/* ************************************************************************
 * Copyright 2018-2021 Advanced Micro Devices, Inc.
 *
 * ************************************************************************ */

#pragma once

#include "bytes.hpp"
#include "cblas_interface.hpp"
#include "flops.hpp"
#include "near.hpp"
#include "norm.hpp"
#include "rocblas.hpp"
#include "rocblas_datatype2string.hpp"
#include "rocblas_init.hpp"
#include "rocblas_math.hpp"
#include "rocblas_random.hpp"
#include "rocblas_test.hpp"
#include "rocblas_vector.hpp"
#include "unit.hpp"
#include "utility.hpp"

template <typename T>
void testing_tbmv_batched_bad_arg(const Arguments& arg)
{
    auto rocblas_tbmv_batched_fn
        = arg.fortran ? rocblas_tbmv_batched<T, true> : rocblas_tbmv_batched<T, false>;

    const rocblas_int M           = 100;
    const rocblas_int K           = 5;
    const rocblas_int lda         = 100;
    const rocblas_int incx        = 1;
    const rocblas_int batch_count = 5;

    const rocblas_fill      uplo   = rocblas_fill_upper;
    const rocblas_operation transA = rocblas_operation_none;
    const rocblas_diagonal  diag   = rocblas_diagonal_non_unit;

    rocblas_local_handle handle{arg};

    size_t size_A = lda * size_t(M);
    size_t size_x = M * size_t(incx);

    // allocate memory on device
    device_batch_vector<T> dA(size_A, 1, batch_count);
    device_batch_vector<T> dx(M, incx, batch_count);
    CHECK_DEVICE_ALLOCATION(dA.memcheck());
    CHECK_DEVICE_ALLOCATION(dx.memcheck());

    EXPECT_ROCBLAS_STATUS(
        rocblas_tbmv_batched_fn(
            handle, uplo, transA, diag, M, K, nullptr, lda, dx.ptr_on_device(), incx, batch_count),
        rocblas_status_invalid_pointer);

    EXPECT_ROCBLAS_STATUS(
        rocblas_tbmv_batched_fn(
            handle, uplo, transA, diag, M, K, dA.ptr_on_device(), lda, nullptr, incx, batch_count),
        rocblas_status_invalid_pointer);

    EXPECT_ROCBLAS_STATUS(rocblas_tbmv_batched_fn(nullptr,
                                                  uplo,
                                                  transA,
                                                  diag,
                                                  M,
                                                  K,
                                                  dA.ptr_on_device(),
                                                  lda,
                                                  dx.ptr_on_device(),
                                                  incx,
                                                  batch_count),
                          rocblas_status_invalid_handle);

    // Adding test to check that if batch_count == 0 we can pass in nullptrs and get a success.
    EXPECT_ROCBLAS_STATUS(
        rocblas_tbmv_batched_fn(handle, uplo, transA, diag, M, K, nullptr, lda, nullptr, incx, 0),
        rocblas_status_success);
}

template <typename T>
void testing_tbmv_batched(const Arguments& arg)
{
    auto rocblas_tbmv_batched_fn
        = arg.fortran ? rocblas_tbmv_batched<T, true> : rocblas_tbmv_batched<T, false>;

    rocblas_int       M           = arg.M;
    rocblas_int       K           = arg.K;
    rocblas_int       lda         = arg.lda;
    rocblas_int       incx        = arg.incx;
    char              char_uplo   = arg.uplo;
    char              char_diag   = arg.diag;
    rocblas_fill      uplo        = char2rocblas_fill(char_uplo);
    rocblas_operation transA      = char2rocblas_operation(arg.transA);
    rocblas_diagonal  diag        = char2rocblas_diagonal(char_diag);
    rocblas_int       batch_count = arg.batch_count;

    rocblas_local_handle handle{arg};

    // argument sanity check before allocating invalid memory
    bool invalid_size = M < 0 || K < 0 || lda < K + 1 || !incx || batch_count < 0;
    if(invalid_size || !M || !batch_count)
    {
        EXPECT_ROCBLAS_STATUS(
            rocblas_tbmv_batched_fn(
                handle, uplo, transA, diag, M, K, nullptr, lda, nullptr, incx, batch_count),
            invalid_size ? rocblas_status_invalid_size : rocblas_status_success);

        return;
    }

    size_t size_A = lda * size_t(M);
    size_t size_x, abs_incx;

    abs_incx = incx >= 0 ? incx : -incx;
    size_x   = M * abs_incx;

    // Naming: dK is in GPU (device) memory. hK is in CPU (host) memory
    host_batch_vector<T> hA(size_A, 1, batch_count);
    host_batch_vector<T> hx(M, incx, batch_count);
    host_batch_vector<T> hx_1(M, incx, batch_count);
    host_batch_vector<T> hx_gold(M, incx, batch_count);
    CHECK_HIP_ERROR(hA.memcheck());
    CHECK_HIP_ERROR(hx.memcheck());
    CHECK_HIP_ERROR(hx_1.memcheck());
    CHECK_HIP_ERROR(hx_gold.memcheck());

    device_batch_vector<T> dA(size_A, 1, batch_count);
    device_batch_vector<T> dx(M, incx, batch_count);
    CHECK_DEVICE_ALLOCATION(dx.memcheck());
    CHECK_DEVICE_ALLOCATION(dA.memcheck());

    // Initialize data on host memory
    rocblas_init_vector(hA, arg, rocblas_client_never_set_nan, true);
    rocblas_init_vector(hx, arg, rocblas_client_never_set_nan, false, true);

    hx_gold.copy_from(hx);
    CHECK_HIP_ERROR(dA.transfer_from(hA));
    CHECK_HIP_ERROR(dx.transfer_from(hx));

    double gpu_time_used, cpu_time_used;
    double rocblas_error_1;
    double rocblas_error_2;

    /* =====================================================================
           ROCBLAS
    =================================================================== */

    if(arg.unit_check || arg.norm_check)
    {
        // pointer mode shouldn't matter here
        CHECK_ROCBLAS_ERROR(rocblas_set_pointer_mode(handle, rocblas_pointer_mode_device));
        CHECK_ROCBLAS_ERROR(rocblas_tbmv_batched_fn(handle,
                                                    uplo,
                                                    transA,
                                                    diag,
                                                    M,
                                                    K,
                                                    dA.ptr_on_device(),
                                                    lda,
                                                    dx.ptr_on_device(),
                                                    incx,
                                                    batch_count));

        // CPU BLAS
        cpu_time_used = get_time_us_no_sync();
        for(int b = 0; b < batch_count; b++)
            cblas_tbmv<T>(uplo, transA, diag, M, K, hA[b], lda, hx_gold[b], incx);
        cpu_time_used = get_time_us_no_sync() - cpu_time_used;

        // copy output from device to CPU
        CHECK_HIP_ERROR(hx_1.transfer_from(dx));

        if(arg.unit_check)
        {
            unit_check_general<T>(1, M, abs_incx, hx_gold, hx_1, batch_count);
        }

        if(arg.norm_check)
        {
            rocblas_error_1
                = norm_check_general<T>('F', 1, M, abs_incx, hx_gold, hx_1, batch_count);
        }
    }

    if(arg.timing)
    {
        int number_cold_calls = arg.cold_iters;
        int number_hot_calls  = arg.iters;

        for(int iter = 0; iter < number_cold_calls; iter++)
        {
            rocblas_tbmv_batched_fn(handle,
                                    uplo,
                                    transA,
                                    diag,
                                    M,
                                    K,
                                    dA.ptr_on_device(),
                                    lda,
                                    dx.ptr_on_device(),
                                    incx,
                                    batch_count);
        }

        hipStream_t stream;
        CHECK_ROCBLAS_ERROR(rocblas_get_stream(handle, &stream));
        gpu_time_used = get_time_us_sync(stream); // in microseconds

        for(int iter = 0; iter < number_hot_calls; iter++)
        {
            rocblas_tbmv_batched_fn(handle,
                                    uplo,
                                    transA,
                                    diag,
                                    M,
                                    K,
                                    dA.ptr_on_device(),
                                    lda,
                                    dx.ptr_on_device(),
                                    incx,
                                    batch_count);
        }

        gpu_time_used = get_time_us_sync(stream) - gpu_time_used;

        ArgumentModel<e_uplo, e_transA, e_diag, e_M, e_K, e_lda, e_incx, e_batch_count>{}
            .log_args<T>(rocblas_cout,
                         arg,
                         gpu_time_used,
                         tbmv_gflop_count<T>(M, K),
                         tbmv_gbyte_count<T>(M, K),
                         cpu_time_used,
                         rocblas_error_1);
    }
}