/* ************************************************************************
 * Copyright 2016-2021 Advanced Micro Devices, Inc.
 *
 * ************************************************************************ */

#include <stdio.h>
#include <stdlib.h>
#include <vector>

#include "testing_common.hpp"

using namespace std;

/* ============================================================================================ */

template <typename Ta, typename Tx = Ta, typename Ty = Tx>
hipblasStatus_t testing_axpy_strided_batched_ex_template(const Arguments& argus)
{
    bool FORTRAN = argus.fortran;
    auto hipblasAxpyStridedBatchedExFn
        = FORTRAN ? hipblasAxpyStridedBatchedExFortran : hipblasAxpyStridedBatchedEx;
    hipblasStatus_t status = HIPBLAS_STATUS_SUCCESS;

    int    N            = argus.N;
    int    incx         = argus.incx;
    int    incy         = argus.incy;
    double stride_scale = argus.stride_scale;
    int    batch_count  = argus.batch_count;

    int abs_incx = incx < 0 ? -incx : incx;
    int abs_incy = incy < 0 ? -incy : incy;

    hipblasStride stridex = size_t(N) * abs_incx * stride_scale;
    hipblasStride stridey = size_t(N) * abs_incy * stride_scale;

    hipblasDatatype_t alphaType     = argus.a_type;
    hipblasDatatype_t xType         = argus.b_type;
    hipblasDatatype_t yType         = argus.c_type;
    hipblasDatatype_t executionType = argus.compute_type;

    hipblasLocalHandle handle(argus);

    // argument sanity check, quick return if input parameters are invalid before allocating invalid
    // memory
    if(N <= 0 || batch_count <= 0)
    {
        CHECK_HIPBLAS_ERROR(hipblasAxpyStridedBatchedExFn(handle,
                                                          N,
                                                          nullptr,
                                                          alphaType,
                                                          nullptr,
                                                          xType,
                                                          incx,
                                                          stridex,
                                                          nullptr,
                                                          yType,
                                                          incy,
                                                          stridey,
                                                          batch_count,
                                                          executionType));
        return HIPBLAS_STATUS_SUCCESS;
    }

    size_t sizeX = stridex * batch_count;
    size_t sizeY = stridey * batch_count;
    if(!sizeX)
        sizeX = 1;
    if(!sizeY)
        sizeY = 1;

    Ta h_alpha = argus.get_alpha<Ta>();

    // Naming: dX is in GPU (device) memory. hK is in CPU (host) memory, plz follow this practice
    host_vector<Tx> hx(sizeX);
    host_vector<Ty> hy_cpu(sizeY);
    host_vector<Ty> hy_host(sizeY);
    host_vector<Ty> hy_device(sizeY);

    device_vector<Tx> dx(sizeX);
    device_vector<Ty> dy(sizeY);
    device_vector<Ta> d_alpha(1);

    double gpu_time_used, hipblas_error_host, hipblas_error_device;

    // Initial Data on CPU
    srand(1);
    hipblas_init<Tx>(hx, 1, N, abs_incx, stridex, batch_count);
    hipblas_init<Ty>(hy_host, 1, N, abs_incy, stridey, batch_count);

    hy_device = hy_host;
    hy_cpu    = hy_host;

    CHECK_HIP_ERROR(hipMemcpy(dx, hx, sizeof(Tx) * sizeX, hipMemcpyHostToDevice));
    CHECK_HIP_ERROR(hipMemcpy(dy, hy_host, sizeof(Ty) * sizeY, hipMemcpyHostToDevice));
    CHECK_HIP_ERROR(hipMemcpy(d_alpha, &h_alpha, sizeof(Ta), hipMemcpyHostToDevice));

    /* =====================================================================
         HIPBLAS
    =================================================================== */
    CHECK_HIPBLAS_ERROR(hipblasSetPointerMode(handle, HIPBLAS_POINTER_MODE_HOST));
    CHECK_HIPBLAS_ERROR(hipblasAxpyStridedBatchedExFn(handle,
                                                      N,
                                                      &h_alpha,
                                                      alphaType,
                                                      dx,
                                                      xType,
                                                      incx,
                                                      stridex,
                                                      dy,
                                                      yType,
                                                      incy,
                                                      stridey,
                                                      batch_count,
                                                      executionType));

    CHECK_HIP_ERROR(hipMemcpy(hy_host, dy, sizeof(Ty) * sizeY, hipMemcpyDeviceToHost));
    CHECK_HIP_ERROR(hipMemcpy(dy, hy_device, sizeof(Ty) * sizeY, hipMemcpyHostToDevice));

    CHECK_HIPBLAS_ERROR(hipblasSetPointerMode(handle, HIPBLAS_POINTER_MODE_DEVICE));
    CHECK_HIPBLAS_ERROR(hipblasAxpyStridedBatchedExFn(handle,
                                                      N,
                                                      d_alpha,
                                                      alphaType,
                                                      dx,
                                                      xType,
                                                      incx,
                                                      stridex,
                                                      dy,
                                                      yType,
                                                      incy,
                                                      stridey,
                                                      batch_count,
                                                      executionType));

    CHECK_HIP_ERROR(hipMemcpy(hy_device, dy, sizeof(Ty) * sizeY, hipMemcpyDeviceToHost));

    if(argus.unit_check || argus.norm_check)
    {
        /* =====================================================================
                    CPU BLAS
        =================================================================== */
        for(int b = 0; b < batch_count; b++)
        {
            cblas_axpy(N, h_alpha, hx.data() + b * stridex, incx, hy_cpu + b * stridey, incy);
        }

        // enable unit check, notice unit check is not invasive, but norm check is,
        // unit check and norm check can not be interchanged their order
        if(argus.unit_check)
        {
            unit_check_general<Ty>(1, N, batch_count, abs_incy, stridey, hy_cpu, hy_host);
            unit_check_general<Ty>(1, N, batch_count, abs_incy, stridey, hy_cpu, hy_device);
        }
        if(argus.norm_check)
        {
            hipblas_error_host = norm_check_general<Ty>(
                'F', 1, N, abs_incy, stridey, hy_cpu, hy_host, batch_count);
            hipblas_error_device = norm_check_general<Ty>(
                'F', 1, N, abs_incy, stridey, hy_cpu, hy_device, batch_count);
        }

    } // end of if unit check

    if(argus.timing)
    {
        hipStream_t stream;
        CHECK_HIPBLAS_ERROR(hipblasGetStream(handle, &stream));
        CHECK_HIPBLAS_ERROR(hipblasSetPointerMode(handle, HIPBLAS_POINTER_MODE_DEVICE));

        int runs = argus.cold_iters + argus.iters;
        for(int iter = 0; iter < runs; iter++)
        {
            if(iter == argus.cold_iters)
                gpu_time_used = get_time_us_sync(stream);

            CHECK_HIPBLAS_ERROR(hipblasAxpyStridedBatchedExFn(handle,
                                                              N,
                                                              d_alpha,
                                                              alphaType,
                                                              dx,
                                                              xType,
                                                              incx,
                                                              stridex,
                                                              dy,
                                                              yType,
                                                              incy,
                                                              stridey,
                                                              batch_count,
                                                              executionType));
        }
        gpu_time_used = get_time_us_sync(stream) - gpu_time_used;

        ArgumentModel<e_N, e_incx, e_stride_x, e_incy, e_stride_y, e_batch_count>{}.log_args<Ta>(
            std::cout,
            argus,
            gpu_time_used,
            axpy_gflop_count<Ta>(N),
            axpy_gbyte_count<Ta>(N),
            hipblas_error_host,
            hipblas_error_device);
    }

    return HIPBLAS_STATUS_SUCCESS;
}

hipblasStatus_t testing_axpy_strided_batched_ex(Arguments argus)
{
    hipblasDatatype_t alphaType     = argus.a_type;
    hipblasDatatype_t xType         = argus.b_type;
    hipblasDatatype_t yType         = argus.c_type;
    hipblasDatatype_t executionType = argus.compute_type;

    hipblasStatus_t status = HIPBLAS_STATUS_SUCCESS;

    if(alphaType == HIPBLAS_R_16F && xType == HIPBLAS_R_16F && yType == HIPBLAS_R_16F
       && executionType == HIPBLAS_R_16F)
    {
        status = testing_axpy_strided_batched_ex_template<hipblasHalf>(argus);
    }
    else if(alphaType == HIPBLAS_R_16F && xType == HIPBLAS_R_16F && yType == HIPBLAS_R_16F
            && executionType == HIPBLAS_R_32F)
    {
        // Not testing accumulation here
        status = testing_axpy_strided_batched_ex_template<hipblasHalf>(argus);
    }
    else if(alphaType == HIPBLAS_R_32F && xType == HIPBLAS_R_16F && yType == HIPBLAS_R_16F
            && executionType == HIPBLAS_R_32F)
    {
        // Not testing accumulation here
        status = testing_axpy_strided_batched_ex_template<float, hipblasHalf>(argus);
    }
    else if(alphaType == HIPBLAS_R_32F && xType == HIPBLAS_R_32F && yType == HIPBLAS_R_32F
            && executionType == HIPBLAS_R_32F)
    {
        status = testing_axpy_strided_batched_ex_template<float>(argus);
    }
    else if(alphaType == HIPBLAS_R_64F && xType == HIPBLAS_R_64F && yType == HIPBLAS_R_64F
            && executionType == HIPBLAS_R_64F)
    {
        status = testing_axpy_strided_batched_ex_template<double>(argus);
    }
    else if(alphaType == HIPBLAS_C_32F && xType == HIPBLAS_C_32F && yType == HIPBLAS_C_32F
            && executionType == HIPBLAS_C_32F)
    {
        status = testing_axpy_strided_batched_ex_template<hipblasComplex>(argus);
    }
    else if(alphaType == HIPBLAS_C_64F && xType == HIPBLAS_C_64F && yType == HIPBLAS_C_64F
            && executionType == HIPBLAS_C_64F)
    {
        status = testing_axpy_strided_batched_ex_template<hipblasDoubleComplex>(argus);
    }
    else
    {
        status = HIPBLAS_STATUS_NOT_SUPPORTED;
    }

    return status;
}