Hello fellow Fortran enthusiasts,

I’ve been diving deep into Fortran programming lately, and one topic that has caught my attention is the use of pointers in Fortran. While Fortran is renowned for its robust array abstractions and efficiency in numerical computing, I’m curious about scenarios where pointers might come into play.

From what I understand, in typical mathematical computing tasks, direct manipulation of memory addresses via pointers might not be necessary, thanks to Fortran’s powerful array operations.

For instance, consider a matrix-matrix multiplication kernel implemented in C:

```
void mxm(double *C, double *A, double *B, int N) {
for (int i = 0; i < N; ++i) {
for (int j = 0; j < N; ++j) {
double *A_ptr = &A[i * N + 0];
double *B_ptr = &B[0 * N + j];
double *C_ptr = &C[i * N + j];
for (int k = 0; k < N; ++k) {
*C_ptr += *A_ptr * *B_ptr;
++A_ptr;
B_ptr += N;
}
}
}
}
```

While this approach may seem unconventional in Fortran, it mirrors a common pattern found in languages like C. Utilizing pointers and pointer arithmetic can potentially enable compilers to optimize computations more effectively by exposing specific compute patterns and enhancing memory locality.

I’m intrigued by the possibility of using pointers in scenarios such as systems programming, scientific computing, or when dealing with data structures like linked lists.

I’d love to hear from the community about their experiences with pointers in Fortran. Have you encountered situations where using pointers proved beneficial, even in the realm of scientific computing? Are there best practices or guidelines to follow when incorporating pointers into Fortran code?

Your insights and experiences would be greatly appreciated.

Best regards and nice to meet you all,

Iñaki