TheRock/rocm-libraries/projects/miopen/test/gtest/bn_3d_spatial_test.cpp

/*******************************************************************************
 *
 * MIT License
 *
 * Copyright (c) 2025 Advanced Micro Devices, Inc.
 *
 * Permission is hereby granted, free of charge, to any person obtaining a copy
 * of this software and associated documentation files (the "Software"), to deal
 * in the Software without restriction, including without limitation the rights
 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 * copies of the Software, and to permit persons to whom the Software is
 * furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice shall be included in all
 * copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 * SOFTWARE.
 *
 *******************************************************************************/

#include <miopen/activ.hpp>
#include <miopen/batch_norm.hpp>
#include <miopen/ford.hpp>

#include "get_handle.hpp"
#include "gtest_common.hpp"
#include "random.hpp"
#include "test_parameter_name_generator.hpp"

// Run CPU emulations in hierarchical reduction mode.
#define MIO_HEIRARCH_SEL 1
#define MIO_BN_TEST_EXPAVGFACTOR 0.1
#define MIO_BN_TEST_EPSILON 1e-5 // FLT_EPSILON
#define MIO_BN_SP_TEST_DEBUG 0
#define MIO_BN_USE_MIX_PREC 1
#if MIO_BN_USE_MIX_PREC == 1
#define PREC_TYPE float
#else
#define PREC_TYPE T
#endif

namespace {

using TestCase = NamedContainer<std::vector<int>>;

//****************************************************
// FORWARD TRAIN
//****************************************************
template <class T, class U>
struct verify_forward_train_3d_bn_spatial
{
    const tensor<T> input;
    const tensor<U> scale;
    const tensor<U> shift;

    std::tuple<tensor<T>, tensor<U>, tensor<U>, tensor<U>, tensor<U>> cpu() const
    {
#if(MIO_BN_TIME_EVERYTHING == 1)
        auto t_start = std::chrono::high_resolution_clock::now();
#endif
        double epsilon      = MIO_BN_TEST_EPSILON;
        double expAvgFactor = MIO_BN_TEST_EXPAVGFACTOR;

        size_t n_batch, channels, depth, height, width;
        std::tie(n_batch, channels, depth, height, width) =
            miopen::tien<5>(input.desc.GetLengths());

        size_t rs_n_batch, rs_channels, rs_depth, rs_height, rs_width;
        auto derivedBnDesc = miopen::TensorDescriptor{};
        miopen::DeriveBNTensorDescriptor(derivedBnDesc, input.desc, miopenBNSpatial);

        std::tie(rs_n_batch, rs_channels, rs_depth, rs_height, rs_width) =
            miopen::tien<5>(derivedBnDesc.GetLengths());

        tensor<U> runMean;
        tensor<U> runVar;

        if(input.desc.GetType() == miopenFloat)
        {
            runMean = tensor<U>{rs_n_batch, rs_channels, rs_depth, rs_height, rs_width}.generate(
                tensor_elem_gen_integer{17});
            runVar = tensor<U>{rs_n_batch, rs_channels, rs_depth, rs_height, rs_width}.generate(
                tensor_elem_gen_integer{17});
        }
        else
        {
            prng::reset_seed();
            runMean = tensor<U>{rs_n_batch, rs_channels, rs_depth, rs_height, rs_width};
            runVar  = tensor<U>{rs_n_batch, rs_channels, rs_depth, rs_height, rs_width};

            const double Data_scale = 0.001;
            for(size_t i = 0; i < runMean.desc.GetElementSize(); i++)
            {
                runMean[i] = prng::gen_descreet_uniform_sign<U>(Data_scale, 100);
                runVar[i]  = prng::gen_descreet_unsigned<U>(Data_scale, 100);
            }
        }
        auto saveMean   = tensor<U>{rs_n_batch, rs_channels, rs_depth, rs_height, rs_width};
        auto saveInvVar = tensor<U>{rs_n_batch, rs_channels, rs_depth, rs_height, rs_width};
        auto out        = input;
        std::fill(out.begin(), out.end(), 0);

        const unsigned int in_cstride = depth * height * width;
        const auto ndhw               = double(in_cstride * n_batch);

        miopen::par_for(channels, 1, [&](int cidx) {
            double elemStd        = 0.;
            double variance_accum = 0.;
            double mean_accum     = 0.;
            double invVar         = 0.;
            double newRunMean     = 0.;
            double adjust         = 0.;

#if(MIO_HEIRARCH_SEL == 1)
            std::vector<double> variance_accum_arr(depth * height, 0.0);
            std::vector<double> mean_accum_arr(depth * height, 0.0);
            std::vector<double> dshift_accum_arr(depth * height, 0.0);
            std::vector<double> dscale_accum_arr(depth * height, 0.0);
#endif

#if(MIO_HEIRARCH_SEL == 0)
            // process the batch per channel
            miopen::ford(n_batch, depth, height, width)(
                [&](size_t bidx, size_t didx, size_t row, size_t column) {
                    // #1 calculate the mean
                    // iterating through the stack of images in the mini_batch
                    mean_accum += input(bidx, cidx, didx, row, column);
                });
#else
            miopen::ford(depth, height, width, n_batch)([&](size_t didx, size_t row, size_t column, size_t bidx) {
                mean_accum_arr[row] += input(bidx, cidx, didx, row, column);
            });

            for(size_t i = 0; i<height; i++) mean_accum += mean_accum_arr[i];
#endif
            mean_accum /= ndhw;

            elemStd        = 0.;
            variance_accum = 0.;

#if(MIO_HEIRARCH_SEL == 0)
            // #2 calculate the variances
            // sigma^2 = (1/batch_mean) * sum( (x_i - batch_mean)^2 )
            miopen::ford(n_batch, depth, height, width)(
                [&](size_t bidx, size_t didx, size_t row, size_t column) {
                    // using out buffer as scratchpad
                    out(bidx, cidx, didx, row, column) = elemStd =
                        (input(bidx, cidx, didx, row, column) - mean_accum); // (x_i - mean)
                    variance_accum += (elemStd * elemStd); // sum{ (x_i - mean)^2 }
                });
#else
            miopen::ford(depth, height, width, n_batch)([&](size_t didx, size_t row, size_t column, size_t bidx) {
                out(bidx,cidx,didx,row,column) = elemStd = input(bidx,cidx,didx,row,column) - mean_accum;
                variance_accum_arr[row] += elemStd * elemStd;
            });

            for(size_t i = 0; i<height; i++) variance_accum += variance_accum_arr[i];
#endif
            variance_accum /= ndhw; // (1/N)*sum{ (x_i - mean)^2 }
            // #3 add epsilon for numeric stability, sqr_root, and invert
            invVar = 1.0 / sqrt(variance_accum + epsilon);

            // #4 apply the normalization
            // x_hat = (x_i - mean) / sqrt(variance_accum + epsilon)
            miopen::ford(n_batch, depth, height, width)(
                [&](size_t bidx, size_t didx, size_t row, size_t column) {
                    // #5 Gamma and Beta adjust
                    // y_i = gamma*x_hat + beta
                    out(bidx, cidx, didx, row, column) =
                        scale(0, cidx, 0, 0, 0) * (invVar * out(bidx, cidx, didx, row, column)) +
                        shift(0, cidx, 0, 0, 0);
                });

            saveMean(0, cidx, 0, 0, 0)   = mean_accum;
            saveInvVar(0, cidx, 0, 0, 0) = invVar;

            newRunMean = runMean(0, cidx, 0, 0, 0) * (1 - expAvgFactor);
            runMean(0, cidx, 0, 0, 0) =
                mean_accum * expAvgFactor + newRunMean; // newMean*factor + tmp
            // var(n+1) = p * var(n-1) + (1 - p)*(b/b-1)*var(n)
            adjust = (n_batch * depth * height * width == 1) ? variance_accum
                                                             : (ndhw / (ndhw - 1)) * variance_accum;
            runVar(0, cidx, 0, 0, 0) =
                (1 - expAvgFactor) * runVar(0, cidx, 0, 0, 0) + expAvgFactor * adjust;
        });

#if(MIO_BN_TIME_EVERYTHING == 1)
        auto t_end = std::chrono::high_resolution_clock::now();

        std::cout << "Wall clock: CPU forward_train_3d_bn_spatial pass time: "
                  << std::chrono::duration<double>(t_end - t_start).count() << " seconds."
                  << std::endl;
#endif
        return std::make_tuple(out, runMean, runVar, saveMean, saveInvVar);
    }

    std::tuple<tensor<T>, tensor<U>, tensor<U>, tensor<U>, tensor<U>> gpu() const
    {
#if(MIO_BN_TIME_EVERYTHING == 1)
        auto t_start = std::chrono::high_resolution_clock::now();
#endif
        auto&& handle = get_handle();

        size_t n_batch, channels, depth, height, width;
        std::tie(n_batch, channels, depth, height, width) =
            miopen::tien<5>(input.desc.GetLengths());

        auto out = input;
        std::fill(out.begin(), out.end(), 0);

        size_t rs_n_batch, rs_channels, rs_depth, rs_height, rs_width;
        auto derivedBnDesc = miopen::TensorDescriptor{};

        miopen::DeriveBNTensorDescriptor(derivedBnDesc, input.desc, miopenBNSpatial);

        std::tie(rs_n_batch, rs_channels, rs_depth, rs_height, rs_width) =
            miopen::tien<5>(derivedBnDesc.GetLengths());

        tensor<U> runMean;
        tensor<U> runVar;

        if(input.desc.GetType() == miopenFloat)
        {
            runMean = tensor<U>{rs_n_batch, rs_channels, rs_depth, rs_height, rs_width}.generate(
                tensor_elem_gen_integer{17});
            runVar = tensor<U>{rs_n_batch, rs_channels, rs_depth, rs_height, rs_width}.generate(
                tensor_elem_gen_integer{17});
        }
        else
        {
            prng::reset_seed();
            runMean = tensor<U>{rs_n_batch, rs_channels, rs_depth, rs_height, rs_width};
            runVar  = tensor<U>{rs_n_batch, rs_channels, rs_depth, rs_height, rs_width};

            const double Data_scale = 0.001;
            for(size_t i = 0; i < runMean.desc.GetElementSize(); i++)
            {
                runMean[i] = prng::gen_descreet_uniform_sign<U>(Data_scale, 100);
                runVar[i]  = prng::gen_descreet_unsigned<U>(Data_scale, 100);
            }
        }

        auto saveMean   = tensor<U>{rs_n_batch, rs_channels, rs_depth, rs_height, rs_width};
        auto saveInvVar = tensor<U>{rs_n_batch, rs_channels, rs_depth, rs_height, rs_width};

        // in buffers
        auto in_dev    = handle.Write(input.data);
        auto scale_dev = handle.Write(scale.data);
        auto shift_dev = handle.Write(shift.data);

        // out buffers
        auto runMean_dev    = handle.Write(runMean.data);
        auto runVar_dev     = handle.Write(runVar.data);
        auto saveMean_dev   = handle.Create<U>(channels);
        auto saveInvVar_dev = handle.Create<U>(channels);
        auto out_dev        = handle.Create<T>(n_batch * channels * depth * height * width);

        double epsilon      = MIO_BN_TEST_EPSILON;
        double expAvgFactor = MIO_BN_TEST_EXPAVGFACTOR;

        float alpha = 1.0;
        float beta  = 0.0;

        miopen::ActivationDescriptor actDesc(miopenActivationPASTHRU, 0.0f, 0.0f, 0.0f);
        miopen::BatchNormForwardTraining(handle,
                                         miopenBNSpatial,
                                         &alpha,
                                         &beta,
                                         miopen::BuildReshaped4DTensorDescriptor(input.desc),
                                         in_dev.get(),
                                         miopen::BuildReshaped4DTensorDescriptor(out.desc),
                                         out_dev.get(),
                                         miopen::BuildReshaped4DTensorDescriptor(scale.desc),
                                         miopen::BuildReshaped4DTensorDescriptor(shift.desc),
                                         miopen::BuildReshaped4DTensorDescriptor(shift.desc),
                                         miopen::BuildReshaped4DTensorDescriptor(shift.desc),
                                         scale_dev.get(),
                                         shift_dev.get(),
                                         expAvgFactor,
                                         runMean_dev.get(),
                                         runVar_dev.get(),
                                         epsilon,
                                         saveMean_dev.get(),
                                         saveInvVar_dev.get(),
                                         actDesc);

        saveMean.data   = handle.Read<U>(saveMean_dev, saveMean.data.size());
        saveInvVar.data = handle.Read<U>(saveInvVar_dev, saveInvVar.data.size());
        runMean.data    = handle.Read<U>(runMean_dev, runMean.data.size());
        runVar.data     = handle.Read<U>(runVar_dev, runVar.data.size());
        out.data        = handle.Read<T>(out_dev, out.data.size());

#if(MIO_BN_TIME_EVERYTHING == 1)
        auto t_end = std::chrono::high_resolution_clock::now();

        std::cout << "Wall clock: GPU forward_train_3d_bn_spatial pass time: "
                  << std::chrono::duration<double>(t_end - t_start).count() << " seconds."
                  << std::endl;
#endif

        return std::make_tuple(out, runMean, runVar, saveMean, saveInvVar);
    }

    void fail(int badtensor) const
    {
        std::cout << "Forward Train Spatial Batch Normalization: " << std::endl;
        std::cout << "Input tensor: " << input.desc.ToString() << std::endl;

        switch(badtensor)
        {
        case(0): std::cout << "Output tensor output failed verification." << std::endl; break;
        case(1): std::cout << "Running Mean output tensor failed verification." << std::endl; break;
        case(2):
            std::cout << "Running Variance output tensor failed verification." << std::endl;
            break;
        case(3): std::cout << "Saved Mean tensor failed verification." << std::endl; break;
        case(4): std::cout << "Saved Variance tensor failed verification." << std::endl; break;
        default: break;
        }
    }
};

//****************************************************
// FORWARD INFERENCE
//****************************************************
template <class T, class U>
struct verify_forward_infer_3d_bn_spatial_recalc
{
    const tensor<T> input;
    const tensor<U> scale;
    const tensor<U> shift;

    tensor<T> cpu() const
    {
#if(MIO_BN_TIME_EVERYTHING == 1)
        auto t_start = std::chrono::high_resolution_clock::now();
#endif
        double epsilon = MIO_BN_TEST_EPSILON;

        size_t n_batch, channels, depth, height, width;
        std::tie(n_batch, channels, depth, height, width) =
            miopen::tien<5>(input.desc.GetLengths());

        auto out = input;
        std::fill(out.begin(), out.end(), 0);

        const unsigned int in_cstride = depth * height * width;
        const auto ndhw               = double(in_cstride * n_batch);

        miopen::par_for(channels, 1, [&](int cidx) {
            double elemStd        = 0.;
            double variance_accum = 0.;
            double mean_accum     = 0.;
            double inhat          = 0.;
            double invVar         = 0.;

            // process the batch per channel
            miopen::ford(depth, height, width, n_batch)(
                [&](size_t didx, size_t row, size_t column, size_t bidx) {
                    // iterating through the stack of images in the mini_batch
                    mean_accum += input(bidx, cidx, didx, row, column);
                });

            mean_accum /= ndhw;

            elemStd        = 0.;
            variance_accum = 0.;

            // #2 calculate the variances
            // sigma^2 = (1/batch_mean) * sum( (x_i - batch_mean)^2 )
            miopen::ford(depth, height, width, n_batch)(
                [&](size_t didx, size_t row, size_t column, size_t bidx) {
                    // using out buffer as scratchpad
                    out(bidx, cidx, didx, row, column) = elemStd =
                        (input(bidx, cidx, didx, row, column) - mean_accum); // (x_i - mean)
                    variance_accum += (elemStd * elemStd); // sum{ (x_i - mean)^2 }
                });

            variance_accum /= ndhw; // (1/N)*sum{ (x_i - mean)^2 }

            // #3 add epsilon for numeric stability, sqr_root, and invert
            invVar = 1.0 / sqrt(variance_accum + epsilon);

            // #4 apply the normalization
            // x_hat = (x_i - mean) / sqrt(variance_accum - epsilon)
            miopen::ford(depth, height, width, n_batch)(
                [&](size_t didx, size_t row, size_t column, size_t bidx) {
                    elemStd = out(bidx,
                                  cidx,
                                  didx,
                                  row,
                                  column); // using saved values from output tensor
                    inhat   = elemStd * invVar;
                    // #5 Gamma and Beta adjust // y_i = gamma*x_hat + beta
                    out(bidx, cidx, didx, row, column) =
                        scale(0, cidx, 0, 0, 0) * inhat + shift(0, cidx, 0, 0, 0);
                });
        });

#if(MIO_BN_TIME_EVERYTHING == 1)
        auto t_end = std::chrono::high_resolution_clock::now();

        std::cout << "Wall clock: CPU forward_infer_3d_bn_spatial_recalc pass time: "
                  << std::chrono::duration<double>(t_end - t_start).count() << " seconds."
                  << std::endl;
#endif
        return out;
    }

    tensor<T> gpu() const
    {
#if(MIO_BN_TIME_EVERYTHING == 1)
        auto t_start = std::chrono::high_resolution_clock::now();
#endif
        auto&& handle = get_handle();
        auto out      = input;
        std::fill(out.begin(), out.end(), 0);

        auto in_dev    = handle.Write(input.data);
        auto scale_dev = handle.Write(scale.data);
        auto shift_dev = handle.Write(shift.data);
        auto out_dev   = handle.Write(out.data);

        float alpha = 1.0;
        float beta  = 0.0;

        double epsilon = MIO_BN_TEST_EPSILON;

        miopen::ActivationDescriptor actDesc(miopenActivationPASTHRU, 0.0f, 0.0f, 0.0f);
        miopen::BatchNormForwardInference(handle,
                                          miopenBNSpatial,
                                          &alpha,
                                          &beta,
                                          miopen::BuildReshaped4DTensorDescriptor(input.desc),
                                          in_dev.get(),
                                          miopen::BuildReshaped4DTensorDescriptor(out.desc),
                                          out_dev.get(),
                                          miopen::BuildReshaped4DTensorDescriptor(scale.desc),
                                          miopen::BuildReshaped4DTensorDescriptor(shift.desc),
                                          miopen::BuildReshaped4DTensorDescriptor(shift.desc),
                                          miopen::BuildReshaped4DTensorDescriptor(shift.desc),
                                          scale_dev.get(),
                                          shift_dev.get(),
                                          nullptr,
                                          nullptr,
                                          epsilon,
                                          actDesc);
        out.data = handle.Read<T>(out_dev, out.data.size());

#if(MIO_BN_TIME_EVERYTHING == 1)
        auto t_end = std::chrono::high_resolution_clock::now();

        std::cout << "Wall clock: GPU forward_infer_3d_bn_spatial_recalc pass time: "
                  << std::chrono::duration<double>(t_end - t_start).count() << " seconds."
                  << std::endl;
#endif
        return out;
    }

    void fail(int) const
    {
        std::cout << "Forward Inference Spatial Batch Normalization Recalc: " << std::endl;
        std::cout << "Input tensor: " << input.desc.ToString() << std::endl;
    }
};

template <class T, class U>
struct verify_forward_infer_3d_bn_spatial_use_est
{
    const tensor<T> input;
    const tensor<U> scale;
    const tensor<U> shift;
    const tensor<U> estMean;
    const tensor<U> estVar;

    tensor<T> cpu() const
    {
#if(MIO_BN_TIME_EVERYTHING == 1)
        auto t_start = std::chrono::high_resolution_clock::now();
#endif
        double epsilon = MIO_BN_TEST_EPSILON;

        size_t n_batch, channels, depth, height, width;
        std::tie(n_batch, channels, depth, height, width) =
            miopen::tien<5>(input.desc.GetLengths());

        auto out = input;
        std::fill(out.begin(), out.end(), 0);

        miopen::par_for(channels, 1, [&](int cidx) {
            double elemStd  = 0.;
            double variance = estVar(0, cidx, 0, 0, 0);
            double mean     = estMean(0, cidx, 0, 0, 0);
            double inhat    = 0.;
            double invVar   = 1.0 / sqrt(variance + epsilon);

            // process the batch per channel
            miopen::ford(n_batch, depth, height, width)(
                [&](size_t bidx, size_t didx, size_t row, size_t column) {
                    elemStd = input(bidx, cidx, didx, row, column) - mean;
                    inhat   = elemStd * invVar;
                    out(bidx, cidx, didx, row, column) =
                        scale(0, cidx, 0, 0, 0) * inhat + shift(0, cidx, 0, 0, 0);
                });
        });

#if(MIO_BN_TIME_EVERYTHING == 1)
        auto t_end = std::chrono::high_resolution_clock::now();

        std::cout << "Wall clock: CPU forward_infer_3d_bn_spatial_use_est pass time: "
                  << std::chrono::duration<double>(t_end - t_start).count() << " seconds."
                  << std::endl;
#endif

        return out;
    }

    tensor<T> gpu() const
    {
#if(MIO_BN_TIME_EVERYTHING == 1)
        auto t_start = std::chrono::high_resolution_clock::now();
#endif
        auto&& handle = get_handle();
        auto out      = input;
        std::fill(out.begin(), out.end(), 0);

        auto in_dev      = handle.Write(input.data);
        auto estMean_dev = handle.Write(estMean.data);
        auto estVar_dev  = handle.Write(estVar.data);
        auto scale_dev   = handle.Write(scale.data);
        auto shift_dev   = handle.Write(shift.data);
        auto out_dev     = handle.Write(out.data);

        float alpha = 1.0;
        float beta  = 0.0;

        double epsilon = MIO_BN_TEST_EPSILON;

        miopen::ActivationDescriptor actDesc(miopenActivationPASTHRU, 0.0f, 0.0f, 0.0f);
        miopen::BatchNormForwardInference(handle,
                                          miopenBNSpatial,
                                          &alpha,
                                          &beta,
                                          miopen::BuildReshaped4DTensorDescriptor(input.desc),
                                          in_dev.get(),
                                          miopen::BuildReshaped4DTensorDescriptor(out.desc),
                                          out_dev.get(),
                                          miopen::BuildReshaped4DTensorDescriptor(scale.desc),
                                          miopen::BuildReshaped4DTensorDescriptor(shift.desc),
                                          miopen::BuildReshaped4DTensorDescriptor(shift.desc),
                                          miopen::BuildReshaped4DTensorDescriptor(shift.desc),
                                          scale_dev.get(),
                                          shift_dev.get(),
                                          estMean_dev.get(),
                                          estVar_dev.get(),
                                          epsilon,
                                          actDesc);
        out.data = handle.Read<T>(out_dev, out.data.size());
#if(MIO_BN_TIME_EVERYTHING == 1)
        auto t_end = std::chrono::high_resolution_clock::now();

        std::cout << "Wall clock: GPU forward_infer_3d_bn_spatial_use_est pass time: "
                  << std::chrono::duration<double>(t_end - t_start).count() << " seconds."
                  << std::endl;
#endif
        return out;
    }

    void fail(int) const
    {
        std::cout << "Forward Inference Spatial Batch Normalization Use Estimated: " << std::endl;
        std::cout << "Input tensor: " << input.desc.ToString() << std::endl;
    }
};

//****************************************************
// BACKWARDS PROPAGATION
//****************************************************
template <class T, class U>
struct verify_backward_3d_bn_spatial_recalc
{
    const tensor<T> x_input;
    const tensor<T> dy_input;
    const tensor<U> scale;

    std::tuple<tensor<T>, tensor<U>, tensor<U>> cpu() const
    {
#if(MIO_BN_TIME_EVERYTHING == 1)
        auto t_start = std::chrono::high_resolution_clock::now();
#endif
        double epsilon = MIO_BN_TEST_EPSILON;

        size_t n_batch, channels, depth, height, width;
        std::tie(n_batch, channels, depth, height, width) =
            miopen::tien<5>(x_input.desc.GetLengths());

        size_t ss_n_batch, ss_channels, ss_depth, ss_height, ss_width;
        auto derivedBnDesc = miopen::TensorDescriptor{};
        miopen::DeriveBNTensorDescriptor(derivedBnDesc, x_input.desc, miopenBNSpatial);
        std::tie(ss_n_batch, ss_channels, ss_depth, ss_height, ss_width) =
            miopen::tien<5>(derivedBnDesc.GetLengths());

        auto dx_out = tensor<T>{n_batch, channels, depth, height, width};
        std::fill(dx_out.begin(), dx_out.end(), 0);

        auto dscale = tensor<U>{ss_n_batch, ss_channels, ss_depth, ss_height, ss_width};
        std::fill(dscale.begin(), dscale.end(), 0);

        auto dshift = tensor<U>{ss_n_batch, ss_channels, ss_depth, ss_height, ss_width};
        std::fill(dshift.begin(), dshift.end(), 0);

        const unsigned int in_dstride = height * width;
        const unsigned int in_cstride = depth * in_dstride;
        const auto ndhw               = double(in_cstride * n_batch);

        miopen::par_for(channels, 1, [&](int cidx) {
            double elemStd = 0.;
            unsigned int xhat_index;
            double mean     = 0.;
            double invVar   = 0.;
            double dyelem   = 0.;
            double variance = 0.;

            std::vector<double> xhat(n_batch * in_cstride, 0.0);

#if(MIO_HEIRARCH_SEL == 1)
            std::vector<double> variance_accum_arr(depth * height, 0.0);
            std::vector<double> mean_accum_arr(depth * height, 0.0);
            std::vector<double> dshift_accum_arr(depth * height, 0.0);
            std::vector<double> dscale_accum_arr(depth * height, 0.0);
#endif

// process the batch per channel
#if(MIO_HEIRARCH_SEL == 0)
            miopen::ford(depth, height, width, n_batch)(
                [&](size_t didx, size_t row, size_t column, size_t bidx) {
                    // #1 calculate the mean
                    mean += x_input(bidx, cidx, didx, row, column);
                });
#else
            miopen::ford(depth, height, width, n_batch)([&](size_t didx, size_t row, size_t column, size_t bidx) {
                mean_accum_arr[row] += x_input(bidx, cidx, didx, row, column);
            });

            for(size_t i = 0; i<height; i++) mean += mean_accum_arr[i];
#endif
            mean /= ndhw;

            elemStd  = 0.;
            variance = 0.;
#if(MIO_HEIRARCH_SEL == 0)
            miopen::ford(depth, height, width, n_batch)(
                [&](size_t didx, size_t row, size_t column, size_t bidx) {
                    // per (x-dims) channel load a block of data into LDS
                    elemStd = x_input(bidx, cidx, didx, row, column) - mean; // (x_i - mean)
                    variance += elemStd * elemStd; // sum{ (x_i - mean)^2 }
                });
#else
            miopen::ford(depth, height, width, n_batch)([&](size_t didx, size_t row, size_t column, size_t bidx) {
                elemStd = x_input(bidx,cidx,didx,row,column) - mean;
                variance_accum_arr[row] += elemStd*elemStd;
            });

            for(size_t i = 0; i<height; i++) variance += variance_accum_arr[i];
#endif
            variance /= ndhw; // (1/(N*H*W))*sum{ (x_i - mean)^2 }
            invVar = 1. / double(sqrt(variance + epsilon));

            dscale(0, cidx, 0, 0, 0) = 0.;

#if(MIO_HEIRARCH_SEL == 0)
            miopen::ford(depth, height, width, n_batch)(
                [&](size_t didx, size_t row, size_t column, size_t bidx) {
                    xhat_index = in_cstride * bidx + in_dstride * didx + width * row + column;
                    // per (x-dims) channel load a block of data into LDS
                    elemStd = x_input(bidx, cidx, didx, row, column) - mean; // (x_i - mean)
                    xhat[xhat_index] = elemStd * invVar;
                    dyelem           = dy_input(bidx, cidx, didx, row, column);
                    dshift(0, cidx, 0, 0, 0) += dyelem;
                    dscale(0, cidx, 0, 0, 0) += xhat[xhat_index] * dyelem;
                });
#else
            miopen::ford(depth, height, width, n_batch)([&](size_t didx, size_t row, size_t column, size_t bidx) {
                xhat_index = in_cstride*bidx + in_dstride*didx + width*row + column;
                //per (x-dims) channel load a block of data into LDS
                elemStd             = x_input(bidx, cidx, didx, row, column) - mean;// (x_i - mean)
                xhat[xhat_index]    = elemStd*invVar;
                dyelem              = dy_input(bidx, cidx, didx, row, column);
                dshift_accum_arr[row] += dyelem;
                dscale_accum_arr[row] += xhat[xhat_index]*dyelem;
                //dscale_accum_arr[row] += x_input(bidx,cidx,row,column);;//dscale_accum_arr[row] += xhat[xhat_index];
                //dscale_accum_arr[row] += 1.0;//DEBUG
            });

            for(size_t i = 0; i<height; i++) {
                dshift(0,cidx,0,0,0) += dshift_accum_arr[i];
                dscale(0,cidx,0,0,0) += dscale_accum_arr[i];
            }
#endif
            miopen::ford(depth, height, width, n_batch)(
                [&](size_t didx, size_t row, size_t column, size_t bidx) {
                    xhat_index = in_cstride * bidx + in_dstride * didx + width * row + column;

                    const double tmp1 =
                        ndhw * dy_input(bidx, cidx, didx, row, column) - dshift(0, cidx, 0, 0, 0);
                    const double tmp2 = -xhat[xhat_index] * dscale(0, cidx, 0, 0, 0);
                    const double tmp3 = (scale(0, cidx, 0, 0, 0) * invVar) / ndhw;
                    dx_out(bidx, cidx, didx, row, column) = tmp3 * (tmp2 + tmp1);
                });
        }); // for (channel)

#if(MIO_BN_TIME_EVERYTHING == 1)
        auto t_end = std::chrono::high_resolution_clock::now();

        std::cout << "Wall clock: CPU backward_3d_bn_spatial_recalc pass time: "
                  << std::chrono::duration<double>(t_end - t_start).count() << " seconds."
                  << std::endl;
#endif

        return std::make_tuple(dx_out, dscale, dshift);
    }

    std::tuple<tensor<T>, tensor<U>, tensor<U>> gpu() const
    {
#if(MIO_BN_TIME_EVERYTHING == 1)
        auto t_start = std::chrono::high_resolution_clock::now();
#endif
        auto&& handle = get_handle();

        size_t n_batch, channels, depth, height, width;
        std::tie(n_batch, channels, depth, height, width) =
            miopen::tien<5>(x_input.desc.GetLengths());

        auto dx_out = tensor<T>{n_batch, channels, depth, height, width};
        std::fill(dx_out.begin(), dx_out.end(), 0);

        size_t ss_n_batch, ss_channels, ss_depth, ss_height, ss_width;
        auto derivedBnDesc = miopen::TensorDescriptor{};
        miopen::DeriveBNTensorDescriptor(derivedBnDesc, x_input.desc, miopenBNSpatial);
        std::tie(ss_n_batch, ss_channels, ss_depth, ss_height, ss_width) =
            miopen::tien<5>(derivedBnDesc.GetLengths());

        auto dscale = tensor<U>{ss_n_batch, ss_channels, ss_depth, ss_height, ss_width};
        std::fill(dscale.begin(), dscale.end(), 0);

        auto dshift = tensor<U>{ss_n_batch, ss_channels, ss_depth, ss_height, ss_width};
        std::fill(dshift.begin(), dshift.end(), 0);

        float alpha = 1.0;
        float beta  = 0.0;

        auto xin_dev    = handle.Write(x_input.data);
        auto dyin_dev   = handle.Write(dy_input.data);
        auto scale_dev  = handle.Write(scale.data);
        auto dscale_dev = handle.Write(dscale.data);
        auto dshift_dev = handle.Write(dshift.data);
        auto dx_out_dev = handle.Write(dx_out.data);

        double epsilon = MIO_BN_TEST_EPSILON;

        miopen::ActivationDescriptor actDesc(miopenActivationPASTHRU, 0.0f, 0.0f, 0.0f);
        miopen::BatchNormBackward(handle,
                                  miopenBNSpatial,
                                  &alpha,
                                  &beta,
                                  &alpha,
                                  &beta,
                                  miopen::BuildReshaped4DTensorDescriptor(x_input.desc),
                                  xin_dev.get(),
                                  miopen::BuildReshaped4DTensorDescriptor(dy_input.desc),
                                  dyin_dev.get(),
                                  miopen::BuildReshaped4DTensorDescriptor(dx_out.desc),
                                  dx_out_dev.get(),
                                  miopen::BuildReshaped4DTensorDescriptor(scale.desc),
                                  miopen::BuildReshaped4DTensorDescriptor(dshift.desc),
                                  miopen::BuildReshaped4DTensorDescriptor(dshift.desc),
                                  miopen::BuildReshaped4DTensorDescriptor(dshift.desc),
                                  scale_dev.get(),
                                  nullptr,
                                  dscale_dev.get(),
                                  dshift_dev.get(),
                                  epsilon,
                                  nullptr,
                                  nullptr,
                                  actDesc);

        dx_out.data = handle.Read<T>(dx_out_dev, dx_out.data.size());
        dscale.data = handle.Read<U>(dscale_dev, dscale.data.size());
        dshift.data = handle.Read<U>(dshift_dev, dshift.data.size());

#if(MIO_BN_TIME_EVERYTHING == 1)
        auto t_end = std::chrono::high_resolution_clock::now();

        std::cout << "Wall clock: GPU backward_3d_bn_spatial_recalc pass time: "
                  << std::chrono::duration<double>(t_end - t_start).count() << " seconds."
                  << std::endl;
#endif
        return std::make_tuple(dx_out, dscale, dshift);
    }

    void fail(int badtensor) const
    {
        std::cout << "Backward Batch Spatial Normalization Recalc Mean and Variance: " << std::endl;
        std::cout << "X Input tensor: " << x_input.desc.ToString() << std::endl;
        std::cout << "Delta Y Input tensor: " << dy_input.desc.ToString() << std::endl;
        switch(badtensor)
        {
        case(0):
            std::cout << "Delta X output tensor output failed verification." << std::endl;
            break;
        case(1): std::cout << "Delta scale output tensor failed verification." << std::endl; break;
        case(2): std::cout << "Delta shift output tensor failed verification." << std::endl; break;
        default: break;
        }
    }
};

template <class T, class U>
struct verify_backward_3d_bn_spatial_use_saved
{
    const tensor<T> x_input;
    const tensor<T> dy_input;
    const tensor<U> scale;
    const tensor<U> savedMean;
    const tensor<U> savedInvVar;

    std::tuple<tensor<T>, tensor<U>, tensor<U>> cpu() const
    {
#if(MIO_BN_TIME_EVERYTHING == 1)
        auto t_start = std::chrono::high_resolution_clock::now();
#endif
        size_t n_batch, channels, depth, height, width;
        std::tie(n_batch, channels, depth, height, width) =
            miopen::tien<5>(x_input.desc.GetLengths());

        auto dx_out = tensor<T>{n_batch, channels, depth, height, width};
        std::fill(dx_out.begin(), dx_out.end(), 0);

        size_t ss_n_batch, ss_channels, ss_depth, ss_height, ss_width;
        auto derivedBnDesc = miopen::TensorDescriptor{};
        miopen::DeriveBNTensorDescriptor(derivedBnDesc, x_input.desc, miopenBNSpatial);
        std::tie(ss_n_batch, ss_channels, ss_depth, ss_height, ss_width) =
            miopen::tien<5>(derivedBnDesc.GetLengths());

        auto dscale = tensor<U>{ss_n_batch, ss_channels, ss_depth, ss_height, ss_width};
        std::fill(dscale.begin(), dscale.end(), 0);

        auto dshift = tensor<U>{ss_n_batch, ss_channels, ss_depth, ss_height, ss_width};
        std::fill(dshift.begin(), dshift.end(), 0);

        const unsigned int in_dstride = height * width;
        const unsigned int in_cstride = depth * in_dstride;
        const auto ndhw               = double(in_cstride * n_batch);

        miopen::par_for(channels, 1, [&](int cidx) {
            double elemStd = 0.;
            unsigned int xhat_index;
            double mean   = savedMean(0, cidx, 0, 0, 0);   // HxW elements
            double invVar = savedInvVar(0, cidx, 0, 0, 0); // HxW elements
            double dyelem = 0.;

            std::vector<double> xhat(n_batch * in_cstride, 0.0);

#if(MIO_HEIRARCH_SEL == 1)
            std::vector<double> dshift_accum_arr(height * depth, 0.0);
            std::vector<double> dscale_accum_arr(height * depth, 0.0);
#endif

            // process the batch per channel
            dscale(0, cidx, 0, 0, 0) = 0.;

#if(MIO_HEIRARCH_SEL == 0)
            miopen::ford(depth, height, width, n_batch)(
                [&](size_t didx, size_t row, size_t column, size_t bidx) {
                    xhat_index = in_cstride * bidx + in_dstride * didx + width * row + column;
                    // per (x-dims) channel load a block of data into LDS
                    elemStd = x_input(bidx, cidx, didx, row, column) - mean; // (x_i - mean)
                    xhat[xhat_index] = elemStd * invVar;
                    dyelem           = dy_input(bidx, cidx, didx, row, column);
                    dshift(0, cidx, 0, 0, 0) += dyelem;
                    dscale(0, cidx, 0, 0, 0) += xhat[xhat_index] * dyelem;
                });
#else
            miopen::ford(depth, height, width, n_batch)([&](size_t didx, size_t row, size_t column, size_t bidx) {
                xhat_index = in_cstride*bidx + in_dstride*didx + width*row + column;
                //per (x-dims) channel load a block of data into LDS
                elemStd             = x_input(bidx,cidx,didx,row,column) - mean;// (x_i - mean)
                xhat[xhat_index]    = elemStd*invVar;
                //printf("xhat[%d]: %lf\n",xhat_index,xhat[xhat_index]);
                dyelem              = dy_input(bidx,cidx,didx, row,column);
                dshift_accum_arr[row] += dyelem;
                dscale_accum_arr[row] += xhat[xhat_index]*dyelem;
                //dscale_accum_arr[row] += 1.0;//DEBUG
            });

            for(size_t i = 0; i<height; i++) {
                dshift(0,cidx,0,0,0) += dshift_accum_arr[i];
                dscale(0,cidx,0,0,0) += dscale_accum_arr[i];
            }
#endif
            miopen::ford(depth, height, width, n_batch)(
                [&](size_t didx, size_t row, size_t column, size_t bidx) {
                    xhat_index = in_cstride * bidx + in_dstride * didx + width * row + column;

                    const double tmp1 =
                        ndhw * dy_input(bidx, cidx, didx, row, column) - dshift(0, cidx, 0, 0, 0);
                    const double tmp2 = -xhat[xhat_index] * dscale(0, cidx, 0, 0, 0);
                    const double tmp3 = (scale(0, cidx, 0, 0, 0) * invVar) / ndhw;
                    dx_out(bidx, cidx, didx, row, column) = tmp3 * (tmp2 + tmp1);
                });
        }); // for (channel)
#if(MIO_BN_TIME_EVERYTHING == 1)
        auto t_end = std::chrono::high_resolution_clock::now();

        std::cout << "Wall clock: CPU backward_bn spatial_use_saved pass time: "
                  << std::chrono::duration<double>(t_end - t_start).count() << " seconds."
                  << std::endl;
#endif
        return std::make_tuple(dx_out, dscale, dshift);
    }

    std::tuple<tensor<T>, tensor<U>, tensor<U>> gpu() const
    {
#if(MIO_BN_TIME_EVERYTHING == 1)
        auto t_start = std::chrono::high_resolution_clock::now();
#endif
        auto&& handle = get_handle();

        size_t n_batch, channels, depth, height, width;
        std::tie(n_batch, channels, depth, height, width) =
            miopen::tien<5>(x_input.desc.GetLengths());

        auto dx_out = tensor<T>{n_batch, channels, depth, height, width};
        std::fill(dx_out.begin(), dx_out.end(), 0);

        size_t ss_n_batch, ss_channels, ss_depth, ss_height, ss_width;
        auto derivedBnDesc = miopen::TensorDescriptor{};
        miopen::DeriveBNTensorDescriptor(derivedBnDesc, x_input.desc, miopenBNSpatial);
        std::tie(ss_n_batch, ss_channels, ss_depth, ss_height, ss_width) =
            miopen::tien<5>(derivedBnDesc.GetLengths());

        auto dscale = tensor<U>{ss_n_batch, ss_channels, ss_depth, ss_height, ss_width};
        std::fill(dscale.begin(), dscale.end(), 0);

        auto dshift = tensor<U>{ss_n_batch, ss_channels, ss_depth, ss_height, ss_width};
        std::fill(dshift.begin(), dshift.end(), 0);

        float alpha = 1.0;
        float beta  = 0.0;

        auto xin_dev         = handle.Write(x_input.data);
        auto dyin_dev        = handle.Write(dy_input.data);
        auto scale_dev       = handle.Write(scale.data);
        auto dscale_dev      = handle.Write(dscale.data);
        auto dshift_dev      = handle.Write(dshift.data);
        auto dx_out_dev      = handle.Write(dx_out.data);
        auto savedMean_dev   = handle.Write(savedMean.data);
        auto savedInvVar_dev = handle.Write(savedInvVar.data);

        double epsilon = MIO_BN_TEST_EPSILON;

        miopen::ActivationDescriptor actDesc(miopenActivationPASTHRU, 0.0f, 0.0f, 0.0f);
        miopen::BatchNormBackward(handle,
                                  miopenBNSpatial,
                                  &alpha,
                                  &beta,
                                  &alpha,
                                  &beta,
                                  miopen::BuildReshaped4DTensorDescriptor(x_input.desc),
                                  xin_dev.get(),
                                  miopen::BuildReshaped4DTensorDescriptor(dy_input.desc),
                                  dyin_dev.get(),
                                  miopen::BuildReshaped4DTensorDescriptor(dx_out.desc),
                                  dx_out_dev.get(),
                                  miopen::BuildReshaped4DTensorDescriptor(scale.desc),
                                  miopen::BuildReshaped4DTensorDescriptor(dshift.desc),
                                  miopen::BuildReshaped4DTensorDescriptor(dshift.desc),
                                  miopen::BuildReshaped4DTensorDescriptor(dshift.desc),
                                  scale_dev.get(),
                                  nullptr,
                                  dscale_dev.get(),
                                  dshift_dev.get(),
                                  epsilon,
                                  savedMean_dev.get(),
                                  savedInvVar_dev.get(),
                                  actDesc);

        dx_out.data = handle.Read<T>(dx_out_dev, dx_out.data.size());
        dscale.data = handle.Read<U>(dscale_dev, dscale.data.size());
        dshift.data = handle.Read<U>(dshift_dev, dshift.data.size());

#if(MIO_BN_TIME_EVERYTHING == 1)
        auto t_end = std::chrono::high_resolution_clock::now();

        std::cout << "Wall clock: GPU backward_3d_bn_spatial_use_saved pass time: "
                  << std::chrono::duration<double>(t_end - t_start).count() << " seconds."
                  << std::endl;
#endif

        return std::make_tuple(dx_out, dscale, dshift);
    }

    void fail(int badtensor) const
    {
        std::cout << "Backward Batch Spatial Normalization Use Saved Mean and Variance: "
                  << std::endl;
        std::cout << "X Input tensor: " << x_input.desc.ToString() << std::endl;
        std::cout << "Delta Y Input tensor: " << dy_input.desc.ToString() << std::endl;
        switch(badtensor)
        {
        case(0):
            std::cout << "Delta X output tensor output failed verification." << std::endl;
            break;
        case(1): std::cout << "Delta scale output tensor failed verification." << std::endl; break;
        case(2): std::cout << "Delta shift output tensor failed verification." << std::endl; break;
        default: break;
        }
    }
};

//====== DRIVERS ===========================================

inline auto GenSmokeTestCases()
{
    return testing::Values(NamedContainer<std::vector<int>>{"dims", {16, 32, 8, 8, 8}, "x"});
}

inline auto GetSmokeTestCases()
{
    static const auto cases = GenSmokeTestCases();
    return cases;
}

inline auto GenFullTestCases()
{
    return MakeNamedParameterCollectionValues<std::vector<int>>(
        "dims", get_3d_bn_spatial_inputs(4), "x");
}

inline auto GetFullTestCases()
{
    static const auto cases = GenFullTestCases();
    return cases;
}

} // namespace

template <typename T>
struct Bn3DSpatialTest : public testing::TestWithParam<TestCase>
{
    static const constexpr uint64_t MaxValue{miopen_type<T>{} == miopenHalf ? 5 : 17};

    void SetUp() override
    {
        prng::reset_seed();
        const std::vector<int> dims = GetParam();
        input                       = tensor<T>{dims}.generate(tensor_elem_gen_integer{MaxValue});
        tolerance = 4e-3 / std::numeric_limits<T>::epsilon(); // ck solver has tolerance of 4e-3
    }

    void Run()
    {
#if(MIO_BN_TIME_EVERYTHING == 1)
        const auto t_start = std::chrono::high_resolution_clock::now();
#endif
        tensor<PREC_TYPE> scale;
        tensor<PREC_TYPE> shift;

        size_t n, c, d, h, w;
        std::tie(n, c, d, h, w) = miopen::tien<5>(input.desc.GetLengths());

        // The condition is derived form bn_spatial_test.cpp as they are known failures
        if(n == 1)
        {
            std::cout << "(n=1) is not supported for BN operation." << std::endl;
            return;
        }

        if((h * w * d > 1024) && (input.desc.GetType() == miopenHalf) && (MIO_BN_USE_MIX_PREC == 0))
        {
            std::cout << "(h*w*d > 1024) is not supported for BN operations "
                      << "when half precision is used but mixed precision disabled." << std::endl;
            return;
        }

        size_t ssn, ssc, ssd, ssh, ssw;
        auto derivedBnDesc = miopen::TensorDescriptor{};
        miopen::DeriveBNTensorDescriptor(derivedBnDesc, input.desc, miopenBNSpatial);
        std::tie(ssn, ssc, ssd, ssh, ssw) = miopen::tien<5>(derivedBnDesc.GetLengths());

        scale                   = tensor<PREC_TYPE>{ssn, ssc, ssd, ssh, ssw};
        shift                   = tensor<PREC_TYPE>{ssn, ssc, ssd, ssh, ssw};
        const double data_scale = 1e-4;

        for(size_t i = 0; i < scale.desc.GetElementSize(); i++)
        {
            scale[i] = prng::gen_descreet_uniform_sign<PREC_TYPE>(data_scale, 100);
            shift[i] = prng::gen_descreet_uniform_sign<PREC_TYPE>(data_scale, 100);
        }

        for(size_t i = 0; i < input.desc.GetElementSize(); i++)
        {
            input[i] = prng::gen_descreet_uniform_sign<T>(1e-5, 100);
        }

// train
#if(MIO_BN_SP_TEST_DEBUG == 1)
        std::cout << "Running forward train spatial with R and S set." << std::endl;
#endif
        const auto outpair =
            Verify(verify_forward_train_3d_bn_spatial<T, PREC_TYPE>{input, scale, shift});
// returns:  std::make_tuple(out,runMean,runVar,saveMean,saveInvVar);

// inference recalc
#if(MIO_BN_SP_TEST_DEBUG == 1)
        std::cout << "Running forward inference spatial recalc." << std::endl;
#endif
        // tolerance = 80;
        // Debug values
        // std::fill(input.begin(), input.end(), 1);
        // std::fill(scale.begin(), scale.end(), 1);
        // std::fill(shift.begin(), shift.end(), 1);
        Verify(verify_forward_infer_3d_bn_spatial_recalc<T, PREC_TYPE>{input, scale, shift});

        // inference use estimated running values
        const auto estMean = std::get<1>(outpair.second);
        const auto estVar  = std::get<2>(outpair.second);
#if(MIO_BN_SP_TEST_DEBUG == 1)
        std::cout << "Running forward inference spatial with R set." << std::endl;
#endif
        Verify(verify_forward_infer_3d_bn_spatial_use_est<T, PREC_TYPE>{
            input, scale, shift, estMean, estVar});

        // backprop recalc
        auto dy_input = std::get<0>(outpair.second);
        miopen::ford(n, c, d, h, w)(
            [&](size_t bidx, size_t cidx, size_t didx, size_t row, size_t column) {
                dy_input(bidx, cidx, didx, row, column) *= 0.1;
            });
#if(MIO_BN_SP_TEST_DEBUG == 2)
        auto debugvals = Verify(verify_backward_3d_bn_spatial_recalc<T>{input, dy_input, scale});
        auto gpuout    = std::get<0>(debugvals.second);
        auto cpuout    = std::get<0>(debugvals.first);

        double maxdiff = 0.;
        int mn         = 0;
        int mc         = 0;
        int md         = 0;
        int mh         = 0;
        int mw         = 0;

        miopen::ford(
            n, c, d, h, w)([&](size_t bidx, size_t cidx, size_t didx, size_t row, size_t column) {
            const double diff =
                fabs(gpuout(bidx, cidx, didx, row, column) - cpuout(bidx, cidx, didx, row, column));
            if(diff > maxdiff)
            {
                maxdiff = diff;
                mn      = bidx;
                mc      = cidx;
                md      = didx;
                mh      = row;
                mw      = column;
            }
            // if(diff > 1.)
            // {
            std::cout << "gpu[" << bidx << ", " << cidx << ", " << didx << ", " << row << ", "
                      << column << "]: " << gpuout(bidx, cidx, didx, row, column) << " :: ";
            std::cout << "cpu[" << bidx << ", " << cidx << ", " << didx << ", " << row << ", "
                      << column << "]: " << cpuout(bidx, cidx, didx, row, column) << " :: ";
            std::cout << "diff: " << diff << std::endl;
            //    }
        });

        if(maxdiff > 0)
        {
            std::cout << "Max diff: " << maxdiff << std::endl;
            std::cout << "gpu[" << mn << ", " << mc << ", " << md << ", " << mh << ", " << mw
                      << "]: " << gpuout(mn, mc, md, mh, mw) << " :: ";
            std::cout << "cpu[" << mn << ", " << mc << ", " << md << ", " << mh << ", " << mw
                      << "]: " << cpuout(mn, mc, md, mh, mw) << std::endl;
        }
#else
#if(MIO_BN_SP_TEST_DEBUG == 1)
        std::cout << "Running back propagation spatial recalc." << std::endl;
#endif
        tolerance = 80 * input.desc.GetElementSize();
        Verify(verify_backward_3d_bn_spatial_recalc<T, PREC_TYPE>{input, dy_input, scale});
#endif

        // backprop use saved values
        const auto savedMean   = std::get<3>(outpair.second);
        const auto savedInvVar = std::get<4>(outpair.second);

#if(MIO_BN_SP_TEST_DEBUG == 3)

        auto debugvals = Verify(verify_backward_3d_bn_spatial_use_saved<T>{
            input, dy_input, scale, savedMean, savedInvVar});
        auto gpuout    = std::get<0>(debugvals.second);
        auto cpuout    = std::get<0>(debugvals.first);

        double maxdiff = 0.;
        int mn         = 0;
        int mc         = 0;
        int md         = 0;
        int mh         = 0;
        int mw         = 0;

        miopen::ford(
            n, c, d, h, w)([&](size_t bidx, size_t cidx, size_t didx, size_t row, size_t column) {
            const double diff =
                fabs(gpuout(bidx, cidx, didx, row, column) - cpuout(bidx, cidx, didx, row, column));
            if(diff > maxdiff)
            {
                maxdiff = diff;
                mn      = bidx;
                mc      = cidx;
                md      = didx;
                mh      = row;
                mw      = column;
            }
            // if(diff > 1.)
            //{
            std::cout << "gpu[" << bidx << ", " << cidx << ", " << didx << ", " << row << ", "
                      << column << "]: " << gpuout(bidx, cidx, didx, row, column) << " :: ";
            std::cout << "cpu[" << bidx << ", " << cidx << ", " << didx << ", " << row << ", "
                      << column << "]: " << cpuout(bidx, cidx, didx, row, column) << " :: ";
            std::cout << "diff: " << diff << std::endl;
            //}
        });

        if(maxdiff > 0)
        {
            std::cout << "Max diff: " << maxdiff << std::endl;
            std::cout << "gpu[" << mn << ", " << mc << ", " << md << ", " << mh << ", " << mw
                      << "]: " << gpuout(mn, mc, md, mh, mw) << " :: ";
            std::cout << "cpu[" << mn << ", " << mc << ", " << md << ", " << mh << ", " << mw
                      << "]: " << cpuout(mn, mc, md, mh, mw) << std::endl;
        }
#else
#if(MIO_BN_SP_TEST_DEBUG == 1)
        std::cout << "Running back propagation spatial with S set." << std::endl;
#endif
        Verify(verify_backward_3d_bn_spatial_use_saved<T, PREC_TYPE>{
            input, dy_input, scale, savedMean, savedInvVar});
#endif

#if(MIO_BN_TIME_EVERYTHING == 1)
        const auto t_end = std::chrono::high_resolution_clock::now();
        std::cout << "Wall clock: full PER_ACTIVATION test pass time: "
                  << std::chrono::duration<double>(t_end - t_start).count() << " seconds."
                  << std::endl;
#endif
    }

private:
    tensor<T> input;
    double tolerance{0.0};

private:
    auto Verify(auto&& v) -> decltype(std::make_pair(v.cpu(), v.gpu()))
    {
        const auto cpu = v.cpu();
        const auto gpu = v.gpu();

        Compare(v, cpu, gpu);

        return std::make_pair(cpu, gpu);
    }

    template <typename... CpuRanges, typename... GpuRanges>
    void Compare(auto&& v, const std::tuple<CpuRanges...>& cpu, const std::tuple<GpuRanges...>& gpu)
    {
        static_assert(sizeof...(CpuRanges) == sizeof...(GpuRanges), "CPU and GPU mismatch");

        miopen::sequence([&](auto... is) {
            miopen::each_args(
                [&](auto i) {
                    const auto& c = std::get<i>(cpu);
                    const auto& g = std::get<i>(gpu);

                    ASSERT_EQ(miopen::range_distance(c), miopen::range_distance(g));

                    using value_type = miopen::range_value<decltype(g)>;

                    const double threshold = std::numeric_limits<value_type>::epsilon() * tolerance;
                    const double error     = miopen::rms_range(c, g);

                    EXPECT_LE(error, threshold);

                    if(error > threshold)
                    {
                        v.fail(i);
                    }
                },
                is...);
        })(std::integral_constant<size_t, sizeof...(CpuRanges)>{});
    }

    template <typename CpuRanges, typename GpuRanges>
    void Compare(auto&& v, const CpuRanges& cpu, const GpuRanges& gpu)
    {
        ASSERT_EQ(miopen::range_distance(cpu), miopen::range_distance(gpu));

        using value_type = miopen::range_value<decltype(gpu)>;

        const double threshold = std::numeric_limits<value_type>::epsilon() * tolerance;
        const double error     = miopen::rms_range(cpu, gpu);

        EXPECT_LE(error, threshold);

        if(error > threshold)
        {
            v.fail(0);
        }
    }
};

struct TestNameGenerator
{
    std::string operator()(const auto& info)
    {
        const auto& dims = info.param;
        std::stringstream ss;
        std::string str;

        ss << "dims_" << GetRangeAsString(dims(), "x") << "_test_id_" << info.index;

        str = ss.str();

        // Name format only supports letters, numbers and underscores.
        std::transform(str.begin(), str.end(), str.begin(), [](char c) {
            return (c == '.') ? 'p' : (std::isalnum(c) ? c : '_');
        });

        return str;
    }
};

using GPU_Bn3DSpatial_FP32 = Bn3DSpatialTest<float>;

TEST_P(GPU_Bn3DSpatial_FP32, TestFloat32) { this->Run(); }

INSTANTIATE_TEST_SUITE_P(Smoke, GPU_Bn3DSpatial_FP32, GetSmokeTestCases(), TestNameGenerator{});
INSTANTIATE_TEST_SUITE_P(Full, GPU_Bn3DSpatial_FP32, GetFullTestCases(), TestNameGenerator{});