/* ************************************************************************
 * Copyright (C) 2016-2025 Advanced Micro Devices, Inc. All rights reserved.
 *
 * 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 <fstream>
#include <iostream>
#include <stdlib.h>
#include <vector>

#include "testing_common.hpp"

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

using hipblasGemvModel
    = ArgumentModel<e_a_type, e_transA, e_M, e_N, e_alpha, e_lda, e_incx, e_beta, e_incy>;

inline void testname_gemv(const Arguments& arg, std::string& name)
{
    hipblasGemvModel{}.test_name(arg, name);
}

template <typename T>
void testing_gemv_bad_arg(const Arguments& arg)
{
    using Ts           = hipblas_internal_type<T>;
    bool FORTRAN       = arg.api == hipblas_client_api::FORTRAN;
    auto hipblasGemvFn = FORTRAN ? hipblasGemv<T, true> : hipblasGemv<T, false>;
    auto hipblasGemvFn_64
        = arg.api == FORTRAN_64 ? hipblasGemv_64<T, true> : hipblasGemv_64<T, false>;

    for(auto pointer_mode : {HIPBLAS_POINTER_MODE_HOST, HIPBLAS_POINTER_MODE_DEVICE})
    {
        hipblasLocalHandle handle(arg);
        CHECK_HIPBLAS_ERROR(hipblasSetPointerMode(handle, pointer_mode));

        hipblasOperation_t transA = HIPBLAS_OP_N;
        int64_t            N      = 100;
        int64_t            M      = 100;
        int64_t            lda    = 100;
        int64_t            incx   = 1;
        int64_t            incy   = 1;

        device_vector<T> d_alpha(1), d_beta(1), d_one(1), d_zero(1);

        Ts h_alpha{1}, h_beta{2}, h_one{1}, h_zero{0};
        if constexpr(is_complex<T>)
            h_one = {1, 0};

        const Ts* alpha = &h_alpha;
        const Ts* beta  = &h_beta;
        const Ts* one   = &h_one;
        const Ts* zero  = &h_zero;

        if(pointer_mode == HIPBLAS_POINTER_MODE_DEVICE)
        {
            CHECK_HIP_ERROR(hipMemcpy(d_alpha, alpha, sizeof(*alpha), hipMemcpyHostToDevice));
            CHECK_HIP_ERROR(hipMemcpy(d_beta, beta, sizeof(*beta), hipMemcpyHostToDevice));
            CHECK_HIP_ERROR(hipMemcpy(d_one, one, sizeof(*one), hipMemcpyHostToDevice));
            CHECK_HIP_ERROR(hipMemcpy(d_zero, zero, sizeof(*zero), hipMemcpyHostToDevice));
            alpha = d_alpha;
            beta  = d_beta;
            one   = d_one;
            zero  = d_zero;
        }

        device_matrix<T> dA(M, N, lda);
        device_vector<T> dx(N, incx);
        device_vector<T> dy(M, incy);

        DAPI_EXPECT(HIPBLAS_STATUS_NOT_INITIALIZED,
                    hipblasGemvFn,
                    (nullptr, transA, M, N, alpha, dA, lda, dx, incx, beta, dy, incy));

        DAPI_EXPECT(HIPBLAS_STATUS_INVALID_ENUM,
                    hipblasGemvFn,
                    (handle,
                     (hipblasOperation_t)HIPBLAS_FILL_MODE_FULL,
                     M,
                     N,
                     alpha,
                     dA,
                     lda,
                     dx,
                     incx,
                     beta,
                     dy,
                     incy));

        if(arg.bad_arg_all)
        {
            DAPI_EXPECT(HIPBLAS_STATUS_INVALID_VALUE,
                        hipblasGemvFn,
                        (handle, transA, M, N, nullptr, dA, lda, dx, incx, beta, dy, incy));

            DAPI_EXPECT(HIPBLAS_STATUS_INVALID_VALUE,
                        hipblasGemvFn,
                        (handle, transA, M, N, alpha, dA, lda, dx, incx, nullptr, dy, incy));

            if(pointer_mode == HIPBLAS_POINTER_MODE_HOST)
            {
                // For device mode in rocBLAS we don't have checks for dA, dx, dy as we may be able to quick return
                DAPI_EXPECT(HIPBLAS_STATUS_INVALID_VALUE,
                            hipblasGemvFn,
                            (handle, transA, M, N, alpha, nullptr, lda, dx, incx, beta, dy, incy));

                DAPI_EXPECT(HIPBLAS_STATUS_INVALID_VALUE,
                            hipblasGemvFn,
                            (handle, transA, M, N, alpha, dA, lda, nullptr, incx, beta, dy, incy));

                DAPI_EXPECT(HIPBLAS_STATUS_INVALID_VALUE,
                            hipblasGemvFn,
                            (handle, transA, M, N, alpha, dA, lda, dx, incx, beta, nullptr, incy));

                // rocBLAS implementation has alpha == 0 and beta == 1 quick return after arg checks, so if we're using 32-bit params,
                // this should fail with invalid-value as c_i32_overflow will rollover to -2147483648
                // Note: that this strategy can't check incx as rocBLAS supports negative. Also depends on implementation so not testing cuBLAS for now

                DAPI_EXPECT((arg.api & c_API_64) ? HIPBLAS_STATUS_SUCCESS
                                                 : HIPBLAS_STATUS_INVALID_VALUE,
                            hipblasGemvFn,
                            (handle,
                             transA,
                             c_i32_overflow,
                             c_i32_overflow,
                             zero,
                             nullptr,
                             c_i32_overflow + 1,
                             nullptr,
                             incx,
                             one,
                             nullptr,
                             incy));
            }

            // With alpha == 0 can have x nullptr
            DAPI_CHECK(hipblasGemvFn,
                       (handle, transA, M, N, zero, nullptr, lda, nullptr, incx, beta, dy, incy));

            // With alpha == 0 && beta == 1, all other ptrs can be nullptr
            DAPI_CHECK(
                hipblasGemvFn,
                (handle, transA, M, N, zero, nullptr, lda, nullptr, incx, one, nullptr, incy));
        }

        // With M == 0 || N == 0, can have all nullptrs
        DAPI_CHECK(
            hipblasGemvFn,
            (handle, transA, 0, N, nullptr, nullptr, lda, nullptr, incx, nullptr, nullptr, incy));
        DAPI_CHECK(
            hipblasGemvFn,
            (handle, transA, M, 0, nullptr, nullptr, lda, nullptr, incx, nullptr, nullptr, incy));
    }
}

template <typename T>
void testing_gemv(const Arguments& arg)
{
    using Ts           = hipblas_internal_type<T>;
    bool FORTRAN       = arg.api == hipblas_client_api::FORTRAN;
    auto hipblasGemvFn = FORTRAN ? hipblasGemv<T, true> : hipblasGemv<T, false>;
    auto hipblasGemvFn_64
        = arg.api == FORTRAN_64 ? hipblasGemv_64<T, true> : hipblasGemv_64<T, false>;

    int64_t M    = arg.M;
    int64_t N    = arg.N;
    int64_t lda  = arg.lda;
    int64_t incx = arg.incx;
    int64_t incy = arg.incy;

    size_t dim_x;
    size_t dim_y;

    hipblasOperation_t transA = char2hipblas_operation(arg.transA);

    if(transA == HIPBLAS_OP_N)
    {
        dim_x = N;
        dim_y = M;
    }
    else
    {
        dim_x = M;
        dim_y = N;
    }

    hipblasLocalHandle handle(arg);

    // argument sanity check, quick return if input parameters are invalid before allocating invalid
    // memory
    bool invalid_size = M < 0 || N < 0 || lda < M || lda < 1 || !incx || !incy;
    if(invalid_size || !M || !N)
    {
        if(!invalid_size || arg.bad_arg_all)
        {
            // cublas backend doesn't support nullptrs with bad input sizes
            DAPI_EXPECT(invalid_size ? HIPBLAS_STATUS_INVALID_VALUE : HIPBLAS_STATUS_SUCCESS,
                        hipblasGemvFn,
                        (handle,
                         transA,
                         M,
                         N,
                         nullptr,
                         nullptr,
                         lda,
                         nullptr,
                         incx,
                         nullptr,
                         nullptr,
                         incy));
        }
        return;
    }

    size_t abs_incx = incx >= 0 ? incx : -incx;
    size_t abs_incy = incy >= 0 ? incy : -incy;

    // Naming: dA is in GPU (device) memory. hA is in CPU (host) memory
    host_matrix<T> hA(M, N, lda);
    host_vector<T> hx(dim_x, incx);
    host_vector<T> hy(dim_y, incy);
    host_vector<T> hy_cpu(dim_y, incy);
    host_vector<T> hy_host(dim_y, incy);
    host_vector<T> hy_device(dim_y, incy);

    device_matrix<T> dA(M, N, lda);
    device_vector<T> dx(dim_x, incx);
    device_vector<T> dy(dim_y, incy);
    device_vector<T> d_alpha(1);
    device_vector<T> d_beta(1);

    // Check device memory allocation
    CHECK_DEVICE_ALLOCATION(dA.memcheck());
    CHECK_DEVICE_ALLOCATION(dx.memcheck());
    CHECK_DEVICE_ALLOCATION(dy.memcheck());
    CHECK_DEVICE_ALLOCATION(d_alpha.memcheck());
    CHECK_DEVICE_ALLOCATION(d_beta.memcheck());

    double hipblas_error_host{0}, hipblas_error_device{0};

    T h_alpha = arg.get_alpha<T>();
    T h_beta  = arg.get_beta<T>();

    // Initial Data on CPU
    hipblas_init_matrix(
        hA, arg, hipblas_client_alpha_sets_nan, hipblas_general_matrix, true, false);
    hipblas_init_vector(hx, arg, hipblas_client_alpha_sets_nan, false, true);
    hipblas_init_vector(hy, arg, hipblas_client_beta_sets_nan);

    // copy vector is easy in STL; hz = hy: save a copy in hz which will be output of CPU BLAS
    hy_cpu = hy;

    // copy data from CPU to device
    CHECK_HIP_ERROR(dA.transfer_from(hA));
    CHECK_HIP_ERROR(dx.transfer_from(hx));
    CHECK_HIP_ERROR(dy.transfer_from(hy));
    CHECK_HIP_ERROR(hipMemcpy(d_alpha, &h_alpha, sizeof(T), hipMemcpyHostToDevice));
    CHECK_HIP_ERROR(hipMemcpy(d_beta, &h_beta, sizeof(T), hipMemcpyHostToDevice));

    /* =====================================================================
           HIPBLAS
    =================================================================== */

    if(arg.unit_check || arg.norm_check)
    {
        CHECK_HIPBLAS_ERROR(hipblasSetPointerMode(handle, HIPBLAS_POINTER_MODE_HOST));
        DAPI_CHECK(hipblasGemvFn,
                   (handle,
                    transA,
                    M,
                    N,
                    reinterpret_cast<Ts*>(&h_alpha),
                    dA,
                    lda,
                    dx,
                    incx,
                    reinterpret_cast<Ts*>(&h_beta),
                    dy,
                    incy));

        CHECK_HIP_ERROR(hy_host.transfer_from(dy));
        CHECK_HIP_ERROR(dy.transfer_from(hy));

        CHECK_HIPBLAS_ERROR(hipblasSetPointerMode(handle, HIPBLAS_POINTER_MODE_DEVICE));
        DAPI_CHECK(hipblasGemvFn,
                   (handle, transA, M, N, d_alpha, dA, lda, dx, incx, d_beta, dy, incy));

        CHECK_HIP_ERROR(hy_device.transfer_from(dy));

        /* =====================================================================
           CPU BLAS
        =================================================================== */

        ref_gemv<T>(
            transA, M, N, h_alpha, hA.data(), lda, hx.data(), incx, h_beta, hy_cpu.data(), 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(arg.unit_check)
        {
            unit_check_general<T>(1, dim_y, abs_incy, hy_cpu, hy_host);
            unit_check_general<T>(1, dim_y, abs_incy, hy_cpu, hy_device);
        }
        if(arg.norm_check)
        {
            hipblas_error_host = norm_check_general<T>('F', 1, dim_y, abs_incy, hy_cpu, hy_host);
            hipblas_error_device
                = norm_check_general<T>('F', 1, dim_y, abs_incy, hy_cpu, hy_device);
        }
    }

    if(arg.timing)
    {
        double      gpu_time_used{0};
        hipStream_t stream;
        CHECK_HIPBLAS_ERROR(hipblasGetStream(handle, &stream));
        CHECK_HIPBLAS_ERROR(hipblasSetPointerMode(handle, HIPBLAS_POINTER_MODE_DEVICE));
        CHECK_HIP_ERROR(dy.transfer_from(hy));

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

            DAPI_DISPATCH(hipblasGemvFn,
                          (handle, transA, M, N, d_alpha, dA, lda, dx, incx, d_beta, dy, incy));
        }
        gpu_time_used = get_time_us_sync(stream) - gpu_time_used;

        hipblasGemvModel{}.log_args<T>(std::cout,
                                       arg,
                                       gpu_time_used,
                                       gemv_gflop_count<T>(transA, M, N),
                                       gemv_gbyte_count<T>(transA, M, N),
                                       hipblas_error_host,
                                       hipblas_error_device);
    }
}