/**
* 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_bfloat16.asc
 * \brief
 */

#include <iostream>
#include <iterator>
#include <vector>
#include "acl/acl.h"
#include "tiling/platform/platform_ascendc.h"
#include "simt_api/asc_fp16.h"

__global__ void math_custom(bfloat16_t* x, bfloat16_t* y, bfloat16_t* 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] = hlog2(x[idx]);
    z[idx] = hlog10(x[idx]);
    z[idx] = hcos(x[idx]);
    z[idx] = hsin(x[idx]);
}

std::vector<uint16_t> add(std::vector<uint16_t>& x, std::vector<uint16_t>& y)
{
    size_t total_byte_size =x.size() * sizeof(uint16_t);
    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;
    bfloat16_t* x_device = nullptr;
    bfloat16_t* y_device = nullptr;
    bfloat16_t* 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);
    // Calc splite params
    uint32_t block_num = 48;
    uint32_t thread_num_per_block = 256;
    uint32_t dyn_ubuf_size = 0;  // No need to alloc dynamic memory.
    // Call kernel funtion with <<<...>>>
    math_custom<<<block_num, thread_num_per_block, dyn_ubuf_size, 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<uint16_t> output((uint16_t *)z_host, (uint16_t *)(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<uint16_t> x(in_shape);
    std::vector<uint16_t> y(in_shape);
    std::vector<uint16_t> output = add(x, y);

    return 0;
}