/**
* Copyright (c) 2026 Huawei Technologies Co., Ltd.
* This program is free software, you can redistribute it and/or modify it under the terms and conditions of
* CANN Open Software License Agreement Version 2.0 (the "License").
* Please refer to the License for details. You may not use this file except in compliance with the License.
* THIS SOFTWARE IS PROVIDED ON AN "AS IS" BASIS, WITHOUT WARRANTIES OF ANY KIND, EITHER EXPRESS OR IMPLIED,
* INCLUDING BUT NOT LIMITED TO NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.
* See LICENSE in the root of the software repository for the full text of the License.
*/

/* !
 * \file math_cpp.asc
 * \brief
 */

#include <iostream>
#include <iterator>
#include <vector>
#include "acl/acl.h"
#include "kernel_operator.h"
#include "asc_simt.h"

__simt_vf__ __launch_bounds__(1024) inline void simt_math_custom(__gm__ float* x, __gm__ float* y, __gm__ float* z, uint64_t total_length)
{
    // Calculate global thread ID
    int32_t idx = blockIdx.x * blockDim.x + threadIdx.x;

    // Maps to the row index of output tensor
    if (idx >= total_length) {
        return;
    }
    z[idx] = AscendC::Simt::Tan(x[idx]);
    z[idx] = AscendC::Simt::Tanh(x[idx]);
    z[idx] = AscendC::Simt::Atan(x[idx]);
    z[idx] = AscendC::Simt::Atanh(x[idx]);
    z[idx] = AscendC::Simt::Exp(x[idx]);
    z[idx] = AscendC::Simt::Log(x[idx]);
    z[idx] = AscendC::Simt::Cos(x[idx]);
    z[idx] = AscendC::Simt::Cosh(x[idx]);
    z[idx] = AscendC::Simt::Sin(x[idx]);
    z[idx] = AscendC::Simt::Sinh(x[idx]);
}

 __global__ __vector__ void math_custom(__gm__ float* x, __gm__ float* y, __gm__ float* z, uint64_t total_length)
{
    asc_vf_call<simt_math_custom>(dim3(1024), x, y, z, total_length);
}

std::vector<float> math(std::vector<float>& x, std::vector<float>& y)
{
    size_t total_byte_size =x.size() * sizeof(float);
    int32_t device_id = 0;
    aclrtStream stream = nullptr;

    uint8_t* x_host = reinterpret_cast<uint8_t *>(x.data());
    uint8_t* y_host = reinterpret_cast<uint8_t *>(y.data());
    uint8_t* z_host = nullptr;
    float* x_device = nullptr;
    float* y_device = nullptr;
    float* z_device = nullptr;
    // Init
    aclInit(nullptr);
    aclrtSetDevice(device_id);
    aclrtCreateStream(&stream);
    // Malloc memory in host and device
    aclrtMallocHost((void **)(&z_host), total_byte_size);
    aclrtMalloc((void **)&x_device, total_byte_size, ACL_MEM_MALLOC_HUGE_FIRST);
    aclrtMalloc((void **)&y_device, total_byte_size, ACL_MEM_MALLOC_HUGE_FIRST);
    aclrtMalloc((void **)&z_device, total_byte_size, ACL_MEM_MALLOC_HUGE_FIRST);
    aclrtMemcpy(x_device, total_byte_size, x_host, total_byte_size, ACL_MEMCPY_HOST_TO_DEVICE);
    aclrtMemcpy(y_device, total_byte_size, y_host, total_byte_size, ACL_MEMCPY_HOST_TO_DEVICE);
    // Call kernel funtion with <<<...>>>
    math_custom<<<2, nullptr, stream>>>(x_device, y_device, z_device, x.size());
    aclrtSynchronizeStream(stream);
    // Copy result from device to host
    aclrtMemcpy(z_host, total_byte_size, z_device, total_byte_size, ACL_MEMCPY_DEVICE_TO_HOST);
    std::vector<float> output((float *)z_host, (float *)(z_host + total_byte_size));
    // Free memory
    aclrtFree(x_device);
    aclrtFree(y_device);
    aclrtFree(z_device);
    aclrtFreeHost(z_host);
    // DeInt
    aclrtDestroyStream(stream);
    aclrtResetDevice(device_id);
    aclFinalize();
    return output;
}

int32_t main(int32_t argc, char* argv[])
{
    constexpr uint32_t in_shape = 48 * 256;
    std::vector<float> x(in_shape);
    for (uint32_t i = 0; i < in_shape; i++) {
        x[i] = i * 1.1f;
    }

    std::vector<float> y(in_shape);
    for (uint32_t i = 0; i < in_shape; i++) {
        y[i] = i + 3.4f;
    }

    std::vector<float> golden(in_shape);
    for (uint32_t i = 0; i < in_shape; i++) {
        golden[i] = x[i] + y[i];
    }

    std::vector<float> output = math(x, y);

    return 0;
}