c++模板杂计

一 杂计

1.1 基础代码

template<typename T>
T add(T a, T b) 
{
    return a + b;
}

template int add<>(int, int); // 显式实例化

int main()
{
    add<int>(1, 2); // 显式指定模板参数 T
    add(1, 2); // 自动推导模板参数 T
    add(1.0f, 2.0f); // 自动推导并且隐式实例化
}

模板是本质上是代码生成器 

1.2 参数中有函数

template <typename T, void (*OpFunc)(T *, T *, T *, T *, int, int, float)>
__mlu_func__ void block3Unary(T *x, T *y, char *nram_buffer, int32_t num_total,
                              int32_t offset_x_half, int32_t offset_aux_a,
                              int32_t offset_aux_b, int32_t num_deal,
                              int32_t num_pong, float coef) {
  if (__is_mpu()) {
    return;
  }
  int32_t num_per_core = num_total / taskDim;
  int32_t num_rem = num_total % taskDim;
  T *addr_x = (T *)x + taskId * num_per_core;
  T *addr_y = (T *)y + taskId * num_per_core;
  if (num_rem > 0 && taskId == taskDim - 1) {
    num_per_core = num_per_core + num_rem;
  }
  int32_t repeat = num_per_core / num_deal;
  int32_t rem = num_per_core % num_deal;
  int32_t align_rem = CEIL_ALIGN(rem, UNARY_ALIGN_NUM);

  T *nram_x = (T *)nram_buffer;
  T *nram_x_half = (T *)nram_buffer + offset_x_half;
  T *nram_aux_a = (T *)nram_buffer + offset_aux_a;
  T *nram_aux_b = (T *)nram_buffer + offset_aux_b;
  int32_t span_handle_size = num_deal * sizeof(T);

  // 3 level pipeline.
  if (repeat > 0) {
    __memcpy_async(nram_x_half, addr_x, span_handle_size, GDRAM2NRAM);
    __sync();
  }

  if (repeat > 1) {
    __memcpy_async(nram_x_half + num_pong, addr_x + num_deal, span_handle_size,
                   GDRAM2NRAM);
    OpFunc(nram_x, nram_x_half, nram_aux_a, nram_aux_b, num_deal, num_deal,
           coef);
    __sync();
  }

在使用的模板时候，需要传入类型参数，和函数的类型