[Need-attention-to]about func "adv_accum_wind" in file "adv_mod.F90"

[BeverlyCrl]

在文件adv_mod.F90的末尾，有一个叫作adv_accum_wind的函数。其性能消耗如下：

其中第一部分浅黄色运用了运算符%，大致含义类似于C语言中的.。这里的消耗主要是来自于chen成员调用的复制与拷贝。
第二部分则是涉及到的下层函数adv_batch。
可以考虑优化这两个部分。
其中第一部分可能需要对调用该函数的数据进行一些了解，可能存在类似于更换库函数之类的优化方法，但不是简单修改代码能够搞定的。
第二部分涉及到整个adv_batch_mod.F90文件，在这个部分中能够找到一些类似迭代步进的代码，这往往是主要耗时的part，详细可以见下面的图片。整个文件中有一些算数运算和循环嵌套的情况，也许可考虑并行，访存优化，数学化简等方式。
（迭代步进指的是下面图片上提到的399~433，能够明显看出这里面的步进逻辑）

以下是图片帮助你找到我描述的位置：

[Jonathan523]

Do you mean code like this one
mesh%half_kds
But in Fortran, that means half_kds is a member of mesh.
mesh here is like the struct in C++

[Jonathan523]

Here’s an example of a basic derived type:

type :: t_pair
  integer :: i
  real :: x
end type

The syntax to create a variable of type t_pair and access its members is:

! Declare
type(t_pair) :: pair
! Initialize
pair%i = 1
pair%x = 0.5

The percentage symbol % is used to access the members of a derived type.

[BeverlyCrl]

Do you mean code like this one mesh%half_kds But in Fortran, that means half_kds is a member of mesh. mesh here is like the struct in C++

I think I mistakenly recorded the meaning of the operator% and let me modify my description of the problem

[BeverlyCrl]

Here’s an example of a basic derived type:

type :: t_pair
  integer :: i
  real :: x
end type

The syntax to create a variable of type t_pair and access its members is:

! Declare
type(t_pair) :: pair
! Initialize
pair%i = 1
pair%x = 0.5

The percentage symbol % is used to access the members of a derived type.

I have rephrased the potential issues here, which may provide new ideas. Thank you for your help. I was not aware of this mistake.