, shuffleMap, tadOnlyShapeInfo, tadOffset),
                              LIBND4J_TYPES);

        sd::DebugHelper::checkErrorCode(stream, "shuffle(...) failed");
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

bool isExperimentalEnabled() {
    return sd::Environment::getInstance()->isExperimentalBuild();
}

void setOmpMinThreads(int threads) {
    minThreads = sd::math::nd4j_max<int>(32, threads);
    minThreads = sd::math::nd4j_min<int>(maxThreads, minThreads);
}

int getDevice() {
    return sd::AffinityManager::currentDeviceId();
}

void setElementThreshold(int num) {
    // this is no-op for CUDA
}

void setTADThreshold(int num) {
    // this is no-op for CUDA
}

////////////////////////////////////////////////////////////////////////
void execSummaryStats(Nd4jPointer *extraPointers,
                                 int opNum,
                                 OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
                                 void *extraParams,
                                 OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo,
                                 bool biasCorrected) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX});

        LaunchContext lc(extraPointers[1], extraPointers[4], extraPointers[5], extraPointers[3]);
        NativeOpExecutioner::execSummaryStats(&lc, opNum,
                dbX->primary(), hXShapeInfo, dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(),
                extraParams,
                dbZ->primary(), hZShapeInfo, dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>(),
                biasCorrected);

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

////////////////////////////////////////////////////////////////////////
void execSummaryStatsTad(Nd4jPointer *extraPointers,
                                 int opNum,
                                 OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
                                 void *extraParams,
                                 OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo,
                                 OpaqueDataBuffer *dbDimension, Nd4jLong *hDimensionShape, Nd4jLong *dDimensionShape,
                                 bool biasCorrected,
								 Nd4jLong *tadShapeInfo, Nd4jLong *tadOffsets) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX, dbDimension});
        InteropDataBuffer::preparePrimaryUse({}, {dbDimension});

        auto dimension = reinterpret_cast<int *>(dbDimension->primary());
        int dimensionLength = static_cast<int>(shape::length(hDimensionShape));

        LaunchContext lc(extraPointers[1], extraPointers[4], extraPointers[5], extraPointers[3]);
        NativeOpExecutioner::execSummaryStats(&lc, opNum,
                dbX->primary(), hXShapeInfo, dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(),
                extraParams,
                dbZ->primary(), hZShapeInfo, dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>(),
                reinterpret_cast<int *>(dbDimension->special()), dimensionLength,
                tadShapeInfo, tadOffsets,
                biasCorrected);

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX, dbDimension});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

////////////////////////////////////////////////////////////////////////
void execReduce3(Nd4jPointer *extraPointers,
                            int opNum,
                            OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
                            void *extraParams,
                            OpaqueDataBuffer *dbY, Nd4jLong *hYShapeInfo, Nd4jLong *dYShapeInfo,
                            OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX, dbY});

        LaunchContext lc(extraPointers[1], extraPointers[4], extraPointers[5], extraPointers[3]);
        NativeOpExecutioner::execReduce3(&lc, opNum,
                dbX->primary(), hXShapeInfo, dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(),
                extraParams,
                dbY->primary(), hYShapeInfo, dbY->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hYShapeInfo).specialAsT<Nd4jLong>(),
                dbZ->primary(), hZShapeInfo, dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>());

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX, dbY});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

////////////////////////////////////////////////////////////////////////
void execReduce3Tad(Nd4jPointer *extraPointers,
                            int opNum,
                            OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
                            void *extraParams,
                            OpaqueDataBuffer *dbY, Nd4jLong *hYShapeInfo, Nd4jLong *dYShapeInfo,
                            OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo,
                            OpaqueDataBuffer *dbDimension, Nd4jLong *hDimensionShape, Nd4jLong *dDimensionShape,
                            Nd4jLong *tadOnlyShapeInfo, Nd4jLong *tadOffsets,
                            Nd4jLong *yTadOnlyShapeInfo, Nd4jLong *yTadOffsets) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX, dbY});
        InteropDataBuffer::preparePrimaryUse({}, {dbDimension});

        auto dimension = reinterpret_cast<int *>(dbDimension->primary());
        int dimensionLength = static_cast<int>(shape::length(hDimensionShape));

        auto tadPack = sd::ConstantTadHelper::getInstance()->tadForDimensions(hXShapeInfo,
                                                                                dimension,
                                                                                shape::length(hDimensionShape));
        auto tadLength = shape::length(tadPack.primaryShapeInfo());
        auto yLength = shape::length(hYShapeInfo);
        auto xLength = shape::length(hXShapeInfo);

        LaunchContext lc(extraPointers[1], extraPointers[4], extraPointers[5], extraPointers[3]);

        if (tadLength == yLength || tadLength == xLength) {
            // nd4j_printf("== way\n","");
            NativeOpExecutioner::execReduce3(&lc, opNum,
                    dbX->primary(), hXShapeInfo, dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(),
                    extraParams,
                    dbY->primary(), hYShapeInfo, dbY->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hYShapeInfo).specialAsT<Nd4jLong>(),
                    dbZ->primary(), hZShapeInfo, dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>(),
                    dimension, dimensionLength,
                    tadOnlyShapeInfo, tadOffsets, yTadOnlyShapeInfo, yTadOffsets);
        } else
            NativeOpExecutioner::execReduce3TAD(&lc, opNum,
                    dbX->primary(), hXShapeInfo, dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(),
                    extraParams,
                    dbY->primary(), hYShapeInfo, dbY->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hYShapeInfo).specialAsT<Nd4jLong>(),
                    dbZ->primary(), hZShapeInfo, dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>(),
                    dimension, dimensionLength,
                    tadOnlyShapeInfo, yTadOffsets, yTadOnlyShapeInfo, yTadOffsets);

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX, dbY});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

////////////////////////////////////////////////////////////////////////
void execReduce3Scalar(Nd4jPointer *extraPointers,int opNum,
                                    OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
                                    void *extraParams,
                                    OpaqueDataBuffer *dbY, Nd4jLong *hYShapeInfo, Nd4jLong *dYShapeInfo,
                                    OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX, dbY});

        LaunchContext lc(extraPointers[1], extraPointers[4], extraPointers[5], extraPointers[3]);
        NativeOpExecutioner::execReduce3Scalar(&lc, opNum,
                dbX->primary(), hXShapeInfo, dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(),
                extraParams,
                dbY->primary(), hYShapeInfo, dbY->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hYShapeInfo).specialAsT<Nd4jLong>(),
                dbZ->primary(), hZShapeInfo, dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>());

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX, dbY});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

////////////////////////////////////////////////////////////////////////
void execScalarBool(Nd4jPointer *extraPointers,
							int opNum,
                            OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
                            OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo,
                            OpaqueDataBuffer *dbScalar, Nd4jLong *hScalarShapeInfo, Nd4jLong *dScalarShapeInfo,
							void *extraParams) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX, dbScalar});

        LaunchContext lc(extraPointers[1], extraPointers[4], extraPointers[5], extraPointers[3]);
        NativeOpExecutioner::execScalarBool(&lc, opNum,
                dbX->primary(), hXShapeInfo, dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(),
                dbZ->primary(), hZShapeInfo, dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>(),
                dbScalar->primary(), hScalarShapeInfo, dbScalar->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hScalarShapeInfo).specialAsT<Nd4jLong>(),
                extraParams);

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX, dbScalar});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

////////////////////////////////////////////////////////////////////////
void execScalarBoolTad(Nd4jPointer *extraPointers,
						   int opNum,
						   OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
						   OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo,
						   OpaqueDataBuffer *dbScalars, Nd4jLong *hScalarShapeInfo, Nd4jLong *dScalarShapeInfo,
						   void *extraParams,
						   OpaqueDataBuffer *dbDimension, Nd4jLong *hDimensionShape, Nd4jLong *dDimensionShape,
                           Nd4jLong *tadShapeInfo, Nd4jLong *tadOffsets,
                           Nd4jLong *tadShapeInfoZ, Nd4jLong *tadOffsetsZ) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX, dbScalars});
        InteropDataBuffer::preparePrimaryUse({}, {dbDimension});

        auto dimension = reinterpret_cast<int *>(dbDimension->primary());
        int dimensionLength = static_cast<int>(shape::length(hDimensionShape));

        LaunchContext lc(extraPointers[1], extraPointers[4], extraPointers[5], extraPointers[3]);
        NativeOpExecutioner::execScalarBool(&lc, opNum,
                dbX->primary(), hXShapeInfo, dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(),
                extraParams,
                dbZ->primary(), hZShapeInfo, dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>(),
                dbScalars->primary(), hScalarShapeInfo, dbScalars->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hScalarShapeInfo).specialAsT<Nd4jLong>(),
                dimension, dimensionLength,
                tadShapeInfo, tadOffsets, tadShapeInfoZ, tadOffsetsZ);

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX, dbScalars});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

////////////////////////////////////////////////////////////////////////
void execScalar(Nd4jPointer *extraPointers,
						int opNum,
                        OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
                        OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo,
                        OpaqueDataBuffer *dbScalar, Nd4jLong *hScalarShapeInfo, Nd4jLong *dScalarShapeInfo,
						void *extraParams) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX, dbScalar});

        LaunchContext lc(extraPointers[1], extraPointers[4], extraPointers[5], extraPointers[3]);
        NativeOpExecutioner::execScalar(&lc, opNum,
                dbX->primary(), hXShapeInfo, dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(),
                dbZ->primary(), hZShapeInfo, dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>(),
                dbScalar->primary(), hScalarShapeInfo, dbScalar->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hScalarShapeInfo).specialAsT<Nd4jLong>(),
                        extraParams);

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX, dbScalar});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

////////////////////////////////////////////////////////////////////////
void execScalarTad(Nd4jPointer *extraPointers,
					 int opNum,
					 OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
					 OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo,
					 OpaqueDataBuffer *dbScalars, Nd4jLong *hScalarShapeInfo, Nd4jLong *dScalarShapeInfo,
					 void *extraParams,
					 OpaqueDataBuffer *dbDimension, Nd4jLong *hDimensionShape, Nd4jLong *dDimensionShape,
                     Nd4jLong *tadShapeInfo, Nd4jLong *tadOffsets,
                     Nd4jLong *tadShapeInfoZ, Nd4jLong *tadOffsetsZ) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX, dbScalars});
        InteropDataBuffer::preparePrimaryUse({}, {dbDimension});

        auto dimension = reinterpret_cast<int *>(dbDimension->primary());
        int dimensionLength = static_cast<int>(shape::length(hDimensionShape));

        cudaStream_t *stream = reinterpret_cast<cudaStream_t *>(extraPointers[1]);

        auto xType = sd::ArrayOptions::dataType(hXShapeInfo);
        auto yType = sd::ArrayOptions::dataType(hScalarShapeInfo);
        auto zType = sd::ArrayOptions::dataType(hZShapeInfo);

        if (yType != xType && yType != sd::DataType::BOOL && !isExperimentalEnabled())
            throw sd::datatype_exception::build("execScalar both operands must have same data type", xType, yType);

        dim3 launchDims(256, 256, 16384);

#ifdef __ND4J_EXPERIMENTAL__
        BUILD_PAIRWISE_SELECTOR(xType, yType, zType, functions::scalar::ScalarTransform, ::executeCudaAlongDimension(launchDims, stream, opNum, dX, dXShapeInfo, dZ, dZShapeInfo, dScalars, extraParams, dimension, dimensionLength, tadShapeInfo, tadOffsets, tadShapeInfoZ, tadOffsetsZ), LIBND4J_TYPES, LIBND4J_TYPES);
#else
        BUILD_SINGLE_SELECTOR_THRICE(xType, functions::scalar::ScalarTransform, ::executeCudaAlongDimension(launchDims, stream, opNum, dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(), dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>(), dbScalars->special(), extraParams, dimension, dimensionLength, tadShapeInfo, tadOffsets, tadShapeInfoZ, tadOffsetsZ), LIBND4J_TYPES);
#endif

        DEBUG_KERNEL(stream, opNum);

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX, dbScalars});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

void execAggregate(Nd4jPointer *extraPointers,
								   int opNum,
                                   void **arguments,
                                   int numArguments,
                                   Nd4jLong **shapes,
                                   int numShapes,
                                   int *indexArguments,
                                   int numIndexArguments,
                                   int **intArrays,
                                   int numIntArrays,
                                   void *realArguments,
                                   int numRealArguments,
                                   sd::DataType dtype) {

}

void batchExecutor(Nd4jPointer *extraPointers,
                               int numAggregates,
                               int opNum,
                               int maxArgs,
                               int maxShapes,
                               int maxIntArrays,
                               int maxIntArraySize,
                               int maxIdx,
                               int maxReals,
                               void *ptrToArguments,
                               sd::DataType dtype) {
}

void execAggregateBatch(Nd4jPointer *extraPointers,
									int numAggregates, int opNum,
									int maxArgs, int maxShapes,
									int maxIntArrays, int maxIntArraySize,
									int maxIdx, int maxReals,
									void *ptrToArguments, sd::DataType dtype) {

}

////////////////////////////////////////////////////////////////////////
void execRandom(Nd4jPointer *extraPointers,
						  int opNum,
                          Nd4jPointer stateHost,
                          OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo,
                          void *extraArguments) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {});

        LaunchContext lc(extraPointers[1], extraPointers[4], extraPointers[5], extraPointers[3]);
        NativeOpExecutioner::execRandom(&lc, opNum, stateHost,
                dbZ->primary(), hZShapeInfo, dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>(),
                extraArguments);

        InteropDataBuffer::registerSpecialUse({dbZ}, {});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

////////////////////////////////////////////////////////////////////////
void execRandom2(Nd4jPointer *extraPointers, int opNum, Nd4jPointer stateHost,
                           OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
                           OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo,
						   void *extraArguments) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX});

        LaunchContext lc(extraPointers[1], extraPointers[4], extraPointers[5], extraPointers[3]);
        NativeOpExecutioner::execRandom(&lc, opNum, stateHost,
                dbX->primary(), hXShapeInfo, dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(),
                dbZ->primary(), hZShapeInfo, dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>(),
                extraArguments);

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

////////////////////////////////////////////////////////////////////////
void execRandom3(Nd4jPointer *extraPointers, int opNum, Nd4jPointer stateHost,
                            OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
                            OpaqueDataBuffer *dbY, Nd4jLong *hYShapeInfo, Nd4jLong *dYShapeInfo,
                            OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo,
							void *extraArguments) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX, dbY});

        LaunchContext lc(extraPointers[1], extraPointers[4], extraPointers[5], extraPointers[3]);
        NativeOpExecutioner::execRandom(&lc, opNum, stateHost,
                dbX->primary(), hXShapeInfo, dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(),
                dbY->primary(), hYShapeInfo, dbY->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hYShapeInfo).specialAsT<Nd4jLong>(),
                dbZ->primary(), hZShapeInfo, dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>(),
                extraArguments);

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX, dbY});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}


Nd4jPointer initRandom(Nd4jPointer *extraPointers, long seed, long bufferSize, Nd4jPointer ptrToBuffer) {

    unsigned long long *ptrHost = reinterpret_cast<unsigned long long *>(extraPointers[0]);
    cudaStream_t *stream = reinterpret_cast<cudaStream_t *>(extraPointers[1]);

    // we don't synchronize at random initialization, it's safe to go unsync here
	// cudaStreamSynchronize(*stream);

    auto ptrDev = reinterpret_cast<unsigned long long *>(ptrToBuffer);
    auto buffer = new sd::random::RandomBuffer(seed, bufferSize, reinterpret_cast<uint64_t *>(ptrHost), reinterpret_cast<uint64_t *>(ptrDev));
    buffer->propagateToDevice(buffer, *stream);

    sd::DebugHelper::checkErrorCode(stream, "initRandom(...) failed A");

	// we generate sequence in the host memory
    sd::random::Xoroshiro128 generator(buffer);
    generator.refreshBuffer();

	// and copy it to gpu
    cudaMemcpyAsync(ptrDev, ptrHost, bufferSize * 8, cudaMemcpyHostToDevice, *stream);
    sd::DebugHelper::checkErrorCode(stream, "initRandom(...) failed B");

    return buffer;
}


void destroyRandom(Nd4jPointer ptrBuffer) {

    sd::random::RandomBuffer *buffer = reinterpret_cast<sd::random::RandomBuffer *> (ptrBuffer);

    // FIXME: it's bad thing, but we can't know in advance, which stream(s) where using this generator in practice
    cudaDeviceSynchronize();

    delete buffer;
}

void refreshBuffer(Nd4jPointer *extraPointers, long seed, Nd4jPointer ptrRandom) {

    sd::random::RandomBuffer *buffer = reinterpret_cast<sd::random::RandomBuffer *> (ptrRandom);

    unsigned long long *ptrHost = reinterpret_cast<unsigned long long *>(extraPointers[0]);
    cudaStream_t *stream = reinterpret_cast<cudaStream_t *>(extraPointers[1]);
    cudaStreamSynchronize(*stream);

    uint64_t *ptrDev = buffer->getDeviceBuffer();

	// update rng state
    buffer->setSeed(seed);
    buffer->setOffset(0);
    buffer->propagateToDevice(buffer, *stream);

	// refresh buffer on host size
    sd::random::Xoroshiro128 generator(buffer);
    generator.refreshBuffer();

	// copy back to gpu
    cudaMemcpyAsync(ptrDev, ptrHost, buffer->getSize() * 8, cudaMemcpyHostToDevice, *stream);
}

void reSeedBuffer(Nd4jPointer *extraPointers, long seed, Nd4jPointer ptrRandom) {

    sd::random::RandomBuffer *buffer = reinterpret_cast<sd::random::RandomBuffer *> (ptrRandom);

    cudaStream_t *stream = reinterpret_cast<cudaStream_t *>(extraPointers[1]);
    cudaStreamSynchronize(*stream);

	// update rng state
    buffer->reSeed(seed);
    buffer->setOffset(0);
    buffer->propagateToDevice(buffer, *stream);
}



/**
    * Return the length of a shape buffer
    * based on the pointer
    * @param buffer  the buffer pointer to check
    * @return
    */
int lengthForShapeBufferPointer(Nd4jPointer buffer) {
    auto shapeBuffer = reinterpret_cast<Nd4jLong *>(buffer);
    return shape::shapeInfoLength(shape::rank(shapeBuffer));
}


/**
  * The pointer to get the address for
  *
  * @param address the address to get the pointer
  * @return the pointer for the given address
  */

Nd4jPointer pointerForAddress(Nd4jLong address) {
	return reinterpret_cast<Nd4jPointer >(address);
}

void tear(Nd4jPointer *extras,
                     OpaqueDataBuffer *dbX, Nd4jLong *xShapeInfo, Nd4jLong *dXShapeInfo,
					 Nd4jPointer *targets,
					 Nd4jLong *zShapeInfo,
					 Nd4jLong *tadShapeInfo,
					 Nd4jLong *tadOffsets) {
    try {
        InteropDataBuffer::prepareSpecialUse({}, {dbX});

        cudaStream_t *stream = reinterpret_cast<cudaStream_t *>(extras[1]);
        dim3 launchDims(512, 512, 512);
        auto xType = sd::ArrayOptions::dataType(xShapeInfo);
        BUILD_SINGLE_SELECTOR(xType, tearKernelGeneric,
                              (launchDims, stream, dbX->special(), dXShapeInfo, targets, zShapeInfo, tadShapeInfo, tadOffsets),
                              LIBND4J_TYPES);

        sd::DebugHelper::checkErrorCode(stream, "tearFloat(...) failed");

        InteropDataBuffer::registerSpecialUse({}, {dbX});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}


void prescanArrayRecursive(Nd4jPointer *extras, int *dZ, int *dX, int numElements, int level) {

    auto stream = reinterpret_cast<cudaStream_t *>(extras[1]);
    auto g_scanBlockSums = reinterpret_cast<int **>(extras[2]);

    int blockSize = 512; // max size of the thread blocks
    int numBlocks = sd::math::nd4j_max<int>(1, static_cast<int>(ceil(static_cast<float>(numElements) / (2.f * blockSize))));
    int numThreads;

    if (numBlocks > 1)
        numThreads = blockSize;
    else if (sd::isPowerOfTwo(numElements))
        numThreads = numElements / 2;
    else
        numThreads = sd::floorPow2(numElements);

    int numEltsPerBlock = numThreads * 2;

    // if this is a non-power-of-2 array, the last block will be non-full
    // compute the smallest power of 2 able to compute its scan.
    int numEltsLastBlock =
            numElements - (numBlocks-1) * numEltsPerBlock;
    int numThreadsLastBlock = sd::math::nd4j_max<int>(1, numEltsLastBlock / 2);
    int np2LastBlock = 0;
    int sharedMemLastBlock = 0;

    if (numEltsLastBlock != numEltsPerBlock) {
        np2LastBlock = 1;

        if(!isPowerOfTwo(numEltsLastBlock))
            numThreadsLastBlock = floorPow2(numEltsLastBlock);

        unsigned int extraSpace = (2 * numThreadsLastBlock) / NUM_BANKS;
        sharedMemLastBlock = sizeof(int) * (2 * numThreadsLastBlock + extraSpace);
    }

    // padding space is used to avoid shared memory bank conflicts
    int extraSpace = numEltsPerBlock / NUM_BANKS;
    int sharedMemSize = sizeof(int) * (numEltsPerBlock + extraSpace);

    // setup execution parameters
    // if NP2, we process the last block separately
    dim3 grid(max(1, numBlocks - np2LastBlock), 1, 1);
    dim3 threads(numThreads, 1, 1);
    dim3 gridOnes(1, 1, 1);
    dim3 threadsOnes(numThreadsLastBlock, 1, 1);

    if (sharedMemSize < 2048)
        sharedMemSize = 2048;

    if (sharedMemLastBlock < 2048)
        sharedMemLastBlock = 2048;

    // execute the scan
    if (numBlocks > 1) {
        sd::prescanLauncher<true, false>(grid, threads, sharedMemSize, stream, dZ, dX, g_scanBlockSums[level], numThreads * 2, 0, 0);
        if (np2LastBlock) {
            sd::prescanLauncher<true, true>(gridOnes, threadsOnes, sharedMemLastBlock, stream, dZ, dX, g_scanBlockSums[level], numEltsLastBlock, numBlocks - 1, numElements - numEltsLastBlock);
        }

        // After scanning all the sub-blocks, we are mostly done.  But now we
        // need to take all of the last values of the sub-blocks and scan those.
        // This will give us a new value that must be sdded to each block to
        // get the final results.
        // recursive (CPU) call
        prescanArrayRecursive(extras, g_scanBlockSums[level], g_scanBlockSums[level], numBlocks, level+1);

        sd::uniformAdd<<<grid, threads, 1024, *stream>>>(dZ, g_scanBlockSums[level], numElements - numEltsLastBlock, 0, 0);

        if (np2LastBlock) {
            sd::uniformAdd<<<1, numThreadsLastBlock, 1024, *stream>>>(dZ, g_scanBlockSums[level], numEltsLastBlock, numBlocks - 1, numElements - numEltsLastBlock);
        }
    } else if (isPowerOfTwo(numElements)) {
        sd::prescanLauncher<false, false>(grid, threads, sharedMemSize, stream, dZ, dX, 0, numThreads * 2, 0, 0);
    } else {
        sd::prescanLauncher<false, true>(grid, threads, sharedMemSize, stream, dZ, dX, 0, numElements, 0, 0);
    }

    sd::DebugHelper::checkErrorCode(stream, "prescanArray(...) failed");
}


void encodeThresholdP1(Nd4jPointer *extras, void *dx, Nd4jLong *hXShapeInfo, Nd4jLong N, int *dz, float threshold) {
    try {
        cudaStream_t *stream = reinterpret_cast<cudaStream_t *>(extras[1]);

        int blockSize = 1024;
        int numBlocks = N / blockSize + (N % blockSize ? 1 : 0);

        dim3 launchDims(numBlocks, blockSize, 1024);
        auto xType = sd::ArrayOptions::dataType(hXShapeInfo);
        BUILD_SINGLE_SELECTOR(xType, encoderKernelP1Generic, (launchDims, stream, dx, N, dz, threshold), LIBND4J_TYPES);

        sd::DebugHelper::checkErrorCode(stream, "encodeThresholdP1Float(...) failed");
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}



void encodeThresholdP2Int(Nd4jPointer *extraPointers, int *dx, Nd4jLong N, int *dz) {
    try {
        cudaStream_t *stream = reinterpret_cast<cudaStream_t *>(extraPointers[1]);
        //encoderKernelP2Float<<<numBlocks, blockSize , 1024 * sizeof(float), *stream>>>(dx, N, dz);
        prescanArrayRecursive(extraPointers, dz, dx + 1, (int) N, 0);
        sd::DebugHelper::checkErrorCode(stream, "encodeThresholdP2Int(...) failed");
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

void encodeThresholdP3(Nd4jPointer *extraPointers, void *dx, Nd4jLong *hXShapeInfo, int *offsets, Nd4jLong N, int *dz){
    try {
        cudaStream_t *stream = reinterpret_cast<cudaStream_t *>(extraPointers[1]);

        int blockSize = 1024;
        int numBlocks = N / blockSize + (N % blockSize ? 1 : 0);

        dim3 launchDims(numBlocks, blockSize, 4096);
        auto xType = sd::ArrayOptions::dataType(hXShapeInfo);
        BUILD_SINGLE_SELECTOR(xType, encoderKernelP3Generic, (launchDims, stream, dx, offsets, N, dz), LIBND4J_TYPES);

        sd::DebugHelper::checkErrorCode(stream, "encodeThresholdP3Float(...) failed");
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

void decodeThreshold(Nd4jPointer *extraPointers, void *dx, Nd4jLong N, void *dz, Nd4jLong *zShapeInfo){
    try {
        cudaStream_t *stream = reinterpret_cast<cudaStream_t *>(extraPointers[1]);

        // we probably want to have smaller blocks here, memory writes are misaligned anyway
        int blockSize = 128;
        int numBlocks = N / blockSize + (N % blockSize ? 1 : 0);

        dim3 launchDims(numBlocks, blockSize, 1024);
        auto zType = sd::ArrayOptions::dataType(zShapeInfo);
        BUILD_SINGLE_SELECTOR(zType, decoderKernelGeneric, (launchDims, stream, dx, N, dz), LIBND4J_TYPES);

        sd::DebugHelper::checkErrorCode(stream, "decodeThresholdFloat(...) failed");
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

////////////////////////////////////////////////////////////////////////
void execReduce3All(Nd4jPointer *extraPointers,
									int opNum,
                                    OpaqueDataBuffer *dbX, Nd4jLong *hXShapeInfo, Nd4jLong *dXShapeInfo,
                            		void *extraParamsVals,
                                    OpaqueDataBuffer *dbY, Nd4jLong *hYShapeInfo, Nd4jLong *dYShapeInfo,
                                    OpaqueDataBuffer *dbZ, Nd4jLong *hZShapeInfo, Nd4jLong *dZShapeInfo,
                                    OpaqueDataBuffer *dbDimension, Nd4jLong *hDimensionShape, Nd4jLong *dDimensionShape,
									Nd4jLong *xTadShapeInfo, Nd4jLong *xOffsets,
									Nd4jLong *yTadShapeInfo, Nd4jLong *yOffsets) {
    try {
        InteropDataBuffer::prepareSpecialUse({dbZ}, {dbX, dbY, dbDimension});
        InteropDataBuffer::preparePrimaryUse({}, {dbDimension});

        auto dimension = reinterpret_cast<int *>(dbDimension->primary());
        int dimensionLength = static_cast<int>(shape::length(hDimensionShape));

        LaunchContext lc(extraPointers[1], extraPointers[4], extraPointers[5], extraPointers[3]);
        NativeOpExecutioner::execReduce3All(&lc, opNum,
                dbX->primary(), hXShapeInfo, dbX->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hXShapeInfo).specialAsT<Nd4jLong>(),
                extraParamsVals,
                dbY->primary(), hYShapeInfo, dbY->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hYShapeInfo).specialAsT<Nd4jLong>(),
                dbZ->primary(), hZShapeInfo, dbZ->special(), ConstantShapeHelper::getInstance()->bufferForShapeInfo(hZShapeInfo).specialAsT<Nd4jLong>(),
                reinterpret_cast<int *>(dbDimension->special()), dimensionLength,
                xTadShapeInfo, xOffsets, yTadShapeInfo, yOffsets);

        InteropDataBuffer::registerSpecialUse({dbZ}, {dbX, dbY});
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}


void sort(Nd4jPointer *extraPointers,
					 void *x, Nd4jLong *xShapeInfo,
					 void *dX, Nd4jLong *dXShapeInfo,
					 bool descending) {
    try {
        cudaStream_t *stream = reinterpret_cast<cudaStream_t *>(extraPointers[1]);

        auto xLength = shape::length(xShapeInfo);
        auto xEWS = shape::elementWiseStride(xShapeInfo);
        auto xType = sd::ArrayOptions::dataType(xShapeInfo);


        // check if xLength is a power of 2, and use bitonic sort, if that's the case
        if ((xLength != 0) && ((xLength & (xLength - 1)) == 0) && (xLength <= 1024 * 1024 * 10)) {
            int numThreads = sd::math::nd4j_min<int>(512, xLength);
            int numBlocks = xLength / numThreads;
            if (xLength % numThreads > 0 || numBlocks == 0)
                numBlocks++;

            dim3 launchDims(numBlocks, numThreads, 32768);

            for (int k = 2; k <= xLength; k = 2 * k) {
                for (int j = k >> 1; j > 0; j = j >> 1) {
                    BUILD_SINGLE_SELECTOR(xType, bitonicSortStepGeneric,
                                          (launchDims, stream, dX, dXShapeInfo, j, k, xLength, descending),
                                          LIBND4J_TYPES);
                }
            }
        } else {
            int numThreads = sd::math::nd4j_min<int>(512, xLength);
            int numBlocks = xLength / numThreads;
            if (xLength % numThreads > 0 || numBlocks == 0)
                numBlocks++;

            numBlocks = sd::math::nd4j_min<int>(512, numBlocks);
            dim3 launchDims(numBlocks, numThreads, 32768);

            int max = 2, dg = 0;
            while (max < xLength) {
                max <<= 1;
                dg++;
            }
            max <<= 1;

            for (int window = 2; window < max; window <<= 1) {
                int n = window;
                int rev = 0;
                do {
                    int half = n >> 1;
                    BUILD_SINGLE_SELECTOR(xType, bitonicArbitraryStepGeneric,
                                          (launchDims, stream, dX, dXShapeInfo, n, xLength, rev, descending),
                                          LIBND4J_TYPES);
                    n >>= 1;
                    rev = 1;
                } while (n > 1);
            }
        }

        sd::DebugHelper::checkErrorCode(stream, "sort(...) failed");
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}


void sortByKey(Nd4jPointer *extraPointers,
               void *x, Nd4jLong *xShapeInfo,
               void *dX, Nd4jLong *dXShapeInfo,
               void *y, Nd4jLong *yShapeInfo,
               void *dy, Nd4jLong *dyShapeInfo,
               bool descending) {
    try {
        auto stream = reinterpret_cast<cudaStream_t *>(extraPointers[1]);

        auto xLength = shape::length(xShapeInfo);
        auto yLength = shape::length(yShapeInfo);
        auto xEWS = shape::elementWiseStride(xShapeInfo);
        auto xType = sd::ArrayOptions::dataType(xShapeInfo);
        auto yType = sd::ArrayOptions::dataType(yShapeInfo);

        if (shape::isEmpty(xShapeInfo) || shape::isEmpty(yShapeInfo))
            return;

        if (xLength != yLength)
            throw std::runtime_error("sortByKey: keys and values must have the same size");


        // check if xLength is a power of 2, and use bitonic sort, if that's the case
        if ((xLength != 0) && ((xLength & (xLength - 1)) == 0) && (xLength <= 1024 * 1024 * 10)) {
            int numThreads = sd::math::nd4j_min<int>(512, xLength);
            int numBlocks = xLength / numThreads;
            if (xLength % numThreads > 0 || numBlocks == 0)
                numBlocks++;

            dim3 launchDims(numBlocks, numThreads, 32768);

            for (int k = 2; k <= xLength; k = 2 * k) {
                for (int j = k >> 1; j > 0; j = j >> 1) {
                    BUILD_DOUBLE_SELECTOR(xType, yType, bitonicSortStepGenericKey,
                                          (launchDims, stream, dX, dXShapeInfo, dy, dyShapeInfo, j, k, xLength, descending),
                                          LIBND4J_TYPES, LIBND4J_TYPES);
                }
            }
        } else {
            int numThreads = sd::math::nd4j_min<int>(512, xLength);
            int numBlocks = xLength / numThreads;
            if (xLength % numThreads > 0 || numBlocks == 0)
                numBlocks++;

            numBlocks = sd::math::nd4j_min<int>(512, numBlocks);
            dim3 launchDims(numBlocks, numThreads, 32768);

            int max = 2, dg = 0;
            while (max < xLength) {
                max <<= 1;
                dg++;
            }
            max <<= 1;

            for (int window = 2; window < max; window <<= 1) {
                int n = window;
                int rev = 0;
                do {
                    int half = n >> 1;
                    BUILD_DOUBLE_SELECTOR(xType, yType, bitonicArbitraryStepGenericKey,
                                          (launchDims, stream, dX, dXShapeInfo, dy, dyShapeInfo, n, xLength, rev, descending),
                                          LIBND4J_TYPES, LIBND4J_TYPES);
                    n >>= 1;
                    rev = 1;
                } while (n > 1);
            }
        }

    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

void sortByValue(Nd4jPointer *extraPointers,
                 void *x, Nd4jLong *xShapeInfo,
                 void *dX, Nd4jLong *dXShapeInfo,
                 void *y, Nd4jLong *yShapeInfo,
                 void *dy, Nd4jLong *dyShapeInfo,
                 bool descending) {
    try {
        auto stream = reinterpret_cast<cudaStream_t *>(extraPointers[1]);

        auto xLength = shape::length(xShapeInfo);
        auto yLength = shape::length(yShapeInfo);
        auto xEWS = shape::elementWiseStride(xShapeInfo);
        auto xType = sd::ArrayOptions::dataType(yShapeInfo);
        auto yType = sd::ArrayOptions::dataType(xShapeInfo);

        if (shape::isEmpty(xShapeInfo) || shape::isEmpty(yShapeInfo))
            return;

        if (xLength != yLength)
            throw std::runtime_error("sortByValue: keys and values must have the same size");


        // check if xLength is a power of 2, and use bitonic sort, if that's the case
        if ((xLength != 0) && ((xLength & (xLength - 1)) == 0) && (xLength <= 1024 * 1024 * 10)) {
            int numThreads = sd::math::nd4j_min<int>(512, xLength);
            int numBlocks = xLength / numThreads;
            if (xLength % numThreads > 0 || numBlocks == 0)
                numBlocks++;

            dim3 launchDims(numBlocks, numThreads, 32768);

            for (int k = 2; k <= xLength; k = 2 * k) {
                for (int j = k >> 1; j > 0; j = j >> 1) {
                    BUILD_DOUBLE_SELECTOR(xType, yType, bitonicSortStepGenericKey,
                                          (launchDims, stream, dy, dyShapeInfo, dX, dXShapeInfo, j, k, xLength, descending),
                                          LIBND4J_TYPES, LIBND4J_TYPES);
                }
            }
        } else {
            int numThreads = sd::math::nd4j_min<int>(512, xLength);
            int numBlocks = xLength / numThreads;
            if (xLength % numThreads > 0 || numBlocks == 0)
                numBlocks++;

            numBlocks = sd::math::nd4j_min<int>(512, numBlocks);
            dim3 launchDims(numBlocks, numThreads, 32768);

            int max = 2, dg = 0;
            while (max < xLength) {
                max <<= 1;
                dg++;
            }
            max <<= 1;

            for (int window = 2; window < max; window <<= 1) {
                int n = window;
                int rev = 0;
                do {
                    int half = n >> 1;
                    BUILD_DOUBLE_SELECTOR(xType, yType, bitonicArbitraryStepGenericKey,
                                          (launchDims, stream, dy, dyShapeInfo, dX, dXShapeInfo, n, xLength, rev, descending),
                                          LIBND4J_TYPES, LIBND4J_TYPES);
                    n >>= 1;
                    rev = 1;
                } while (n > 1);
            }
        }
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}



void sortTadByKey(Nd4jPointer *extraPointers,
                             void *x, Nd4jLong *xShapeInfo,
                             void *dX, Nd4jLong *dXShapeInfo,
                             void *y, Nd4jLong *yShapeInfo,
                             void *dy, Nd4jLong *dyShapeInfo,
                             int *dimension,
                             int dimensionLength,
                             bool descending) {
    try {
        auto stream = reinterpret_cast<cudaStream_t *>(extraPointers[1]);
        auto context = extraPointers[0] == 0 ? LaunchContext::defaultContext()
                                             : reinterpret_cast<LaunchContext *>(extraPointers[0]);
        auto tadPack = sd::ConstantTadHelper::getInstance()->tadForDimensions(xShapeInfo, dimension, dimensionLength);
        dim3 launchDims((int) tadPack.numberOfTads(), 256, 2048);
        auto xType = sd::ArrayOptions::dataType(xShapeInfo);
        auto yType = sd::ArrayOptions::dataType(yShapeInfo);
        BUILD_DOUBLE_SELECTOR(xType, yType, oesTadGenericKey,
                              (launchDims, stream, dX, dXShapeInfo, dy, dyShapeInfo, nullptr, dimensionLength, tadPack.platformShapeInfo(), tadPack.platformOffsets(), descending),
                              LIBND4J_TYPES, LIBND4J_TYPES);

        sd::DebugHelper::checkErrorCode(stream, "sortTadKey(...) failed");
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

void sortTadByValue(Nd4jPointer *extraPointers,
                               void *x, Nd4jLong *xShapeInfo,
                               void *dX, Nd4jLong *dXShapeInfo,
                               void *y, Nd4jLong *yShapeInfo,
                               void *dy, Nd4jLong *dyShapeInfo,
                               int *dimension,
                               int dimensionLength,
                               bool descending) {
    try {
        auto stream = reinterpret_cast<cudaStream_t *>(extraPointers[1]);
        auto context = extraPointers[0] == 0 ? LaunchContext::defaultContext()
                                             : reinterpret_cast<LaunchContext *>(extraPointers[0]);
        auto tadPack = sd::ConstantTadHelper::getInstance()->tadForDimensions(xShapeInfo, dimension, dimensionLength);
        dim3 launchDims((int) tadPack.numberOfTads(), 256, 2048);
        auto xType = sd::ArrayOptions::dataType(yShapeInfo);
        auto yType = sd::ArrayOptions::dataType(xShapeInfo);

        BUILD_DOUBLE_SELECTOR(xType, yType, oesTadGenericKey,
                              (launchDims, stream, dy, dyShapeInfo, dX, dXShapeInfo, nullptr, dimensionLength, tadPack.platformShapeInfo(), tadPack.platformOffsets(), descending),
                              LIBND4J_TYPES, LIBND4J_TYPES);

        sd::DebugHelper::checkErrorCode(stream, "sortTadValue(...) failed");
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}


void sortTad(Nd4jPointer *extraPointers,
						void *x, Nd4jLong *xShapeInfo,
						void *dX, Nd4jLong *dXShapeInfo,
						int *dimension,
						int dimensionLength,
						Nd4jLong *tadShapeInfo,
						Nd4jLong *tadOffsets,
						bool descending) {
    try {
        // to be implemented
        auto stream = reinterpret_cast<cudaStream_t *>(extraPointers[1]);
        auto context = extraPointers[0] == 0 ? LaunchContext::defaultContext()
                                             : reinterpret_cast<LaunchContext *>(extraPointers[0]);
        auto tadPack = sd::ConstantTadHelper::getInstance()->tadForDimensions(xShapeInfo, dimension, dimensionLength);
        dim3 launchDims((int) tadPack.numberOfTads(), 512, 33768);
        auto xType = sd::ArrayOptions::dataType(xShapeInfo);
        BUILD_SINGLE_SELECTOR(xType, oesTadGeneric,
                              (launchDims, stream, dX, dXShapeInfo, nullptr, dimensionLength, tadShapeInfo, tadOffsets, descending),
                              LIBND4J_TYPES);

        sd::DebugHelper::checkErrorCode(stream, "sortTad(...) failed");
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

void sortCooIndices(Nd4jPointer *extraPointers, Nd4jLong *indices, void *values, Nd4jLong length, int rank) {
	throw std::runtime_error("sortCooIndices:: Not implemented yet");
}


Nd4jLong encodeBitmap(Nd4jPointer *extraPointers,
								void *dx, Nd4jLong *hXShapeInfo,
								Nd4jLong N,
								int *dz,
								float threshold) {
    try {

        cudaStream_t *stream = reinterpret_cast<cudaStream_t *>(extraPointers[1]);
        int *resultPointer = reinterpret_cast<int *>(extraPointers[2]);
        int *reductionPointer = reinterpret_cast<int *>(extraPointers[3]);

        dim3 launchDims(512, 512, 32768);
        auto xType = sd::ArrayOptions::dataType(hXShapeInfo);
        BUILD_SINGLE_SELECTOR(xType, cudaEncodeBitmapGeneric,
                              (launchDims, stream, dx, N, dz, resultPointer, reductionPointer, threshold),
                              LIBND4J_TYPES);

        sd::DebugHelper::checkErrorCode(stream, "encodeBitmapFloat(...) failed");

        Nd4jLong dZ = (Nd4jLong) resultPointer[0];
        resultPointer[0] = 0;

        return dZ;
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
        return 0;
    }
}


void decodeBitmap(Nd4jPointer *extraPointers,
							void *dx,
							Nd4jLong N,
							void *dz, Nd4jLong *zShapeInfo) {
    try {
        cudaStream_t *stream = reinterpret_cast<cudaStream_t *>(extraPointers[1]);
        dim3 launchDims(512, 512, 16384);
        auto xType = sd::ArrayOptions::dataType(zShapeInfo);
        BUILD_SINGLE_SELECTOR(xType, cudaDecodeBitmapGeneric, (launchDims, stream, dx, N, dz), LIBND4J_TYPES);

        sd::DebugHelper::checkErrorCode(stream, "decodeBitmapFloat(...) failed");
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

Nd4jLong* mmapFile(Nd4jPointer *extraPointers, const char *fileName, Nd4jLong length) {
	return nullptr;
}

void munmapFile(Nd4jPointer *extraPointers, Nd4jLong* ptrMap, Nd4jLong length) {

}


sd::graph::ResultWrapper* executeFlatGraph(Nd4jPointer *extraPointers, Nd4jPointer flatBufferPointer) {
    try {
        return sd::graph::GraphExecutioner::executeFlatBuffer(flatBufferPointer);
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
        return nullptr;
    }
}

Nd4jLong getResultWrapperSize(sd::graph::ResultWrapper* ptr) {
    return ptr->size();
}
Nd4jPointer getResultWrapperPointer(sd::graph::ResultWrapper* ptr) {
    return ptr->pointer();
}


const char* getAllCustomOps() {
	return sd::ops::OpRegistrator::getInstance()->getAllCustomOperations();
}


sd::ShapeList* _calculateOutputShapes(Nd4jPointer* extraPointers, sd::ops::DeclarableOp* op, Nd4jPointer* inputBuffers, Nd4jPointer* inputShapes, int numInputShapes, double* tArgs, int numTArgs, Nd4jLong *iArgs, int numIArgs, bool *bArgs, int numBArgs, int *dArgs, int numDArgs) {
    sd::graph::VariableSpace varSpace;
    Context block(2, &varSpace);
    sd::ShapeList inShapes;

    for (int e = 0; e < numIArgs; e++)
        block.getIArguments()->push_back(iArgs[e]);

    for (int e = 0; e < numTArgs; e++)
        block.getTArguments()->push_back(tArgs[e]);

	for (int e = 0; e < numBArgs; e++)
		block.getBArguments()->push_back(bArgs[e]);

    for (int e = 0; e < numDArgs; e++)
        block.getDArguments()->push_back((sd::DataType) dArgs[e]);

	for (int e = 0; e < numInputShapes; e++) {
		auto shape_ = reinterpret_cast<Nd4jLong *>(inputShapes[e]);

		// we shouldn't copy buffer if that's empty array
		void *buffer_ = sd::ArrayOptions::arrayType(shape_) == ArrayType::EMPTY ? nullptr : inputBuffers[e];
        void *bufferD_ = sd::ArrayOptions::arrayType(shape_) == ArrayType::EMPTY ? nullptr : inputBuffers[e + numInputShapes];

		auto array = new sd::NDArray(buffer_, bufferD_, shape_);

		// block should contain references to proper variable
		varSpace.putVariable(1, e, array);
		block.pickInput(1, e);

		inShapes.push_back(shape_);
	}

    auto shapeList = op->calculateOutputShape(&inShapes, block);

    if (varSpace.launchContext()->getWorkspace() != nullptr)
        shapeList->detach();

    return shapeList;
}

sd::ShapeList* calculateOutputShapes2(Nd4jPointer* extraPointers, Nd4jLong hash, Nd4jPointer* inputBuffers, Nd4jPointer* inputShapes, int numInputShapes, double* tArgs, int numTArgs, Nd4jLong *iArgs, int numIArgs, bool *bArgs, int numBArgs, int *dArgs, int numDArgs) {
    try {
        auto op = sd::ops::OpRegistrator::getInstance()->getOperation(hash);

        return _calculateOutputShapes(extraPointers, op, inputBuffers, inputShapes, numInputShapes, tArgs, numTArgs,
                                      iArgs, numIArgs, bArgs, numBArgs, dArgs, numDArgs);
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
        return nullptr;
    }
}

sd::ShapeList* _calculateOutputShapes(Nd4jPointer* extraPointers, sd::ops::DeclarableOp* op, Nd4jPointer* inputShapes, int numInputShapes, double* tArgs, int numTArgs, Nd4jLong *iArgs, int numIArgs) {
    Context block(1);
	sd::ShapeList inShapes;

	for (int e = 0; e < numIArgs; e++)
		block.getIArguments()->push_back(iArgs[e]);

	for (int e = 0; e < numTArgs; e++)
		block.getTArguments()->push_back(tArgs[e]);

	for (int e = 0; e < numInputShapes; e++)
		inShapes.push_back(reinterpret_cast<Nd4jLong *>(inputShapes[e]));

	auto shapeList = op->calculateOutputShape(&inShapes, block);

	return shapeList;
}

sd::ShapeList* calculateOutputShapes(Nd4jPointer* extraPointers, Nd4jLong hash, Nd4jPointer* inputShapes, int numInputShapes, double* tArgs, int numTArgs, Nd4jLong *iArgs, int numIArgs) {
    try {
        auto op = sd::ops::OpRegistrator::getInstance()->getOperation(hash);

        return _calculateOutputShapes(extraPointers, op, inputShapes, numInputShapes, tArgs, numTArgs, iArgs, numIArgs);
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
        return nullptr;
    }
}

Nd4jLong getShapeListSize(sd::ShapeList* list) {
    return list->size();
}

Nd4jLong* getShape(sd::ShapeList* list, Nd4jLong i) {
    return list->at(i);
}

static FORCEINLINE Nd4jStatus realExec(sd::ops::DeclarableOp* op, Nd4jPointer* extraPointers, Nd4jLong hash, Nd4jPointer* inputBuffers, Nd4jPointer* inputShapes, int numInputs, Nd4jPointer* outputBuffers, Nd4jPointer* outputShapes, int numOutputs, double* tArgs, int numTArgs, Nd4jLong *iArgs, int numIArgs, bool* bArgs, int numBArgs, bool isInplace) {
	if (op == nullptr)
		nd4j_printf("Can't find requested operation: [%lld]\n", hash);

	// we're using the same fake nodeId everywhere here

	std::vector<sd::NDArray*> inputs(numInputs);
	std::vector<sd::NDArray*> outputs(numOutputs);
	std::vector<double> ttArgs(numTArgs);
	std::vector<bool> bbArgs(numBArgs);
	std::vector<Nd4jLong> iiArgs(numIArgs);

	// filling block now with inputs
	for (int e = 0; e < numInputs; e++) {
		auto shape = reinterpret_cast<Nd4jLong *>(inputShapes[e]);
		void *buffer = sd::ArrayOptions::arrayType(shape) == ArrayType::EMPTY ? nullptr : inputBuffers[e];
        void *bufferD = sd::ArrayOptions::arrayType(shape) == ArrayType::EMPTY ? nullptr : inputBuffers[e + numInputs];

		inputs[e] = new sd::NDArray(buffer, bufferD, shape);
	}

	// if not inplace - transferring output arrays

	if (!isInplace)
		for (int e = 0; e < numOutputs; e++) {
			// we want to keep original output shape intact
			auto shape = shape::copyShape(reinterpret_cast<Nd4jLong *>(outputShapes[e]));
			void *buffer = sd::ArrayOptions::arrayType(shape) == ArrayType::EMPTY ? nullptr : outputBuffers[e];
            void *bufferD = sd::ArrayOptions::arrayType(shape) == ArrayType::EMPTY ? nullptr : outputBuffers[e + numOutputs];

			// FIXME: revisit this.
			bool canNullify = true;
			for (int i = 0; i < numInputs; i++) {
				void *ibuffer = sd::ArrayOptions::arrayType(shape) == ArrayType::EMPTY ? nullptr : inputBuffers[i];
				if (ibuffer == buffer) {
					canNullify = false;
					break;
				}
			}

			if (canNullify && buffer != nullptr)
				memset((uint8_t *) buffer, '\0', shape::length(shape) * DataTypeUtils::sizeOfElement(ArrayOptions::dataType(shape)));

			auto array = new sd::NDArray(buffer, bufferD, shape);
			outputs[e] = array;
		}

	for (int e = 0; e < numIArgs; e++)
		iiArgs[e] = iArgs[e];

	for (int e = 0; e < numTArgs; e++)
		ttArgs[e] = tArgs[e];

    for (int e = 0; e < numBArgs; e++)
        bbArgs[e] = bArgs[e];


	// hypothetically at this point we have everything filled
	auto dZ = op->execute(inputs, outputs, ttArgs, iiArgs, bbArgs, std::vector<sd::DataType>(), isInplace);
	//auto dZ = op->execute(inputs, ttArgs, iiArgs, isInplace);


	if (!isInplace)
		for (int e = 0; e < numOutputs; e++) {
			//shape::printShapeInfoLinear("JVM output shape", (int *) outputShapes[e]);
			//shape::printShapeInfoLinear("C++ output shape", (int *) outputs[e]->shapeInfo());
			//outputs[e]->printIndexedBuffer("C++ raw output");
			//outputs[e]->printBuffer("C++ indexed output");

			if (outputs[e]->ordering() != shape::order(reinterpret_cast<Nd4jLong *>(outputShapes[e])))
				outputs[e]->streamline(shape::order(reinterpret_cast<Nd4jLong *>(outputShapes[e])));
		}

	for (auto v: inputs)
		delete v;

	for (auto v: outputs)
		delete v;

	return Status::OK();
}


int execCustomOp(Nd4jPointer* extraPointers, Nd4jLong hash, Nd4jPointer* inputBuffers, Nd4jPointer* inputShapes, int numInputs, Nd4jPointer* outputBuffers, Nd4jPointer* outputShapes, int numOutputs, double* tArgs, int numTArgs, Nd4jLong *iArgs, int numIArgs, bool* bArgs, int numBArgs, bool isInplace) {
    try {
        auto op = sd::ops::OpRegistrator::getInstance()->getOperation(hash);

        return realExec(op, extraPointers, hash, inputBuffers, inputShapes, numInputs, outputBuffers, outputShapes,
                        numOutputs, tArgs, numTArgs, iArgs, numIArgs, bArgs, numBArgs, isInplace);
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
        return 1;
    }
}

int execCustomOp2(Nd4jPointer* extraPointers, Nd4jLong hash, Nd4jPointer opContext) {
    try {
        auto op = sd::ops::OpRegistrator::getInstance()->getOperation(hash);
        auto context = reinterpret_cast<Context *>(opContext);

        auto result = op->execute(context);

        auto res = cudaStreamSynchronize(*context->launchContext()->getCudaStream());
        if (res != 0)
            throw sd::cuda_exception::build("customOp execution failed", res);

        for (auto v:context->fastpath_in()) {
            if (!v->isEmpty())
                v->syncToDevice();
        }

        for (auto v:context->fastpath_out()) {
            if (!v->isEmpty())
                v->syncToDevice();
        }

        return result;
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
        return 1;
    }
}

int registerGraph(Nd4jPointer *extraPointers, Nd4jLong graphId, Nd4jPointer flatBufferPointer) {
    try {
        auto graph = sd::graph::GraphExecutioner::importFromFlatPointer(flatBufferPointer);

        sd::graph::GraphHolder::getInstance()->registerGraph(graphId, graph);

        return ND4J_STATUS_OK;
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
        return 1;
    }
}


static VariablesSet* executeStoredGraphT(Nd4jPointer *extraPointers, Nd4jLong graphId, Nd4jPointer *inputBuffers, Nd4jPointer *inputShapes, int* inputIndices, int numInputs) {
	auto graph = sd::graph::GraphHolder::getInstance()->pullGraph(graphId);
	auto varSpace = graph->getVariableSpace()->clone();

	std::vector<sd::NDArray*> handles;

	for (int e = 0; e < numInputs; e++) {
		auto idx = inputIndices[e];

		// we'll delete this array later, together with cloned VariableSpace
		auto array = new sd::NDArray(inputBuffers[e], reinterpret_cast<Nd4jLong *>(inputShapes[e]));
		handles.emplace_back(array);

		if (varSpace->hasVariable(idx)) {
			auto var = varSpace->getVariable(idx);
			if (var->hasNDArray())
				delete var->getNDArray();

			var->setNDArray(array);
		} else
			varSpace->putVariable(idx, array);
	}

	auto dZ = sd::graph::GraphExecutioner::execute(graph, varSpace);
	auto varSet = new sd::graph::VariablesSet(dZ);

	if (dZ == ND4J_STATUS_OK) {
		// pull back results, and provide them
		auto outputs = graph->fetchOutputs();
		for (int e = 0; e < outputs->size(); e++) {
			// we're only getting variable ID/Index from original grap. values will be taken from cloned workspace
			std::pair<int, int> varId(outputs->at(e)->id(), outputs->at(e)->index());

			auto var = varSpace->getVariable(varId);

			varSet->push_back(var->clone());
		}

		delete outputs;
	}

	delete varSpace;

	return varSet;
}

VariablesSet* executeStoredGraph(Nd4jPointer *extraPointers, Nd4jLong graphId, Nd4jPointer *inputBuffers, Nd4jPointer *inputShapes, int* inputIndices, int numInputs) {
    try {
        return executeStoredGraphT(extraPointers, graphId, inputBuffers, inputShapes, inputIndices, numInputs);
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
        return nullptr;
    }
}

Nd4jLong getVariablesSetSize(sd::graph::VariablesSet* set) {
    return set->size();
}

Nd4jStatus getVariablesSetStatus(sd::graph::VariablesSet* set) {
    return set->status();
}

sd::graph::Variable* getVariable(sd::graph::VariablesSet* set, Nd4jLong i) {
    return set->at(i);
}

int getVariableId(sd::graph::Variable* variable) {
    return variable->id();
}

int getVariableIndex(sd::graph::Variable* variable) {
    return variable->index();
}

const char* getVariableName(sd::graph::Variable* variable) {
    return variable->getName()->c_str();
}

Nd4jLong* getVariableShape(sd::graph::Variable* variable) {
    return variable->getNDArray()->shapeInfo();
}

void* getVariableBuffer(sd::graph::Variable* variable) {
    return variable->getNDArray()->buffer();
}

int unregisterGraph(Nd4jPointer *extraPointers, Nd4jLong graphId) {
    try {
        sd::graph::GraphHolder::getInstance()->dropGraphAny(graphId);

        return ND4J_STATUS_OK;
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
        return 1;
    }
}

void deletePointerArray(Nd4jPointer pointer) {
    Nd4jPointer *ptr = reinterpret_cast<Nd4jPointer *>(pointer);
    delete[] ptr;
}

void deleteCharArray(Nd4jPointer pointer) {
    auto ptr = reinterpret_cast<char *>(pointer);
    delete[] ptr;
}

void deleteIntArray(Nd4jPointer pointer) {
	auto ptr = reinterpret_cast<int *>(pointer);
	delete[] ptr;
}

void deleteLongArray(Nd4jPointer pointer) {
	auto ptr = reinterpret_cast<Nd4jLong *>(pointer);
	delete[] ptr;
}

void deleteVariablesSet(sd::graph::VariablesSet* pointer) {
	delete pointer;
}

void deleteShapeList(Nd4jPointer shapeList) {
    sd::ShapeList* list = reinterpret_cast<sd::ShapeList*>(shapeList);

    //list->destroy();
    delete list;
}

const char* getAllOperations() {
    return sd::OpTracker::getInstance()->exportOperations();
}

Nd4jPointer getGraphState(Nd4jLong id) {
    return (Nd4jPointer) new sd::graph::GraphState(id);
}


void deleteGraphState(Nd4jPointer state) {
    auto stateP = reinterpret_cast<sd::graph::GraphState*>(state);
    delete stateP;
}


Nd4jStatus execCustomOpWithScope(Nd4jPointer *extraPointers, sd::graph::GraphState *state, Nd4jLong opHash, Nd4jLong *scopes, int numScopes, Nd4jPointer *inputBuffers, Nd4jPointer *inputShapes, int numInputs, Nd4jPointer *outputBuffers, Nd4jPointer *outputShapes, int numOutputs) {
    /**
     * That's basically exec, with VariableSpace provided in GraphState:
     * depending on operation (i.e. while of if), different logic executors could be used
     */

    auto graph = state->graph();
    auto varSpace = state->variableSpace();

    // Node is dynamically created, and has nothing beyond it: only inputs and outputs
    // this node has id of 0, and inputs are
    Node node(OpType_LOGIC, opHash, 0);

    // mapping inputs
    for (int e = 0; e < numInputs; e++) {
        auto buffer = inputBuffers[e];
        auto shapeInfo = reinterpret_cast<Nd4jLong *>(inputShapes[e]);

        auto array = new sd::NDArray(buffer, shapeInfo, varSpace->launchContext());

        // now we just put array to VarSpace
        varSpace->putVariable(0, e, array);
        node.pickInput(0, e);
    }

    // mapping scopes
    for (int e = 0; e < numScopes; e++) {
        // we should check scope existence in GraphState/Graph
        int scopeId = (int) scopes[e];
        if (!state->hasScope(scopeId)) {
            // nd4j_printf("execCustomOpWithScope: referenced scope [%i] doesn't exist\n", scopeId);
            return Status::THROW();
        }
        node.pickInput(scopeId, 0);
    }

    auto dZ = LogicExecutor::processNode(graph, &node);
    if (dZ != Status::OK())
        return dZ;

    // mapping outputs

    for (int e = 0; e < numOutputs; e++) {
        auto buffer = outputBuffers[e];
        auto shapeInfo = reinterpret_cast<Nd4jLong *>(outputShapes[e]);

        NDArray array(buffer, shapeInfo, varSpace->launchContext());

        // now we just put array to VarSpace to the same ID
        //varSpace->putVariable(0, e, array);

        auto t = varSpace->getVariable(0, e)->getNDArray();
        array.assign(t);
    }

    // removing input variables
    for (int e = 0; e < numInputs; e++) {
        varSpace->dropVariable(0, e);
    }

    // after some bla-bla-bla we should have Graph and Node for current op
    return Status::OK();
}


Nd4jStatus execCustomOpWithScope(Nd4jPointer *extraPointers, Nd4jPointer state, Nd4jLong opHash, Nd4jLong *scopes, int numScopes, Nd4jPointer *inputBuffers, Nd4jPointer *inputShapes, int numInputs, Nd4jPointer *outputBuffers, Nd4jPointer *outputShapes, int numOutputs) {
    try {
        return execCustomOpWithScope(extraPointers, reinterpret_cast<sd::graph::GraphState *>(state), opHash, scopes,
                                     numScopes, inputBuffers, inputShapes, numInputs, outputBuffers, outputShapes,
                                     numOutputs);
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
        return 1;
    }
}

void deleteResultWrapper(Nd4jPointer ptr) {
	// just 0 room for compiler s@!t
	auto p = reinterpret_cast<sd::graph::ResultWrapper *>(ptr);
	delete p;
}

int estimateThreshold(Nd4jPointer *extraPointers, Nd4jPointer dX, Nd4jLong *dXShapeInfo, int N, float threshold) {
	throw std::runtime_error("estimateThreshold: Not implemented yet");
}

/*
 * TypeDef:
 *     void convertTypes(Nd4jPointer *extras, int srcType, Nd4jPointer dX, long N, int dstType, Nd4jPointer dZ);
 */
void convertTypes(Nd4jPointer *extras, int srcType, Nd4jPointer dX, Nd4jLong N, int dstType, Nd4jPointer dZ) {
    try {
        auto dx = reinterpret_cast<void *>(dX);
        auto dz = reinterpret_cast<void *>(dZ);

        if (srcType == ND4J_FLOAT8) {
            if (dstType == ND4J_FLOAT8) {
                // convertKernel<double, sd::float8>(extras, dx, N, dz);
            } else if (dstType == ND4J_INT8) {
                //sd::TypeCast::convertGenericCuda<sd::float8, sd::int8>(extras, dx, N, dz);
            } else if (dstType == ND4J_UINT8) {
                //sd::TypeCast::convertGenericCuda<sd::float8, sd::uint8>(extras, dx, N, dz);
            } else if (dstType == ND4J_FLOAT16) {
                //sd::TypeCast::convertGenericCuda<sd::float8, float16>(extras, dx, N, dz);
            } else if (dstType == ND4J_INT16) {
                //sd::TypeCast::convertGenericCuda<sd::float8, sd::int16>(extras, dx, N, dz);
            } else if (dstType == ND4J_UINT16) {
                //sd::TypeCast::convertGenericCuda<sd::float8, sd::uint16>(extras, dx, N, dz);
            } else if (dstType == ND4J_FLOAT24) {

            } else if (dstType == ND4J_FLOAT32) {
                //sd::TypeCast::convertGenericCuda<sd::float8, float>(extras, dx, N, dz);
            } else if (dstType == ND4J_DOUBLE) {
                //sd::TypeCast::convertGenericCuda<sd::float8, double>(extras, dx, N, dz);
            } else {
                nd4j_printf("Unsupported types conversion: [%i] -> [%i]\n", srcType, dstType);
            }
        } else if (srcType == ND4J_INT8) {
            if (dstType == ND4J_FLOAT8) {
                //sd::TypeCast::convertGenericCuda<sd::int8, sd::float8>(extras, dx, N, dz);
            } else if (dstType == ND4J_INT8) {
                //convertKernel<sd::int8, sd::int8>(extras, dx, N, dz);
            } else if (dstType == ND4J_UINT8) {
                sd::TypeCast::convertGenericCuda<int8_t, uint8_t>(extras, dx, N, dz);
            } else if (dstType == ND4J_FLOAT16) {
                sd::TypeCast::convertGenericCuda<int8_t, float16>(extras, dx, N, dz);
            } else if (dstType == ND4J_INT16) {
                sd::TypeCast::convertGenericCuda<int8_t, int16_t>(extras, dx, N, dz);
            } else if (dstType == ND4J_UINT16) {
                sd::TypeCast::convertGenericCuda<int8_t, uint16_t>(extras, dx, N, dz);
            } else if (dstType == ND4J_FLOAT24) {
                // TODO: eventually we might want to add it
            } else if (dstType == ND4J_FLOAT32) {
                sd::TypeCast::convertGenericCuda<int8_t, float>(extras, dx, N, dz);
            } else if (dstType == ND4J_DOUBLE) {
                sd::TypeCast::convertGenericCuda<int8_t, double>(extras, dx, N, dz);
            } else {
                nd4j_printf("Unsupported types conversion: [%i] -> [%i]\n", srcType, dstType);
            }
        } else if (srcType == ND4J_UINT8) {
            if (dstType == ND4J_FLOAT8) {
                //sd::TypeCast::convertGenericCuda<uint8_t, sd::float8>(extras, dx, N, dz);
            } else if (dstType == ND4J_INT8) {
                sd::TypeCast::convertGenericCuda<uint8_t, int8_t>(extras, dx, N, dz);
            } else if (dstType == ND4J_UINT8) {
                sd::TypeCast::convertGenericCuda<uint8_t, uint8_t>(extras, dx, N, dz);
            } else if (dstType == ND4J_FLOAT16) {
                sd::TypeCast::convertGenericCuda<uint8_t, float16>(extras, dx, N, dz);
            } else if (dstType == ND4J_INT16) {
                sd::TypeCast::convertGenericCuda<uint8_t, int16_t>(extras, dx, N, dz);
            } else if (dstType == ND4J_UINT16) {
                sd::TypeCast::convertGenericCuda<uint8_t, uint16_t>(extras, dx, N, dz);
            } else if (dstType == ND4J_FLOAT24) {
                // TODO: still might want to add
            } else if (dstType == ND4J_FLOAT32) {
                sd::TypeCast::convertGenericCuda<uint8_t, float>(extras, dx, N, dz);
            } else if (dstType == ND4J_DOUBLE) {
                sd::TypeCast::convertGenericCuda<uint8_t, double>(extras, dx, N, dz);
            } else {
                nd4j_printf("Unsupported types conversion: [%i] -> [%i]\n", srcType, dstType);
            }
        } else if (srcType == ND4J_FLOAT16) {
            if (dstType == ND4J_FLOAT8) {
                //sd::TypeCast::convertGenericCuda<float16, sd::float8>(extras, dx, N, dz);
            } else if (dstType == ND4J_INT8) {
                sd::TypeCast::convertGenericCuda<float16, int8_t>(extras, dx, N, dz);
            } else if (dstType == ND4J_UINT8) {
                sd::TypeCast::convertGenericCuda<float16, uint8_t>(extras, dx, N, dz);
            } else if (dstType == ND4J_FLOAT16) {
                sd::TypeCast::convertGenericCuda<float16, float16>(extras, dx, N, dz);
            } else if (dstType == ND4J_INT16) {
                sd::TypeCast::convertGenericCuda<float16, int16_t>(extras, dx, N, dz);
            } else if (dstType == ND4J_UINT16) {
                sd::TypeCast::convertGenericCuda<float16, uint16_t>(extras, dx, N, dz);
            } else if (dstType == ND4J_FLOAT24) {
                // TODO: .... ^^^
            } else if (dstType == ND4J_FLOAT32) {
                sd::TypeCast::convertGenericCuda<float16, float>(extras, dx, N, dz);
            } else if (dstType == ND4J_DOUBLE) {
                sd::TypeCast::convertGenericCuda<float16, double>(extras, dx, N, dz);
            } else if (dstType == ND4J_THRESHOLD) {
                //sd::convertToThreshold<float16>(nullptr, dx, N, dz);
            } else {
                nd4j_printf("Unsupported types conversion: [%i] -> [%i]\n", srcType, dstType);
            }
        } else if (srcType == ND4J_INT16) {
            if (dstType == ND4J_FLOAT8) {
                //sd::TypeCast::convertGenericCuda<int16_t, sd::float8>(extras, dx, N, dz);
            } else if (dstType == ND4J_INT8) {
                sd::TypeCast::convertGenericCuda<int16_t, int8_t>(extras, dx, N, dz);
            } else if (dstType == ND4J_UINT8) {
                sd::TypeCast::convertGenericCuda<int16_t, uint8_t>(extras, dx, N, dz);
            } else if (dstType == ND4J_FLOAT16) {
                sd::TypeCast::convertGenericCuda<int16_t, float16>(extras, dx, N, dz);
            } else if (dstType == ND4J_INT16) {
                sd::TypeCast::convertGenericCuda<int16_t, int16_t>(extras, dx, N, dz);
            } else if (dstType == ND4J_UINT16) {
                sd::TypeCast::convertGenericCuda<int16_t, uint16_t>(extras, dx, N, dz);
            } else if (dstType == ND4J_FLOAT24) {
                // TODO...
            } else if (dstType == ND4J_FLOAT32) {
                sd::TypeCast::convertGenericCuda<int16_t, float>(extras, dx, N, dz);
            } else if (dstType == ND4J_DOUBLE) {
                sd::TypeCast::convertGenericCuda<int16_t, double>(extras, dx, N, dz);
            } else {
                printf("Unsupported types conversion: [%i] -> [%i]\n", srcType, dstType);
            }
        } else if (srcType == ND4J_FLOAT24) {

        } else if (srcType == ND4J_FLOAT32) {
            if (dstType == ND4J_FLOAT8) {
                //sd::TypeCast::convertGenericCuda<float, sd::float8>(extras, dx, N, dz);
            } else if (dstType == ND4J_INT8) {
                sd::TypeCast::convertGenericCuda<float, int8_t>(extras, dx, N, dz);
            } else if (dstType == ND4J_UINT8) {
                sd::TypeCast::convertGenericCuda<float, uint8_t>(extras, dx, N, dz);
            } else if (dstType == ND4J_FLOAT16) {
                sd::TypeCast::convertGenericCuda<float, float16>(extras, dx, N, dz);
            } else if (dstType == ND4J_INT16) {
                sd::TypeCast::convertGenericCuda<float, int16_t>(extras, dx, N, dz);
            } else if (dstType == ND4J_UINT16) {
                sd::TypeCast::convertGenericCuda<float, uint16_t>(extras, dx, N, dz);
            } else if (dstType == ND4J_FLOAT24) {

            } else if (dstType == ND4J_DOUBLE) {
                sd::TypeCast::convertGenericCuda<float, double>(extras, dx, N, dz);
            } else if (dstType == ND4J_THRESHOLD) {
                //sd::convertToThreshold<float>(nullptr, dx, N, dz);
            } else {
                nd4j_printf("Unsupported types conversion: [%i] -> [%i]\n", srcType, dstType);
            }
        } else if (srcType == ND4J_DOUBLE) {
            if (dstType == ND4J_FLOAT8) {
                //sd::TypeCast::convertGenericCuda<double, sd::float8>(extras, dx, N, dz);
            } else if (dstType == ND4J_INT8) {
                sd::TypeCast::convertGenericCuda<double, int8_t>(extras, dx, N, dz);
            } else if (dstType == ND4J_UINT8) {
                sd::TypeCast::convertGenericCuda<double, uint8_t>(extras, dx, N, dz);
            } else if (dstType == ND4J_FLOAT16) {
                sd::TypeCast::convertGenericCuda<double, float16>(extras, dx, N, dz);
            } else if (dstType == ND4J_INT16) {
                sd::TypeCast::convertGenericCuda<double, int16_t>(extras, dx, N, dz);
            } else if (dstType == ND4J_UINT16) {
                sd::TypeCast::convertGenericCuda<double, uint16_t>(extras, dx, N, dz);
            } else if (dstType == ND4J_FLOAT24) {

            } else if (dstType == ND4J_FLOAT32) {
                sd::TypeCast::convertGenericCuda<double, float>(extras, dx, N, dz);
            } else if (dstType == ND4J_DOUBLE) {
                //
            } else if (dstType == ND4J_THRESHOLD) {
                //sd::convertToThreshold<double>(nullptr, dx, N, dz);
            } else {
                nd4j_printf("Unsupported types conversion: [%i] -> [%i]\n", srcType, dstType);
            }
        } else if (srcType == ND4J_THRESHOLD) {
            if (dstType == ND4J_FLOAT16) {
                //sd::convertFromThreshold<float16>(nullptr, dx, N, dz);
            } else if (dstType == ND4J_FLOAT32) {
                //sd::convertFromThreshold<float>(nullptr, dx, N, dz);
            } else if (dstType == ND4J_DOUBLE) {
                //sd::convertFromThreshold<double>(nullptr, dx, N, dz);
            } else {
                nd4j_printf("Unsupported types conversion: [%i] -> [%i]\n", srcType, dstType);
            }
        } else {
            nd4j_printf("Unsupported types conversion: [%i] -> [%i]\n", srcType, dstType);
        }
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

Nd4jPointer createUtf8String(Nd4jPointer *extraPointers, const char *string, int length) {
    auto u = new sd::utf8string(string, length);
    return reinterpret_cast<Nd4jPointer>(u);
}

Nd4jLong getUtf8StringLength(Nd4jPointer *extraPointers, Nd4jPointer ptr) {
    return reinterpret_cast<sd::utf8string*>(ptr)->_length;
}
char* getUtf8StringBuffer(Nd4jPointer *extraPointers, Nd4jPointer ptr) {
    return reinterpret_cast<sd::utf8string*>(ptr)->_buffer;
}

void deleteUtf8String(Nd4jPointer *extraPointers, Nd4jPointer ptr) {
    delete(reinterpret_cast<sd::utf8string*>(ptr));
}

///////////////////////////////////////////////////////////////////
template<typename T, typename I>
__global__ static void scatterUpdateCuda(const int opCode, const int numOfSubArrs,
										      void* vx, const Nd4jLong *xShapeInfo, const Nd4jLong *xOffsets,
										      void* vy, const Nd4jLong *yShapeInfo, const Nd4jLong *yOffsets,
										      const void* vindexes) {

    __shared__ T *x, *y;
    __shared__ Nd4jLong arrLenX, arrLenY;
    auto indexes = reinterpret_cast<const I*>(vindexes);

    for (int e = 0; e < numOfSubArrs; e++ ) {

        const auto xIndex = indexes[e];
        const bool isOwner = xIndex < gridDim.x ? blockIdx.x == xIndex : blockIdx.x == xIndex % gridDim.x;

        if (!isOwner)
            continue;

        if (threadIdx.x == 0) {
            x = reinterpret_cast<T*>(vx) + xOffsets[xIndex];
            y = reinterpret_cast<T*>(vy) + yOffsets[e];
            arrLenX = shape::length(xShapeInfo);
            arrLenY = shape::length(yShapeInfo);
        }
        __syncthreads();

        if (arrLenX != arrLenY)
            return;

        for (Nd4jLong i = threadIdx.x; i < arrLenX; i += blockDim.x) {

            const auto xOffset = shape::getIndexOffset(i, xShapeInfo);
            const auto yOffset = shape::getIndexOffset(i, yShapeInfo);

            switch (opCode) {
                case 0:
                    x[xOffset] += y[yOffset];
                    break;
                case 1:
                    x[xOffset] -= y[yOffset];
                    break;
                case 2:
                    x[xOffset] *= y[yOffset];
                    break;
                case 3:
                    x[xOffset] /= y[yOffset];
                    break;
                case 4:
                    x[xOffset] = y[yOffset] - x[xOffset];
                    break;
                case 5:
                    x[xOffset] = y[yOffset] / x[xOffset];
                    break;
                case 6:
                    x[xOffset] = y[yOffset];
                    break;
                default:
                    continue;
            }
        }
        __syncthreads();
    }
}

template<typename T, typename I>
__host__ static void scatterUpdateCudaLauncher(const cudaStream_t* stream, const int opCode, const int numOfSubArrs, void* vx, const Nd4jLong *xShapeInfo, const Nd4jLong *xOffsets, void* vy, const Nd4jLong *yShapeInfo, const Nd4jLong *yOffsets, const void* indexes) {

    scatterUpdateCuda<T, I><<<512, 256, MAX_NUM_THREADS, *stream>>>(opCode, numOfSubArrs, vx, xShapeInfo, xOffsets, vy, yShapeInfo, yOffsets, indexes);
}


//////////////////////////////////////////////////////////////////////////
void scatterUpdate(Nd4jPointer *extraPointers, int opCode, int numOfSubArrs,
                      			void* hX, Nd4jLong* hXShapeInfo, Nd4jLong* hXOffsets,
                      			void* dX, Nd4jLong* dXShapeInfo, Nd4jLong* dXOffsets,
                      			void* hY, Nd4jLong* hYShapeInfo, Nd4jLong* hYOffsets,
                      			void* dY, Nd4jLong* dYShapeInfo, Nd4jLong* dYOffsets,
                      			void* hIindexes, Nd4jLong* hIndicesShapeInfo, void* dIindexes, Nd4jLong* dIndicesShapeInfo) {
    try {
        auto stream = reinterpret_cast<cudaStream_t *>(extraPointers[1]);

        auto type = ArrayOptions::dataType(hXShapeInfo);
        auto iType = ArrayOptions::dataType(hIndicesShapeInfo);

        BUILD_DOUBLE_SELECTOR(type, iType, scatterUpdateCudaLauncher,
                              (stream, opCode, numOfSubArrs, dX, dXShapeInfo, dXOffsets, dY, dYShapeInfo, dYOffsets, dIindexes),
                              LIBND4J_TYPES, INDEXING_TYPES);

        sd::DebugHelper::checkErrorCode(stream, "scatterUpdate(...) failed");
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

void inspectArray(Nd4jPointer *extraPointers, Nd4jPointer buffer, Nd4jLong *shapeInfo, Nd4jPointer specialBuffer, Nd4jLong *specialShapeInfo, Nd4jPointer debugInfo) {
    try {
        LaunchContext lc(extraPointers[1], extraPointers[4], extraPointers[5], extraPointers[3]);
        auto p = reinterpret_cast<sd::DebugInfo *>(debugInfo);
        NDArray array(buffer, specialBuffer, shapeInfo, &lc);
        sd::DebugHelper::retrieveDebugStatistics(p, &array);
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

void __global__ tryPointerKernel(void* p, int len) {
    auto buf = reinterpret_cast<int8_t*>(p);
    auto tid = threadIdx.x + blockIdx.x * blockDim.x;
    __shared__ int b;
    if (tid < len)
        atomicAdd(&b, buf[tid]);

    __syncthreads();

    if (threadIdx.x ==0 && blockIdx.x == 0)
        printf("Pointer check complete: %i\n", b);
}

void tryPointer(Nd4jPointer extra, Nd4jPointer p, int len) {
    try {
        cudaStream_t stream;
        cudaStreamCreate(&stream);

        tryPointerKernel << < 256, 512, len + 64, stream >> > (p, len);
        auto e = cudaStreamSynchronize(stream);

        if (e != 0)
            throw sd::cuda_exception::build("tryPointer failed", e);

        cudaStreamDestroy(stream);
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

int dataTypeFromNpyHeader(void *header) {
    return (int) cnpy::dataTypeFromHeader(reinterpret_cast<char *>(header));
}
sd::ConstantDataBuffer* shapeBuffer(int rank, Nd4jLong *shape, Nd4jLong *strides, sd::DataType dtype, char order, Nd4jLong ews, bool empty) {
    try {
        auto buffer = new ConstantDataBuffer();
        *buffer = sd::ConstantShapeHelper::getInstance()->bufferForShapeInfo(
                ShapeDescriptor(dtype, order, shape, strides, rank, ews, empty));
        return buffer;
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
        return nullptr;
    }
}

void deleteShapeBuffer(sd::ConstantDataBuffer* ptr) {
    delete ptr;
}

void deleteTadPack(sd::TadPack* ptr) {
    delete ptr;
}

bool isBlasVersionMatches(int major, int minor, int build) {
    auto result = major == Environment::getInstance()->_blasMajorVersion && minor == Environment::getInstance()->_blasMinorVersion && build == Environment::getInstance()->_blasPatchVersion;

    if (!result) {
        nd4j_printf("CUDA/cuBLAS version mismatch. Expected: %i.%i.%i but got %i.%i.%i instead\n", Environment::getInstance()->_blasMajorVersion, Environment::getInstance()->_blasMinorVersion, Environment::getInstance()->_blasPatchVersion, major, minor, build);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(152);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage("CUDA/cuBLAS version mismatch");
    }

    return result;
}

sd::ConstantDataBuffer* constantBufferLong(sd::DataType dtype, Nd4jLong *data, int length) {
    return sd::ConstantHelper::getInstance()->constantBuffer(ConstantDescriptor(data, length), dtype);
}

sd::ConstantDataBuffer* constantBufferDouble(sd::DataType dtype, double *data, int length) {
    return sd::ConstantHelper::getInstance()->constantBuffer(ConstantDescriptor(data, length), dtype);
}

sd::ConstantDataBuffer* constantBuffer(sd::DataType dtype, sd::ConstantDescriptor *descriptor) {
    return sd::ConstantHelper::getInstance()->constantBuffer(*descriptor, dtype);
}


Nd4jPointer getConstantDataBufferPrimary(sd::ConstantDataBuffer* dbf) {
    return dbf->primary();
}
Nd4jPointer getConstantDataBufferSpecial(sd::ConstantDataBuffer* dbf) {
    return dbf->special();
}
Nd4jLong getConstantDataBufferLength(sd::ConstantDataBuffer* dbf) {
    return dbf->length();
}
Nd4jLong getConstantDataBufferSizeOf(sd::ConstantDataBuffer* dbf) {
    return dbf->sizeOf();
}


sd::graph::Context* createGraphContext(int nodeId) {
    return new sd::graph::Context(nodeId);
}

sd::graph::RandomGenerator* getGraphContextRandomGenerator(sd::graph::Context* ptr) {
    return &ptr->randomGenerator();
}

void markGraphContextInplace(sd::graph::Context* ptr, bool reallyInplace) {
    ptr->markInplace(reallyInplace);
}

void setGraphContextCudaContext(sd::graph::Context* ptr, void *stream, void *reductionPointer, void *allocationPointer) {
    ptr->setCudaContext(stream, reductionPointer, allocationPointer);
}

void setGraphContextInputArray(sd::graph::Context* ptr, int index, void *buffer, void *shapeInfo, void *specialBuffer, void *specialShapeInfo) {
    ptr->setInputArray(index, buffer, shapeInfo, specialBuffer, specialShapeInfo);
}

void setGraphContextOutputArray(sd::graph::Context* ptr, int index, void *buffer, void *shapeInfo, void *specialBuffer, void *specialShapeInfo) {
    ptr->setOutputArray(index, buffer, shapeInfo, specialBuffer, specialShapeInfo);
}

void setGraphContextInputBuffer(OpaqueContext* ptr, int index, OpaqueDataBuffer *buffer, void *shapeInfo, void *specialShapeInfo) {
    ptr->setInputArray(index, buffer, shapeInfo, specialShapeInfo);
}

void setGraphContextOutputBuffer(OpaqueContext* ptr, int index, OpaqueDataBuffer *buffer, void *shapeInfo, void *specialShapeInfo) {
    ptr->setOutputArray(index, buffer, shapeInfo, specialShapeInfo);
}

void setGraphContextTArguments(sd::graph::Context* ptr, double *arguments, int numberOfArguments) {
    ptr->setTArguments(arguments, numberOfArguments);
}

void setGraphContextIArguments(sd::graph::Context* ptr, Nd4jLong *arguments, int numberOfArguments) {
    ptr->setIArguments(arguments, numberOfArguments);
}

void setGraphContextBArguments(sd::graph::Context* ptr, bool *arguments, int numberOfArguments) {
    ptr->setBArguments(arguments, numberOfArguments);
}

void setGraphContextDArguments(OpaqueContext* ptr, int *arguments, int numberOfArguments) {
    std::vector<sd::DataType> dtypes(numberOfArguments);
    for (int e = 0; e < numberOfArguments; e++)
        dtypes[e] = (sd::DataType) arguments[e];

    ptr->setDArguments(dtypes);
}

void deleteGraphContext(sd::graph::Context* ptr) {
    delete ptr;
}


sd::graph::RandomGenerator* createRandomGenerator(Nd4jLong rootSeed, Nd4jLong nodeSeed) {
    try {
        return new sd::graph::RandomGenerator(rootSeed, nodeSeed);
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
        return nullptr;
    }
}

Nd4jLong getRandomGeneratorRootState(sd::graph::RandomGenerator* ptr) {
    return ptr->rootState();
}

Nd4jLong getRandomGeneratorNodeState(sd::graph::RandomGenerator* ptr) {
    return ptr->nodeState();
}

void setRandomGeneratorStates(sd::graph::RandomGenerator* ptr, Nd4jLong rootSeed, Nd4jLong nodeSeed) {
    ptr->setStates(rootSeed, nodeSeed);
}

int getRandomGeneratorRelativeInt(sd::graph::RandomGenerator* ptr, Nd4jLong index) {
    return ptr->relativeInt(index);
}

Nd4jLong getRandomGeneratorRelativeLong(sd::graph::RandomGenerator* ptr, Nd4jLong index) {
    return ptr->relativeLong(index);
}

void deleteRandomGenerator(sd::graph::RandomGenerator* ptr) {
    delete ptr;
}


Nd4jPointer shapeBufferForNumpy(Nd4jPointer npyArray) {
    try {
        cnpy::NpyArray arr = cnpy::loadNpyFromPointer(reinterpret_cast<char *>(npyArray));
        unsigned int shapeSize = arr.shape.size();
        std::vector<Nd4jLong> shape(shapeSize);
        bool _empty = false;
        for (unsigned int i = 0; i < shapeSize; i++) {
            shape[i] = arr.shape[i];

            if (arr.shape[i] == 0)
                _empty = true;
        }

        auto dtype = cnpy::dataTypeFromHeader(reinterpret_cast<char *>(npyArray));

        Nd4jLong *shapeBuffer;
        if (shape.size() == 1 && shape[0] == 0) {
            // scalar case
            shapeBuffer = sd::ShapeBuilders::createScalarShapeInfo(dtype);
        } else if (_empty) {
            if (shapeSize > 0)
                shapeBuffer = sd::ShapeBuilders::emptyShapeInfo(dtype, arr.fortranOrder ? 'f' : 'c', shape);
            else
                shapeBuffer = sd::ShapeBuilders::emptyShapeInfo(dtype);
        } else {
            shapeBuffer = sd::ShapeBuilders::createShapeInfo(dtype, arr.fortranOrder ? 'f' : 'c', shape);
        }
        return reinterpret_cast<Nd4jPointer>(sd::ConstantShapeHelper::getInstance()->createFromExisting(shapeBuffer,
                                                                                                          true));
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
        return nullptr;
    }
}

const char* runLightBenchmarkSuit(bool printOut) {
    try {
        sd::LightBenchmarkSuit suit;
        auto result = suit.runSuit();

        if (printOut)
            nd4j_printf("%s\n", result.data());

        auto chars = new char[result.length() + 1];
        std::memcpy(chars, result.data(), result.length());
        chars[result.length()] = (char) 0x0;

        return chars;
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
        return nullptr;
    }
}

const char* runFullBenchmarkSuit(bool printOut) {
    try {
        sd::FullBenchmarkSuit suit;
        auto result = suit.runSuit();

        if (printOut)
            nd4j_printf("%s\n", result.data());

        auto chars = new char[result.length() + 1];
        std::memcpy(chars, result.data(), result.length());
        chars[result.length()] = (char) 0x0;

        return chars;
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
        return nullptr;
    }
}

Nd4jLong getCachedMemory(int deviceId) {
    return sd::ConstantHelper::getInstance()->getCachedAmount(deviceId);
}

sd::LaunchContext* defaultLaunchContext() {
    return LaunchContext::defaultContext();
}

Nd4jPointer lcScalarPointer(OpaqueLaunchContext* lc) {
    return lc->getScalarPointer();
}

Nd4jPointer lcReductionPointer(OpaqueLaunchContext* lc) {
    return lc->getReductionPointer();
}

Nd4jPointer lcAllocationPointer(OpaqueLaunchContext* lc) {
    return lc->getAllocationPointer();
}

Nd4jPointer lcExecutionStream(OpaqueLaunchContext* lc) {
    return lc->getCudaStream();
}

Nd4jPointer lcCopyStream(OpaqueLaunchContext* lc) {
    return lc->getCudaSpecialStream();
}

Nd4jPointer lcBlasHandle(OpaqueLaunchContext* lc) {
    return lc->getCublasHandle();
}

Nd4jPointer lcSolverHandle(OpaqueLaunchContext* lc) {
    return lc->getCusolverHandle();
}

int lastErrorCode() {
    return sd::LaunchContext::defaultContext()->errorReference()->errorCode();
}

const char* lastErrorMessage() {
    return sd::LaunchContext::defaultContext()->errorReference()->errorMessage();
}

void ctxShapeFunctionOverride(OpaqueContext* ptr, bool reallyOverride) {
    ptr->setShapeFunctionOverride(reallyOverride);
}

void ctxPurge(OpaqueContext* ptr) {
    ptr->clearFastPath();
}

int  binaryLevel() {
    return 0;
}

int optimalLevel() {
    return 0;
}

bool isMinimalRequirementsMet() {
    return true;
}

bool isOptimalRequirementsMet() {
    return true;
}

void ctxAllowHelpers(OpaqueContext* ptr, bool reallyAllow) {
    ptr->allowHelpers(reallyAllow);
}

void ctxSetExecutionMode(OpaqueContext* ptr, int execMode) {
    if (execMode < 0 || execMode > 2)
        execMode = 0;

    ptr->setExecutionMode((samediff::ExecutionMode) execMode);
}

OpaqueDataBuffer* allocateDataBuffer(Nd4jLong elements, int dataType, bool allocateBoth) {
    try {
        auto dtype = DataTypeUtils::fromInt(dataType);
        return new sd::InteropDataBuffer(elements * DataTypeUtils::sizeOf(dtype), dtype, allocateBoth);
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
        return nullptr;
    }
}

Nd4jPointer dbPrimaryBuffer(OpaqueDataBuffer *dataBuffer) {
    return dataBuffer->primary();
}

Nd4jPointer dbSpecialBuffer(OpaqueDataBuffer *dataBuffer) {
    return dataBuffer->special();
}

void deleteDataBuffer(OpaqueDataBuffer *dataBuffer) {
    delete dataBuffer;
}

void dbSetPrimaryBuffer(OpaqueDataBuffer *dataBuffer, Nd4jPointer primaryBuffer, Nd4jLong numBytes) {
    dataBuffer->setPrimary(primaryBuffer, numBytes);
}

void dbSetSpecialBuffer(OpaqueDataBuffer *dataBuffer, Nd4jPointer specialBuffer, Nd4jLong numBytes) {
    dataBuffer->setSpecial(specialBuffer, numBytes);
}

void dbAllocatePrimaryBuffer(OpaqueDataBuffer *dataBuffer) {
    dataBuffer->dataBuffer()->allocatePrimary();
}

void dbAllocateSpecialBuffer(OpaqueDataBuffer *dataBuffer) {
    dataBuffer->dataBuffer()->allocateSpecial();
}

void dbExpandBuffer(OpaqueDataBuffer *dataBuffer, Nd4jLong elements) {
    try {
        dataBuffer->dataBuffer()->expand(elements * DataTypeUtils::sizeOf(dataBuffer->dataBuffer()->getDataType()));
    } catch (std::exception &e) {
        sd::LaunchContext::defaultContext()->errorReference()->setErrorCode(1);
        sd::LaunchContext::defaultContext()->errorReference()->setErrorMessage(e.what());
    }
}

OpaqueDataBuffer* dbCreateView(OpaqueDataBuffer *dataBuffer, Nd4jLong length, Nd4jLong offset) {
    return new InteropDataBuffer(*dataBuffer, length, offset);
}

void dbSyncToSpecial(OpaqueDataBuffer *dataBuffer) {
    dataBuffer->dataBuffer()->syncToSpecial();
}

void dbSyncToPrimary(OpaqueDataBuffer *dataBuffer) {
    dataBuffer->dataBuffer()->syncToPrimary(nullptr);
}

void dbTickHostRead(OpaqueDataBuffer *dataBuffer) {
    dataBuffer->dataBuffer()->readPrimary();
}

void dbTickHostWrite(OpaqueDataBuffer *dataBuffer) {
    dataBuffer->dataBuffer()->writePrimary();
}

void dbTickDeviceRead(OpaqueDataBuffer *dataBuffer) {
    dataBuffer->dataBuffer()->readSpecial();
}

void dbTickDeviceWrite(OpaqueDataBuffer *dataBuffer) {
    dataBuffer->dataBuffer()->writeSpecial();
}

void dbExpand(OpaqueDataBuffer *dataBuffer, Nd4jLong elements) {
    dataBuffer->expand(elements);
}

void dbClose(OpaqueDataBuffer *dataBuffer) {
    dataBuffer->getDataBuffer()->close();
}

int dbDeviceId(OpaqueDataBuffer *dataBuffer) {
    return dataBuffer->deviceId();
}

void dbSetDeviceId(OpaqueDataBuffer *dataBuffer, int deviceId) {
    dataBuffer->setDeviceId(deviceId);
}

int dbLocality(OpaqueDataBuffer *dataBuffer) {
    auto p = dataBuffer->dataBuffer()->isPrimaryActual();
    auto d = dataBuffer->dataBuffer()->isSpecialActual();

    if (p && d)
        return 0;
    else if (p)
        return -1;
    else
        return 1;
}