Porting to MIOpen

Porting to MIOpen#

The following is a summary of the key differences between MIOpen and cuDNN.

  • Calling miopenFindConvolution*Algorithm() is mandatory before calling any Convolution API

  • The typical calling sequence for MIOpen Convolution APIs is:

    • miopenConvolution*GetWorkSpaceSize() (returns the workspace size required by Find())

    • miopenFindConvolution*Algorithm() (returns performance information for various algorithms)

    • miopenConvolution*()

MIOpen supports:

  • 4D tensors in the NCHW and NHWC storage format; the cuDNN __“\*Nd\*”__ APIs don’t have a corresponding MIOpen API

  • __`float(fp32)`__ datatype

  • __2D Convolutions__ and __3D Convolutions__

  • __2D Pooling__

MIOpen doesn’t support:

  • __Preferences__ for convolutions

  • Softmax modes (MIOpen implements the __SOFTMAX_MODE_CHANNEL__ flavor)

  • __Transform-Tensor__, __Dropout__, __RNNs__, and __Divisive Normalization__

Useful MIOpen environment variables include:

  • MIOPEN_ENABLE_LOGGING=1: Logs all the MIOpen APIs called, including the parameters passed to those APIs

  • MIOPEN_DEBUG_AMD_ROCM_PRECOMPILED_BINARIES=0: Disables the Winograd convolution algorithm

  • MIOPEN_DEBUG_GCN_ASM_KERNELS=0: Disables hand-tuned ASM kernels for the direct convolution algorithm (the fall-back is to kernels written in high-level language)

  • MIOPEN_DEBUG_CONV_FFT=0: Disables the FFT convolution algorithm

  • MIOPEN_DEBUG_CONV_DIRECT=0: Disables the direct convolution algorithm

cuDNN versus MIOpen APIs#

The following sections compare cuDNN and MIOpen APIs with similar functions.

Handle operations#

cuDNN

MIOpen

 
cudnnStatus_t
cudnnCreate(
    cudnnHandle_t *handle)

miopenStatus_t
miopenCreate(
    miopenHandle_t *handle)

cudnnStatus_t
cudnnDestroy(
    cudnnHandle_t handle)

miopenStatus_t
miopenDestroy(
    miopenHandle_t handle)

cudnnStatus_t
cudnnSetStream(
    cudnnHandle_t handle,
    cudaStream_t streamId)

miopenStatus_t
miopenSetStream(
    miopenHandle_t handle,
    miopenAcceleratorQueue_t streamID)

cudnnStatus_t
cudnnGetStream(
    cudnnHandle_t handle,
    cudaStream_t *streamId)

miopenStatus_t
miopenGetStream(
    miopenHandle_t handle,
    miopenAcceleratorQueue_t  *streamID)

Tensor operations#

cuDNN

MIOpen

 
cudnnStatus_t
cudnnCreateTensorDescriptor(
    cudnnTensorDescriptor_t *tensorDesc)

miopenStatus_t
miopenCreateTensorDescriptor(
    miopenTensorDescriptor_t
    *tensorDesc)

cudnnStatus_t
cudnnDestroyTensorDescriptor(
    cudnnTensorDescriptor_t tensorDesc)

miopenStatus_t
miopenDestroyTensorDescriptor(
    miopenTensorDescriptor_t tensorDesc)

cudnnStatus_t
cudnnSetTensor(
    cudnnHandle_t handle,
    const cudnnTensorDescriptor_t yDesc,
    void *y,
    const void *valuePtr)

miopenStatus_t
miopenSetTensor(
    miopenHandle_t handle,
    const miopenTensorDescriptor_t yDesc,
    void *y,
    const void *alpha)

cudnnStatus_t
cudnnSetTensor4dDescriptor(
    cudnnTensorDescriptor_t tensorDesc,
    cudnnTensorFormat_t format,
    cudnnDataType_t dataType,
    int n,
    int c,
    int h,
    int w)

    miopenStatus_t miopenSet4dTensorDescriptor(
        miopenTensorDescriptor_t tensorDesc,
        miopenDataType_t dataType,
        int n,
        int c,
        int h,
        int w)

(Only ``NCHW`` format is supported)

cudnnStatus_t
cudnnGetTensor4dDescriptor(
    cudnnTensorDescriptor_t tensorDesc,
    cudnnDataType_t *dataType,
    int *n,
    int *c,
    int *h,
    int *w,
    int *nStride,
    int *cStride,
    int *hStride,
    int *wStride)

miopenStatus_t
miopenGet4dTensorDescriptor(
    miopenTensorDescriptor_t tensorDesc,
    miopenDataType_t *dataType,
    int *n,
    int *c,
    int *h,
    int *w,
    int *nStride,
    int *cStride,
    int *hStride,
    int *wStride)

cudnnStatus_t
cudnnAddTensor(
    cudnnHandle_t handle,
    const void *alpha,
    const cudnnTensorDescriptor_t aDesc,
    const void *A,
    const void *beta,
    const cudnnTensorDescriptor_t cDesc,
    void *C)

    miopenStatus_t
    miopenOpTensor(
        miopenHandle_t handle,
        miopenTensorOp_t tensorOp,
        const void *alpha1,
        constmiopenTensorDescriptor_t  aDesc,
        const void *A,
        const void *alpha2,
        const miopenTensorDescriptor_t bDesc,
        const void *B,
        const void *beta,
        const miopenTensorDescriptor_t  cDesc,
        void *C)

For forward bias, use ``miopenConvolutionForwardBias``.

cudnnStatus_t
cudnnOpTensor(
    cudnnHandle_t handle,
    const cudnnOpTensorDescriptor_t opTensorDesc,
    const void *alpha1,
    const cudnnTensorDescriptor_t aDesc,
    const void *A,
    const void *alpha2,
    const cudnnTensorDescriptor_t bDesc,
    const void *B,
    const void *beta,
    const cudnnTensorDescriptor_t cDesc,
    void *C)

miopenStatus_t
miopenOpTensor(
    miopenHandle_t handle,
    miopenTensorOp_t tensorOp,
    const void *alpha1,
    const miopenTensorDescriptor_t aDesc,
    const void *A, const void *alpha2,
    const miopenTensorDescriptor_t  bDesc,
    const void *B,
    const void *beta,
    const miopenTensorDescriptor_t  cDesc,
    void *C)

cudnnStatus_t
cudnnOpTensor(
    cudnnHandle_t handle,
    const cudnnOpTensorDescriptor_t opTensorDesc,
    const void *alpha1,
    const cudnnTensorDescriptor_t aDesc,
    const void *A,
    const void *alpha2,
    const cudnnTensorDescriptor_t bDesc,
    const void *B,
    const void *beta,
    const cudnnTensorDescriptor_t cDesc,
    void *C)

miopenStatus_t
miopenOpTensor(
    miopenHandle_t handle,
    miopenTensorOp_t tensorOp,
    const void *alpha1,
    const miopenTensorDescriptor_t aDesc,
    const void *A, const void *alpha2,
    const miopenTensorDescriptor_t  bDesc,
    const void *B,
    const void *beta,
    const miopenTensorDescriptor_t  cDesc,
    void *C)

cudnnStatus_t
cudnnScaleTensor(
    cudnnHandle_t handle,
    const cudnnTensorDescriptor_t yDesc,
    void *y,
    const void *alpha)

miopenStatus_t
miopenScaleTensor(
    miopenHandle_t handle,
    const miopenTensorDescriptor_t yDesc,
    void *y,
    const void *alpha)

Filter operations#

cuDNN

MIOpen

 
cudnnStatus_t
cudnnCreateFilterDescriptor(
    cudnnFilterDescriptor_t *filterDesc)

All FilterDescriptor APIs are substituted by their respective TensorDescriptor API.

Convolution operations#

cuDNN

MIOpen

 
cudnnStatus_t
cudnnCreateConvolutionDescriptor(
    cudnnConvolutionDescriptor_t *convDesc)

miopenStatus_t
miopenCreateConvolutionDescriptor(
    miopenConvolutionDescriptor_t *convDesc)

cudnnStatus_t
cudnnDestroyConvolutionDescriptor(
    cudnnConvolutionDescriptor_t convDesc)

miopenStatus_t
miopenDestroyConvolutionDescriptor(
    miopenConvolutionDescriptor_t convDesc)

cudnnStatus_t
cudnnGetConvolution2dDescriptor(
    const cudnnConvolutionDescriptor_t convDesc,
    int *pad_h,
    int *pad_y,
    int *u,
    int *v,
    int *upscalex,
    int *upscaley,
    cudnnConvolutionMode_t *mode)

miopenStatus_t
miopenGetConvolutionDescriptor(
    miopenConvolutionDescriptor_t convDesc,
    miopenConvolutionMode_t *mode,
    int *pad_h,
    int *pad_y,
    int *u,
    int *v,
    int *upscalex,
    int *upscaley)

cudnnStatus_t
cudnnGetConvolution2dForwardOutputDim(
    const cudnnConvolutionDescriptor_t convDesc,
    const cudnnTensorDescriptor_t inputTensorDesc,
    const cudnnFilterDescriptor_t filterDesc,
    int *n,
    int *c,
    int *h,
    int *w)

miopenStatus_t
miopenGetConvolutionForwardOutputDim(
    miopenConvolutionDescriptor_t convDesc,
    const miopenTensorDescriptor_t inputTensorDesc,
    const miopenTensorDescriptor_t filterDesc,
    int *n,
    int *c,
    int *h,
    int *w)

cudnnStatus_t
cudnnGetConvolutionForwardWorkspaceSize(
    cudnnHandle_t handle,
    const cudnnTensorDescriptor_t xDesc,
    const cudnnFilterDescriptor_t wDesc,
    const cudnnConvolutionDescriptor_t convDesc,
    const cudnnTensorDescriptor_t yDesc,
    cudnnConvolutionFwdAlgo_t algo,
    size_t *sizeInBytes)

miopenStatus_t
miopenConvolutionForwardGetWorkSpaceSize(
    miopenHandle_t handle,
    const miopenTensorDescriptor_t wDesc,
    const miopenTensorDescriptor_t xDesc,
    const miopenConvolutionDescriptor_t convDesc,
    const miopenTensorDescriptor_t yDesc,
    size_t *workSpaceSize)

cudnnStatus_t
cudnnGetConvolutionBackwardFilterWorkspaceSize(
    cudnnHandle_t handle,
    const cudnnTensorDescriptor_t xDesc,
    const cudnnTensorDescriptor_t dyDesc,
    const cudnnConvolutionDescriptor_t convDesc,
    const cudnnFilterDescriptor_t gradDesc,
    cudnnConvolutionBwdFilterAlgo_t algo,
    size_t *sizeInBytes)

miopenStatus_t
miopenConvolutionBackwardWeightsGetWorkSpaceSize(
    miopenHandle_t handle,
    const miopenTensorDescriptor_t dyDesc,
    const miopenTensorDescriptor_t xDesc,
    const miopenConvolutionDescriptor_t convDesc,
    const miopenTensorDescriptor_t dwDesc,
    size_t *workSpaceSize)

cudnnStatus_t
cudnnGetConvolutionBackwardDataWorkspaceSize(
    cudnnHandle_t handle,
    const cudnnFilterDescriptor_t wDesc,
    const cudnnTensorDescriptor_t dyDesc,
    const cudnnConvolutionDescriptor_t convDesc,
    const cudnnTensorDescriptor_t dxDesc,
    cudnnConvolutionBwdDataAlgo_t algo,
    size_t *sizeInBytes

miopenStatus_t
miopenConvolutionBackwardDataGetWorkSpaceSize(
    miopenHandle_t handle,
    const miopenTensorDescriptor_t dyDesc,
    const miopenTensorDescriptor_t wDesc,
    const miopenConvolutionDescriptor_t convDesc,
    const miopenTensorDescriptor_t dxDesc,
    size_t *workSpaceSize)

cudnnStatus_t
cudnnConvolutionForward(
    cudnnHandle_t handle,
    const void *alpha,
    const cudnnTensorDescriptor_t xDesc,
    const void *x,
    const cudnnFilterDescriptor_t wDesc,
    const void *w,
    const cudnnConvolutionDescriptor_t convDesc,
    cudnnConvolutionFwdAlgo_t algo,
    void *workSpace,
    size_t workSpaceSizeInBytes,
    const void *beta,
    const cudnnTensorDescriptor_t yDesc,
    void *y)

miopenStatus_t
miopenConvolutionForward(
    miopenHandle_t handle,
    const void *alpha,
    const miopenTensorDescriptor_t xDesc,
    const void *x,
    const miopenTensorDescriptor_t wDesc,
    const void *w,
    const miopenConvolutionDescriptor_t convDesc,
    miopenConvFwdAlgorithm_t algo,
    const void *beta,
    const miopenTensorDescriptor_t yDesc,
    void *y,
    void *workSpace,
    size_t workSpaceSize)

    cudnnStatus_t
    cudnnFindConvolutionForwardAlgorithm(
        cudnnHandle_t handle,
        const cudnnTensorDescriptor_t xDesc,
        const cudnnFilterDescriptor_t wDesc,
        const cudnnConvolutionDescriptor_t convDesc,
        const cudnnTensorDescriptor_t yDesc,
        const int requestedAlgoCount,
        int *returnedAlgoCount,
        cudnnConvolutionFwdAlgoPerf_t *perfResults)

.. code-block:: cpp

    cudnnStatus_t
    cudnnFindConvolutionForwardAlgorithmEx(
        cudnnHandle_t handle,
        const cudnnTensorDescriptor_t xDesc,
        const void *x,
        const cudnnFilterDescriptor_t wDesc,
        const void *w,
        const cudnnConvolutionDescriptor_t convDesc,
        const cudnnTensorDescriptor_t yDesc,
        void *y,
        const int requestedAlgoCount,
        int *returnedAlgoCount,
        cudnnConvolutionFwdAlgoPerf_t *perfResults,
        void *workSpace,
        size_t workSpaceSizeInBytes)

.. code-block:: cpp

    cudnnStatus_t
    cudnnGetConvolutionForwardAlgorithm(
        cudnnHandle_t handle,
        const cudnnTensorDescriptor_t xDesc,
        const cudnnFilterDescriptor_t wDesc,
        const cudnnConvolutionDescriptor_t convDesc,
        const cudnnTensorDescriptor_t yDesc,
        cudnnConvolutionFwdPreference_t preference,
        size_t memoryLimitInBytes,
        cudnnConvolutionFwdAlgo_t *algo)
FindConvolution() is mandatory.

Allocate workspace prior to running this API. A table with times and memory requirements for different algorithms is returned. You can choose the top-most algorithm if you want only the fastest algorithm.

miopenStatus_t
miopenFindConvolutionForwardAlgorithm(
    miopenHandle_t handle,
    const miopenTensorDescriptor_t xDesc,
    const void *x,
    const miopenTensorDescriptor_t wDesc,
    const void *w,
    const miopenConvolutionDescriptor_t convDesc,
    const miopenTensorDescriptor_t yDesc,
    void *y,
    const int requestAlgoCount,
    int *returnedAlgoCount,
    miopenConvAlgoPerf_t *perfResults,
    void *workSpace,
    size_t workSpaceSize,
    bool exhaustiveSearch)

cudnnStatus_t
cudnnConvolutionBackwardBias(
    cudnnHandle_t handle,
    const void *alpha,
    const cudnnTensorDescriptor_t dyDesc,
    const void *dy,
    const void *beta,
    const cudnnTensorDescriptor_t dbDesc,
    void *db)

miopenStatus_t
miopenConvolutionBackwardBias(
    miopenHandle_t handle,
    const void *alpha,
    const miopenTensorDescriptor_t dyDesc,
    const void *dy,
    const void *beta,
    const miopenTensorDescriptor_t dbDesc,
    void *db)

    cudnnStatus_t
    cudnnFindConvolutionBackwardFilterAlgorithm(
        cudnnHandle_t handle,
        const cudnnTensorDescriptor_t xDesc,
        const cudnnTensorDescriptor_t dyDesc,
        const cudnnConvolutionDescriptor_t convDesc,
        const cudnnFilterDescriptor_t dwDesc,
        const int requestedAlgoCount,
        int *returnedAlgoCount,
        cudnnConvolutionBwdFilterAlgoPerf_t *perfResults)

.. code-block:: cpp

    cudnnStatus_t
    cudnnFindConvolutionBackwardFilterAlgorithmEx(
        cudnnHandle_t handle,
        const cudnnTensorDescriptor_t xDesc,
        const void *x,
        const cudnnTensorDescriptor_t dyDesc,
        const void *y,
        const cudnnConvolutionDescriptor_t convDesc,
        const cudnnFilterDescriptor_t dwDesc,
        void *dw,
        const int requestedAlgoCount,
        int *returnedAlgoCount,
        cudnnConvolutionBwdFilterAlgoPerf_t *perfResults,
        void *workSpace,
        size_t workSpaceSizeInBytes)

.. code-block:: cpp

    cudnnStatus_t
    cudnnGetConvolutionBackwardFilterAlgorithm(
        cudnnHandle_t handle,
        const cudnnTensorDescriptor_t xDesc,
        const cudnnTensorDescriptor_t dyDesc,
        const cudnnConvolutionDescriptor_t convDesc,
        const cudnnFilterDescriptor_t dwDesc,
        cudnnConvolutionBwdFilterPreference_t preference,
        size_t memoryLimitInBytes,
        cudnnConvolutionBwdFilterAlgo_t *algo)
FindConvolution() is mandatory.

Allocate workspace prior to running this API. A table with times and memory requirements for different algorithms is returned. You can choose the top-most algorithm if you want only the fastest algorithm.

miopenStatus_t
miopenFindConvolutionBackwardWeightsAlgorithm(
    miopenHandle_t handle,
    const miopenTensorDescriptor_t dyDesc,
    const void *dy,
    const miopenTensorDescriptor_t xDesc,
    const void *x,
    const miopenConvolutionDescriptor_t convDesc,
    const miopenTensorDescriptor_t dwDesc,
    void *dw,
    const int requestAlgoCount,
    int *returnedAlgoCount,
    miopenConvAlgoPerf_t *perfResults,
    void *workSpace,
    size_t workSpaceSize,
    bool exhaustiveSearch)

    cudnnStatus_t
    cudnnFindConvolutionBackwardDataAlgorithm(
        cudnnHandle_t handle,
        const cudnnFilterDescriptor_t wDesc,
        const cudnnTensorDescriptor_t dyDesc,
        const cudnnConvolutionDescriptor_t convDesc,
        const cudnnTensorDescriptor_t dxDesc,
        const int requestedAlgoCount,
        int *returnedAlgoCount,
        cudnnConvolutionBwdDataAlgoPerf_t *perfResults)

.. code-block:: cpp

    cudnnStatus_t
    cudnnFindConvolutionBackwardDataAlgorithmEx(
        cudnnHandle_t handle,
        const cudnnFilterDescriptor_t wDesc,
        const void *w,
        const cudnnTensorDescriptor_t dyDesc,
        const void *dy,
        const cudnnConvolutionDescriptor_t convDesc,
        const cudnnTensorDescriptor_t dxDesc,
        void *dx,
        const int requestedAlgoCount,
        int *returnedAlgoCount,
        cudnnConvolutionBwdDataAlgoPerf_t *perfResults,
        void *workSpace,
        size_t workSpaceSizeInBytes)

.. code-block:: cpp

    cudnnStatus_t
    cudnnGetConvolutionBackwardDataAlgorithm(
        cudnnHandle_t handle,
        const cudnnFilterDescriptor_t wDesc,
        const cudnnTensorDescriptor_t dyDesc,
        const cudnnConvolutionDescriptor_t convDesc,
        const cudnnTensorDescriptor_t dxDesc,
        cudnnConvolutionBwdDataPreference_t preference,
        size_t memoryLimitInBytes,
        cudnnConvolutionBwdDataAlgo_t *algo)
FindConvolution() is mandatory.

Allocate workspace prior to running this API. A table with times and memory requirements for different algorithms is returned. You can choose the top-most algorithm if you want only the fastest algorithm.

miopenStatus_t
miopenFindConvolutionBackwardDataAlgorithm(
    miopenHandle_t handle,
    const miopenTensorDescriptor_t dyDesc,
    const void *dy,
    const miopenTensorDescriptor_t wDesc,
    const void *w,
    const miopenConvolutionDescriptor_t convDesc,
    const miopenTensorDescriptor_t dxDesc,
    const void *dx,
    const int requestAlgoCount,
    int *returnedAlgoCount,
    miopenConvAlgoPerf_t *perfResults,
    void *workSpace,
    size_t workSpaceSize,
    bool exhaustiveSearch)

cudnnStatus_t
cudnnConvolutionBackwardFilter(
    cudnnHandle_t handle,
    const void *alpha,
    const cudnnTensorDescriptor_t xDesc,
    const void *x,
    const cudnnTensorDescriptor_t dyDesc,
    const void *dy,
    const cudnnConvolutionDescriptor_t convDesc,
    cudnnConvolutionBwdFilterAlgo_t algo,
    void *workSpace,
    size_t workSpaceSizeInBytes,
    const void *beta,
    const cudnnFilterDescriptor_t dwDesc,
    void *dw)

miopenStatus_t
miopenConvolutionBackwardWeights(
    miopenHandle_t handle,
    const void *alpha,
    const miopenTensorDescriptor_t dyDesc,
    const void *dy,
    const miopenTensorDescriptor_t xDesc,
    const void *x,
    const miopenConvolutionDescriptor_t convDesc,
    miopenConvBwdWeightsAlgorithm_t algo,
    const void *beta,
    const miopenTensorDescriptor_t dwDesc,
    void *dw,
    void *workSpace,
    size_t workSpaceSize)

cudnnStatus_t
cudnnConvolutionBackwardData(
    cudnnHandle_t handle,
    const void *alpha,
    const cudnnFilterDescriptor_t wDesc,
    const void *w,
    const cudnnTensorDescriptor_t dyDesc,
    const void *dy,
    const cudnnConvolutionDescriptor_t convDesc,
    cudnnConvolutionBwdDataAlgo_t algo,
    void *workSpace,
    size_t workSpaceSizeInBytes,
    const void *beta,
    const cudnnTensorDescriptor_t dxDesc,
    void *dx)

miopenStatus_t
miopenConvolutionBackwardData(
    miopenHandle_t handle,
    const void *alpha,
    const miopenTensorDescriptor_t dyDesc,
    const void *dy,
    const miopenTensorDescriptor_t wDesc,
    const void *w,
    const miopenConvolutionDescriptor_t convDesc,
    miopenConvBwdDataAlgorithm_t algo,
    const void *beta,
    const miopenTensorDescriptor_t dxDesc,
    void *dx,
    void *workSpace,
    size_t workSpaceSize)

Softmax operations#

cuDNN

MIOpen

 
cudnnStatus_t
cudnnSoftmaxForward(
    cudnnHandle_t handle,
    cudnnSoftmaxAlgorithm_t algo,
    cudnnSoftmaxMode_t mode,
    const void *alpha,
    const cudnnTensorDescriptor_t xDesc,
    const void *x,
    const void *beta,
    const cudnnTensorDescriptor_t yDesc,
    void *y)

miopenStatus_t
miopenSoftmaxForward(
    miopenHandle_t handle,
    const void *alpha,
    const miopenTensorDescriptor_t xDesc,
    const void *x,
    const void *beta,
    const miopenTensorDescriptor_t yDesc,
    void *y)

cudnnStatus_t
cudnnSoftmaxBackward(
    cudnnHandle_t handle,
    cudnnSoftmaxAlgorithm_t algo,
    cudnnSoftmaxMode_t mode,
    const void *alpha,
    const cudnnTensorDescriptor_t yDesc,
    const void *y,
    const cudnnTensorDescriptor_t dyDesc,
    const void *dy,
    const void *beta,
    const cudnnTensorDescriptor_t dxDesc,
    void *dx)

miopenStatus_t
miopenSoftmaxBackward(
    miopenHandle_t handle,
    const void *alpha,
    const miopenTensorDescriptor_t yDesc,
    const void *y,
    const miopenTensorDescriptor_t dyDesc,
    const void *dy,
    const void *beta,
    const miopenTensorDescriptor_t dxDesc,
    void *dx)

Pooling operations#

cuDNN

MIOpen

 
cudnnStatus_t
cudnnCreatePoolingDescriptor(
    cudnnPoolingDescriptor_t *poolingDesc)

miopenStatus_t
miopenCreatePoolingDescriptor(
    miopenPoolingDescriptor_t *poolDesc)

cudnnStatus_t
cudnnSetPooling2dDescriptor(
    cudnnPoolingDescriptor_t poolingDesc,
    cudnnPoolingMode_t mode,
    cudnnNanPropagation_t maxpoolingNanOpt,
    int windowHeight,
    int windowWidth,
    int verticalPadding,
    int horizontalPadding,
    int verticalStride,
    int horizontalStride)

miopenStatus_t
miopenSet2dPoolingDescriptor(
    miopenPoolingDescriptor_t poolDesc,
    miopenPoolingMode_t mode,
    int windowHeight,
    int windowWidth,
    int pad_h,
    int pad_w,
    int u,
    int v)

cudnnStatus_t
cudnnGetPooling2dDescriptor(
    const cudnnPoolingDescriptor_t poolingDesc,
    cudnnPoolingMode_t *mode,
    cudnnNanPropagation_t *maxpoolingNanOpt,
    int *windowHeight,
    int *windowWidth,
    int *verticalPadding,
    int *horizontalPadding,
    int *verticalStride,
    int *horizontalStride)

miopenStatus_t
miopenGet2dPoolingDescriptor(
    const miopenPoolingDescriptor_t poolDesc,
    miopenPoolingMode_t *mode,
    int *windowHeight,
    int *windowWidth,
    int *pad_h,
    int *pad_w,
    int *u,
    int *v)

cudnnStatus_t
cudnnGetPooling2dForwardOutputDim(
    const cudnnPoolingDescriptor_t poolingDesc,
    const cudnnTensorDescriptor_t inputTensorDesc,
    int *n,
    int *c,
    int *h,
    int *w)

miopenStatus_t
miopenGetPoolingForwardOutputDim(
    const miopenPoolingDescriptor_t poolDesc,
    const miopenTensorDescriptor_t tensorDesc,
    int *n,
    int *c,
    int *h,
    int *w)

cudnnStatus_t
cudnnDestroyPoolingDescriptor(
    cudnnPoolingDescriptor_t poolingDesc)

miopenStatus_t
miopenDestroyPoolingDescriptor(
    miopenPoolingDescriptor_t poolDesc)

cudnnStatus_t
cudnnPoolingForward(
    cudnnHandle_t handle,
    const cudnnPoolingDescriptor_t poolingDesc,
    const void *alpha,
    const cudnnTensorDescriptor_t xDesc,
    const void *x,
    const void *beta,
    const cudnnTensorDescriptor_t yDesc,
    void *y)

miopenStatus_t
miopenPoolingForward(
    miopenHandle_t handle,
    const miopenPoolingDescriptor_t poolDesc,
    const void *alpha,
    const miopenTensorDescriptor_t xDesc,
    const void *x,
    const void *beta,
    const miopenTensorDescriptor_t yDesc,
    void *y,
    bool do_backward,
    void *workSpace,
    size_t workSpaceSize)

NA

 
miopenStatus_t
miopenPoolingGetWorkSpaceSize(
    const miopenTensorDescriptor_t yDesc,
    size_t *workSpaceSize)

cudnnStatus_t
cudnnPoolingBackward(
    cudnnHandle_t handle,
    const cudnnPoolingDescriptor_t poolingDesc,
    const void *alpha,
    const cudnnTensorDescriptor_t yDesc,
    const void *y,
    const cudnnTensorDescriptor_t dyDesc,
    const void *dy,
    const cudnnTensorDescriptor_t xDesc,
    const void *x,
    const void *beta,
    const cudnnTensorDescriptor_t dxDesc,
    void *dx)

miopenStatus_t
miopenPoolingBackward(
    miopenHandle_t handle,
    const miopenPoolingDescriptor_t poolDesc,
    const void *alpha,
    const miopenTensorDescriptor_t yDesc,
    const void *y,
    const miopenTensorDescriptor_t dyDesc,
    const void *dy,
    const miopenTensorDescriptor_t xDesc,
    const void *x,
    const void *beta,
    const miopenTensorDescriptor_t dxDesc,
    void *dx,
    const void *workspace)

Activation operations#

cuDNN

MIOpen

 
cudnnStatus_t
cudnnCreateActivationDescriptor(
    cudnnActivationDescriptor_t *activationDesc)

miopenStatus_t
miopenCreateActivationDescriptor(
    miopenActivationDescriptor_t *activDesc)

cudnnStatus_t
cudnnSetActivationDescriptor(
    cudnnActivationDescriptor_t activationDesc,
    cudnnActivationMode_t mode,
    cudnnNanPropagation_t reluNanOpt,
    double reluCeiling)

miopenStatus_t
miopenSetActivationDescriptor(
    const miopenActivationDescriptor_t activDesc,
    miopenActivationMode_t mode,
    double activAlpha,
    double activBeta,
    double activPower)

cudnnStatus_t
cudnnGetActivationDescriptor(
    const cudnnActivationDescriptor_t activationDesc,
    cudnnActivationMode_t *mode,
    cudnnNanPropagation_t *reluNanOpt,
    double *reluCeiling)

miopenStatus_t
miopenGetActivationDescriptor(
    const miopenActivationDescriptor_t activDesc,
    miopenActivationMode_t *mode,
    double *activAlpha,
    double *activBeta,
    double *activPower)

cudnnStatus_t
cudnnDestroyActivationDescriptor(
    cudnnActivationDescriptor_t activationDesc)

miopenStatus_t
miopenDestroyActivationDescriptor(
    miopenActivationDescriptor_t activDesc)

cudnnStatus_t
cudnnActivationForward(
    cudnnHandle_t handle,
    cudnnActivationDescriptor_t activationDesc,
    const void *alpha,
    const cudnnTensorDescriptor_t xDesc,
    const void *x,
    const void *beta,
    const cudnnTensorDescriptor_t yDesc,
    void *y)

miopenStatus_t
miopenActivationForward(
    miopenHandle_t handle,
    const miopenActivationDescriptor_t activDesc,
    const void *alpha,
    const miopenTensorDescriptor_t xDesc,
    const void *x,
    const void *beta,
    const miopenTensorDescriptor_t yDesc,
    void *y)

cudnnStatus_t
cudnnActivationBackward(
    cudnnHandle_t handle,
    cudnnActivationDescriptor_t activationDesc,
    const void *alpha,
    const cudnnTensorDescriptor_t yDesc,
    const void *y,
    const cudnnTensorDescriptor_t dyDesc,
    const void *dy,
    const cudnnTensorDescriptor_t xDesc,
    const void *x,
    const void *beta,
    const cudnnTensorDescriptor_t dxDesc,
    void *dx)

miopenStatus_t
miopenActivationBackward(
    miopenHandle_t handle,
    const miopenActivationDescriptor_t activDesc,
    const void *alpha,
    const miopenTensorDescriptor_t yDesc,
    const void *y,
    const miopenTensorDescriptor_t dyDesc,
    const void *dy,
    const miopenTensorDescriptor_t xDesc,
    const void *x,
    const void *beta,
    const miopenTensorDescriptor_t dxDesc,
    void *dx)

LRN operations#

cuDNN

MIOpen

 
cudnnStatus_t
cudnnCreateLRNDescriptor(
    cudnnLRNDescriptor_t *normDesc)

miopenStatus_t
miopenCreateLRNDescriptor(
    miopenLRNDescriptor_t
    *lrnDesc)

cudnnStatus_t
cudnnSetLRNDescriptor(
    cudnnLRNDescriptor_t normDesc,
    unsigned lrnN,
    double lrnAlpha,
    double lrnBeta,
    double lrnK)

miopenStatus_t
miopenSetLRNDescriptor(
    const miopenLRNDescriptor_t lrnDesc,
    miopenLRNMode_t mode,
    unsigned lrnN,
    double lrnAlpha,
    double lrnBeta,
    double lrnK)

cudnnStatus_t
cudnnGetLRNDescriptor(
    cudnnLRNDescriptor_t normDesc,
    unsigned* lrnN,
    double* lrnAlpha,
    double* lrnBeta,
    double* lrnK)

miopenStatus_t
miopenGetLRNDescriptor(
    const miopenLRNDescriptor_t lrnDesc,
    miopenLRNMode_t *mode,
    unsigned *lrnN,
    double *lrnAlpha,
    double *lrnBeta,
    double *lrnK)

cudnnStatus_t
cudnnDestroyLRNDescriptor(
    cudnnLRNDescriptor_t lrnDesc)

miopenStatus_t
miopenDestroyLRNDescriptor(
    miopenLRNDescriptor_t lrnDesc)

cudnnStatus_t
cudnnLRNCrossChannelForward(
    cudnnHandle_t handle,
    cudnnLRNDescriptor_t normDesc,
    cudnnLRNMode_t lrnMode,
    const void* alpha,
    const cudnnTensorDescriptor_t xDesc,
    const void *x,
    const void *beta,
    const cudnnTensorDescriptor_t yDesc,
    void *y)

miopenStatus_t
miopenLRNForward(
    miopenHandle_t handle,
    const miopenLRNDescriptor_t lrnDesc,
    const void *alpha,
    const miopenTensorDescriptor_t xDesc,
    const void *x,
    const void *beta,
    const miopenTensorDescriptor_t yDesc,
    void *y,
    bool do_backward,
    void  *workspace)

cudnnStatus_t
cudnnLRNCrossChannelBackward(
    cudnnHandle_t handle,
    cudnnLRNDescriptor_t normDesc,
    cudnnLRNMode_t lrnMode,
    const void* alpha,
    const cudnnTensorDescriptor_t yDesc,
    const void *y,
    const cudnnTensorDescriptor_t dyDesc,
    const void *dy,
    const cudnnTensorDescriptor_t xDesc,
    const void *x,
    const void *beta,
    const cudnnTensorDescriptor_t dxDesc,
    void *dx)

miopenStatus_t
miopenLRNBackward(
    miopenHandle_t handle,
    const miopenLRNDescriptor_t lrnDesc,
    const void *alpha,
    const miopenTensorDescriptor_t yDesc,
    const void *y,
    const miopenTensorDescriptor_t dyDesc,
    const void *dy,
    const miopenTensorDescriptor_t xDesc,
    const void *x, const void *beta,
    const miopenTensorDescriptor_t dxDesc,
    void *dx,
    const void *workspace)

NA

 
miopenStatus_t
miopenLRNGetWorkSpaceSize(
    const miopenTensorDescriptor_t yDesc,
    size_t *workSpaceSize)

cudnnStatus_t
cudnnDeriveBNTensorDescriptor(
    cudnnTensorDescriptor_t derivedBnDesc,
    const cudnnTensorDescriptor_t xDesc,
    cudnnBatchNormMode_t mode)

miopenStatus_t
miopenDeriveBNTensorDescriptor(
    miopenTensorDescriptor_t derivedBnDesc,
    const miopenTensorDescriptor_t xDesc,
    miopenBatchNormMode_t bn_mode)

Batch normalization operations#

cuDNN

MIOpen

 
cudnnStatus_t
cudnnBatchNormalizationForwardTraining(
    cudnnHandle_t handle,
    cudnnBatchNormMode_t mode,
    void *alpha,
    void *beta,
    const cudnnTensorDescriptor_t xDesc,
    const void *x,
    const cudnnTensorDescriptor_t yDesc,
    void *y,
    const cudnnTensorDescriptor_t
        bnScaleBiasMeanVarDesc,
    void *bnScale,
    void *bnBias,
    double exponentialAverageFactor,
    void *resultRunningMean,
    void *resultRunningVariance,
    double epsilon,
    void *resultSaveMean,
    void *resultSaveInvVariance)

miopenStatus_t
miopenBatchNormalizationForwardTraining(
    miopenHandle_t handle,
    miopenBatchNormMode_t bn_mode,
    void *alpha,
    void *beta,
    const miopenTensorDescriptor_t xDesc,
    const void *x,
    const miopenTensorDescriptor_t yDesc,
    void *y,
    const miopenTensorDescriptor_t
        bnScaleBiasMeanVarDesc,
    void *bnScale,
    void *bnBias,
    double expAvgFactor,
    void *resultRunningMean,
    void *resultRunningVariance,
    double epsilon,
    void *resultSaveMean,
    void *resultSaveInvVariance)

cudnnStatus_t
cudnnnBatchNormalizationForwardInference(
    cudnnHandle_t handle,
    cudnnBatchNormMode_t mode,
    void *alpha,
    void *beta,
    const cudnnTensorDescriptor_t xDesc,
    const void *x,
    const cudnnTensorDescriptor_t yDesc,
    void *y,
    const cudnnTensorDescriptor_t
        bnScaleBiasMeanVarDesc,
    const void *bnScale,
    void *bnBias,
    const void *estimatedMean,
    const void *estimatedVariance,
    double epsilon)

miopenStatus_t
miopenBatchNormalizationForwardInference(
    miopenHandle_t handle,
    miopenBatchNormMode_t bn_mode,
    void *alpha,
    void *beta,
    const miopenTensorDescriptor_t xDesc,
    const void *x,
    const miopenTensorDescriptor_t yDesc,
    void *y,
    const miopenTensorDescriptor_t
        bnScaleBiasMeanVarDesc,
    void *bnScale,
    void *bnBias,
    void *estimatedMean,
    void *estimatedVariance,
    double epsilon)

cudnnStatus_t
cudnnBatchNormalizationBackward(
    cudnnHandle_t handle,
    cudnnBatchNormMode_t mode,
    const void *alphaDataDiff,
    const void *betaDataDiff,
    const void *alphaParamDiff,
    const void *betaParamDiff,
    const cudnnTensorDescriptor_t xDesc,
    const void *x,
    const cudnnTensorDescriptor_t dyDesc,
    const void *dy,
    const cudnnTensorDescriptor_t dxDesc,
    void *dx,
    const cudnnTensorDescriptor_t
        bnScaleBiasDiffDesc,
    const void *bnScale,
    void *resultBnScaleDiff,
    void *resultBnBiasDiff,
    double epsilon,
    const void *savedMean,
    const void *savedInvVariance)

miopenStatus_t
miopenBatchNormalizationBackward(
    miopenHandle_t handle,
    miopenBatchNormMode_t bn_mode,
    const void *alphaDataDiff,
    const void *betaDataDiff,
    const void *alphaParamDiff,
    const void *betaParamDiff,
    const miopenTensorDescriptor_t xDesc,
    const void *x,
    const miopenTensorDescriptor_t dyDesc,
    const void *dy,
    const miopenTensorDescriptor_t dxDesc,
    void *dx,
    const miopenTensorDescriptor_t
        bnScaleBiasDiffDesc,
    const void *bnScale,
    void *resultBnScaleDiff,
    void *resultBnBiasDiff,
    double epsilon,
    const void *savedMean,
    const void *savedInvVariance)
Contents